Sie sind auf Seite 1von 330

T he S UCC ~"'S'rtn. .

I;..' \' ESTO R'S G uide to Spoiling

1 ;..' \'E ST ~ I E ;..'T 'I' RE;,\ DS

"lid T UR;.." ' C I'O I\ 'TS...

'OURTH L D , T , O ..

The Successful Investor's Guide
to Spotting Investment Trends
and Turning Points

Fourth Edition

Martin J. Pring

New York Chicago San francisco
lisbon London Madrid Mexico City Milan
New Deihl San Juan Seoul Singapore
Sydney Toronto
Library of Congress Cataloging-in-Publication nata

Pring, Martin J.
Technical analysis explained / the successful investor's guide to spotting
investment trends and turning points / by Martin Pring.-4th ed.
p. em.
ISBN 0-07-138193-7
1. Investment analysis. I. Title

HG4529 .P75 2002

332.63'22-dc21 2001044981
To my son, Thomas William Pring
McGraw-Hill iZ
A DivisionofThdf&G_.H4lComptmies

Copyright ©2002 by McGraw-Hill. All rights reserved. Printed in the United States
ofAmerica. Except as permitted under the United States Copyright Act of 1976, no
part of this publication may be reproduced or distributed in any form or by any
means, or stored in a data base or retrieval system, without the prior written per-
mission of the publisher.
67890 DOC/DOC 0876
ISBN: 0-07-138193-7
The sponsoring editor for this book was Stephen Isaacs and the production super-
visor was Clare Stanley. It was set in New Baskerville by MacAllister Publishing
Services, LLC.
Printed and bound by R R Donnelley & Sons Company.
This publication is designed to provide accurate and authoritative information in
regard to the subject matter covered. It is sold with the understanding that neither
the author nor the publisher is engaged in rendering legal, accounting, or other
professional service. If legal advice or other expert assistance is required, the ser-
vices of a competent professional person should be sought.
-From a Declaration of Principles jointly adopted
by a Committee of the American Bar
Association and a Committee of Publishers

McGraw-Hill books are available at special quantity discounts to use as premiums

and sales promotions, or for use in corporate training programs. For more infor-
mation, please write to the Director of Special Sales, Professional Publishing,
McGraw-Hili, Two Penn Plaza, New York, NY 10121-2298. Or contact your local

O This book is printed on recycled, acid-free paper containing a minimum of

50% recycled de-inked fiber.
Preface xi
Acknowledgments xiv

Introduction 1

Part r, Trend-Determining Techniques 13

1 The Market Cycle Model 15
..........2. Financial Markets and the Business Cycle 26
/3. Dow Theory 36
_4. Typical Parameters for IntermediateTrends 50
.> 5. Price Patterns 63
" 6. Smaller Price Patterns 98
7. One- and Two-Bar Price Patterns 111
8. Trendlines 136
9. Moving Averages 154
10. Momentum Principles 179
11. Individual Momentum Indicators I 211

VnI Ix

/"", 12 . Individual Momentum Indlcaton D 237 30. Technical Analysis of Global Stock Markets 564
__ 13. Candle Charts 257 31. Technical Analysis of individUal Stocks 574
/ 14 . Point and figure Charting 277 EpUogue 593
__ 15. Miscellaneous Techniques for Detennlnlng
Appendix TIle Elliott Wave 595
Trends 287
Glossary 600
./ 16. The Concept of Relative Strength 304
Bibliography 604
17. Putting the Indicators Together: The OJ
Transports 1990·2001 322 Index 609

Part II. Market Structure 329

,./ 18. Price: The Major Averages 331
/ , 19 . Price: Group Rotation 352
.,./ 20. TIme: Longer-Tenn Cycles 364
21. Practical Identification of Cycles 391
/ 22. Volume: General Principles 397
/"" 23. Volume Oscillators 409
,./ 24. Breadth 435

Part m. Other Aspects of Market Behavior 461

25. Why Interest Rates Affect the Stock Market 463
/'" 26. Sentiment Indicators 485
27. Applying Technical Analysis to the Theory
of Contrary Opinion 511
28. Checkpoints for Identifying Primary Stock
Market Peaks and Troughs 526
29 . Automated Trading Systems 539
There is no reason why anyone cannot make a substantial amount of money
in the financial marJ<.e~ but there are many reasons why many people will
not. As with most 'eiicteilv';;rs in life, the key to success is knowledge and
action. This book has been written in an attempt to shed some light on the
internal workings of the markets and to help expand the knowledge compo-
nent, leaving the action to the patience, discipline, and objectivity of the indi-
vidual investor.
The mid- to late 1980s saw the expansion of investment and trading
opportunities to a global ~c~ in terms of both the cash and the futures
markets. In the 1990s, innoVatlohs in the communications industry enabled
anyone to plot data on an intraday basis for relatively little cost. Today,
numerous charting sites have sprung up on the Internet, so virtually any-
one now has the ability to practice technical analysis. As a consequence of
the technological revolution, time horizons have been greatly shortened. I
am not sure that this is a good thing because short-term trends experience
more random noise than longer-term ones. This means that the technical
indicators are not as effective. The fou~edition of Technical Analysis
Explained has been expanded and totally reVised to keep abreast of many of
these changes, and to include some technical innovations and evolvement
in my own thinking since the publication of the third edition.
Every chapter has been thoroughly reworked and expanded. In the inter-
est of efficiency, some have been dropped and others substituted.
Considerable attention continues to be focused on the U.S. equity market,
but many of the marketplace examples feature international stock indexes,
currencies, commodities, and precious metals. Special chapters also feature
technical analysis of the credit markets and global equities. In most cases,
the marketplace examples have been updated, but some older ones from

xii Preface Preface xiii

, previous editions have been left in deliberately to give the book some his- on positions that go against the direction of the main trend. If a whipsaw is
torical perspective. These historical examples also underscore the point that going to develop, it will usually arise from a contratrend signal.
nothing has really changed in the last 100 years. The same true and tried To be successful, technical analysis should be regarded as the art of assess-
principles are as relevant today as they always were. I have no doubt what- ing the technical position of a particular security with the aid of several. s~i­
soever that this win continue to be so in the future. entifically researched indicators. Although many of the mec~amstlc
. Thus, tech?ical analysis could be applied in New York in 1850, in Tokyo techniques described in this book offer reliable indications of changmg mar-
m 1950, and in Moscow in 2150. This is true because price action in finan- ket conditions, all suffer from the common characteristic that they can, and
cial markets is a reflection of human nature, and human nature remains often do, fail to operate satisfactorily. This attribute presents no problem to
more or le~s c?nstan~ Technical principles can also be applied to any freely the consciously disciplined investor or trader, since a good working knowl-
traded entity many time frame. A trend reversal signal on a 5-minute bar edge of the principles underlying major price movements in financial
c~a~t is base~ o? the same indicators as one on a monthly chart; only the kets and a balanced view of the overall technical position offer a supenor
SIgnificance IS different, Shorter time frames reflect shorter trends and are framework within which to operate.
therefore less significant. There is, after all, no substitute for independent thought. The action of
Several new c~ap~~r~'b-ave.been added in this edition. One of the areas the technical indicators illustrates the underlying characteristics of any mar-
I become more Impressed With the more I study it is the concept of one- ket, and it is up to the analyst to put the pieces of the jigsaw puzzle together
and two-bar price reversals. They are also 'refe~t for intraday and swing and develop a working hypothesis.
traders. Consequently, a brand new chapter on this subject has been intro- The task is by no means easy, as initial success can lead to overconfidence
~uced in this edition. In view of their growing popularity, candlesticks, pre- and arrogance. Charles H. Dow, the father of technical analysis, once wrote
viously relegated to an Appendix, now receive full treatment. A third words to the effect that "Pride of opinion caused the downfall of more men
chapter has been added to the momentum section so that material on the on Wall Street than all the other opinions put together." This is true
Directional Movement System, the Chande Momentum Oscillator the because markets are essentially a reflection of people in action. Normally,
Relative Momentum Index, and the parabolic (not a true momentumindi- such activity develops on a reasonably predictable path. Since people can
c~tor, but valuable nonetheless) could be included. In addition, the expan- -and do-change their minds, price trends in the market can deviate unex-
sion of~e momentum section has left room for some of my own new ideas pectedly from their anticipated course. To avoid serious trou?le, investo~,
concermng momentum interpretation, such as extreme swings, mega- and especially traders, must adjust their attitudes as changes m the techm-
?verboughts and mega-oversolds, and so on. The volume section also cal position emerge..
mcludes an additional chapter so that indicators such as the Demand Index In addition to pecuniary rewards, a study of the market can also reveal
and Chaikin Money Flow could be covered. Greater emphasis is also placed much about human nature, both from observing other people in action and
on volume morne? tum. The concept of relative strength is a very important from the aspect of self-development. As investors react to the constant strug-
and underappreclated :urn of technical analysis. It too has been upgraded gle through which the market will undoubtedly put them, they will also learn
to full chapter status. Finally, a new chapter on applying technical analysis a little about their own makeup. Washington Irving might wen have been
to contrary opinion theory expands our coverage of the psychological referring to this challenge of the markets when he wrote, "Little minds are
aspects of trading and investing. taxed and subdued by misfortune but great minds rise above it."
Since the .1970s, the time horizon of virtually all market participants has '
shn:nk consld~rably.As a result, technical analysis has become very popular Martin J. Pring
for l1~plementing short-term timing strategies. This use may lead to great dis-
ap~Ol~.tment: In my experience, there is a rough correlation between the
~elIablbtyof the technical indicators and the time span being monitored. This
IS why most of the discussion here has been oriented toward intermediate
and long-term trends. Even short-term traders with a 1- to 3-week time hori-
zo~ need t? have some un?e~tandingof the direction and maturity of the
main or pnmary trend. This IS because mistakes are usually made by taking


Acknowledgments Introduction
This book has evolved over the last 20 years and has grown from a text of To investors willing to buy and hold common stocks for the long term, the
under 250 pages in the first edition to over 600 in the current one. stock market has offered excellent rewards over the years in terms of both
The material for the fourth edition of this book has been gathered from dividend growth and capital appreciation. The market is even more chal-
a ~ubstantial number of sources, and I am deeply indebted to the manyorga- lenging, fulfilling, and rewarding to resourceful investors willing to learn
nizations that have given their permission to reproduce charts and figures the art of market timing through a study of technical analysis.
without which this book would not have been possible. Special thnks go to The advantages of this approach over the buy-and-hold approa~e~~~?
Tim Hayes and Ned Davis at Ned Davis research for providing many of the particularly marked between 1966 and 1982. The market made no headway
charts on sentiment and flow of funds indicators. at all, as measured by the Dow Jones Industrial Average (DJIA), in the 16
Appreciation also goes to Danny Pring who came up with the title for the years between 1966 and 1982. Yet there were some substantial price fluctu-
first edition of this book. It's a great title and has followed it ever since. ations. Although the DJIA failed to record a net advance between 1966 and
Thanks also to my colleagues at Pring Research includingJimmie Sigsway 1982, the period included five major advances totaling over :5~0 Dow
whose efficiency and dedication allowed my writing to progress in a serene points. The potential rewards of market timing were therefore slg?lficant.
atmosphere uninterrupted by the construction going on around us. Anyone A long-term investor fortunate enough to sell at the five tops 10 1966,
who has experienced the rebuilding of a homez'offlce complex will know 1968, 1973, 1979, and 1981 and to reinvest the money at the troughs of 1966,
what I mean. She is also a grat mother in lawl 1970, 1974, 1980, and 1982 would have seen the total investment (exclud-
Above all thanks go to my wife Lisa, who tirelessly reconstructed most of ing transactions costs and capital gains tax) grow from a theoretical $1000
the charts and figures from my pathetic PowerPoint illustrations while (that is, $1 for every Dow point) in 1966 to over $10,000 by October 1983.
~imultaneously running a home, taking care of our son Thomas, maintain- In contrast, an investor following a buy-and-hold approach would have real-
Ing our web site ( and managing a major house renovation. ized a mere $250 gain over the same period. Even during the spectacular
rise that began in August 1982, technical analysis would have proved use-
ful, since that period witnessed a considerable variation in performance
between different industry groups.
A bull market, such as the one that occurred in the 1980s and 1990s, is a
once-in-a-generation affair. In fact, it was a record in 200 years of recorded
U.S. stock market history. This implies that the opening decade of the
twenty-first century will be a more difficult and challenging period, and that
market timing will prove to be of crucial importance.

2 3

In. pr.lcti~e, of COU Be, it is impossible to buy and ~II consistently at exact ical forces. The an of technical analysis, for it is an art, is to itkntifJ atrmt.!
rummg points, but the enormous potential of th is approach still leaves rn.omal at a rr:lalively early s~ and ritU on that tmuI until the weight of the ~­
plenty of mam for error, even when commission costs and taxes are included. denu shows Of' pruves that 1M trmd has rr:vmed. The evidence in .15
in the calculation. Th e rewaeds for identifying major marketjunclures and represented by the numerous scientijiaJlly denved indicators described In
taking the appropri ate action can be substantial.
this book. .
O~ginally, te~hnical an alysis was applied. principally in the equity market, Human nature remains more or less constant and tends to react to sim-
but Its populanty has gradually expanded to embrace commod ities debt ilar situations in consistent ways. Bystudying the nature of previous market
instru ments, currencies, an d other international markets. In the mrs of the turning points. it is possible to deve lop some characteristics th~t.can help
old market, participan ts had a fairly long time horizon. stretching over to ide ntify market tops and bottoms. Therefore. technical ana9su IS based on
months or yea rs. Th ere have always been short-term traders and scalpers. the assumption that people will continue to makt the same mistaJaes thrJ have made
bu t the technological revolution in communications has shortened the time in the past H uman relationships are extre ~ely complex a?d never repe~t
ho rizon ofj~t. about.everyo~e involved in markets. When holding periods in identi cal combinatio ns. Th e markets. which are a reflection of peopl ~ In
are length~. It ~ posslbl~ t~ m~ulge in ~e luxury of fundamental analysis. action . never duplicate their performance exactly: ~ut the ~u~n~e of Sim-
but when time 15 short. tim ing 15 everything. In such an environme nt tech- ilar ch ara cteristi cs is sufficient to enable technicians to identify Jun cture
nical analysis really comes into its own. • points. Since no single indicato r has signaled. or indeed could signal. every
To be successful, ~e technical approach involves taking a position con-
~ry to ~e ~X~tauons the cro wd.. Th is requires the patience. objectiv-
Ity, and diSCipline to acqu ire a fin ancial asset at a time of depression and
top or bottom. technical analysts have developed an arsenal of tools to help
isolate these points.
gl~m. and liquidate it in an environment of eupho ria and excessive opti-
mism. Th e level of pessimism or op timism will depend on the turning point.
Sho rt-term peaks and troughs are associated with more moderate extremes Three Branches of
in sen.time m than longer-term ones. The aim of this book is to exp lain the Technical Analysis
~mcal c~aracteristi~ to be expe cted at all of these market turn ing Technical analysis can be broken down into three essential areas: sentiment,
po mts , parucularIy major ones, and to help to assess them objectively. flow-of-funds . and market structure indicators. Data and indicato~ for all
three areas are available for the U.S. stock marke t. For other finan cial mar-
kets th e statistics are more or less confined to the market structure indica-
Technical Analysis Defined ton: The m.yor exceptions are futures markets based in the ~nited States .
for which short-term sentiment data are available. The fol1owmg comments
. During the course of the book when it is time to emphasize a specifi c bu t on sentime nt and flow-of.funds indicators relate to the U.S. stock market.
Important po int, it will be highlighted in the following way:

Sentiment Indicators
Sentiment or expeetational indicators monitor the action s of ?ifferent mar-
ket participants. such as insiders, mutual funds man agers and investors, and
floor specialists.Jus t as the pendulum ofa clock continually moves from .one
extreme to another, so the sentiment indexes (which monitor the emouons
of investors ) move from one extreme a t a bear market bo ttom to another
The te~hnical approach to investment is essentially a reflection of the idea at a bull market top. The assumption on which these indicators are bas.ed
~at pnces move In tr~nds that are de term ined by the changing atti tudes of is that different groups of investors are consistent in their actions at major
mvestora toward a variety of econ omic, mon etary. political, and psycholog- market turning points. For example. insiders (that is. key employees or

4 TechnicalAnalysis Explained Introduction 5

major stockholders of a company) and New York Stock Exchange (NYSE) A superior approach to flow-of-funds analysis is derived from an exami-
members as a group have a tendency to be correct at market turning points; nation of liquidity trends in the banking system, which measures financial
in aggregate, their transactions are on the buy side toward market bottoms pressure not only on the stock market, but o~ the economy as well.
and on the sell side toward tops.
Conversely, advisory services as a group are often wrong at market turn-
ing points, since they consistently become bullish at market tops and bear- Market Structure Indicators
ish at market troughs. Indexes derived from such data show that certain
This area of technical analysis is the main concern of this book, embracing
readings have historically corresponded to market tops, while others have
market structure or the character Of the marketindicators. .These indicators mon-
been associated with market bottoms. Since the consensus or majority opin- 'J '::'1..::-'-<4.1'
itor the trend of various price indexes, market breadth, cycles, volume, and
ion is normally wrong at market turning points, these indicators of market
so on in order to evaluate the health of thlpre~hing trend.
psychology are a useful basis from which to form a contrary opinion.
Indicators that monitor the trend of a price include moving averages,
peak-and-trough analysis, price patterns, and trendlines. Such techniques
can also be applied to the sentiment and flow-of-funds indicators discussed
Flow-of-Funds Indicators
previously. This is because these indicators also move in trends. When the
The area of technical analysis that involves what are loosely termed flow-of- trend of psychology, as reflected in these series, reverses, prices are also likely
funds indicators analyzes the financial position ofvarious investor groups in to change direction.
an attempt to measure their potential capacity for buying or selling stocks. Most of the time, price and internal measures, such as market breadth,
Since there has to be a purchase for each sale, the ex post, or actual dollar momentum, and volume, rise and fall together, but toward the end of mar-
balance between supply and demand for stock, must always be equal. The ket movements, the paths of many of these indicators diverge from the
price at which a stock transaction takes place has to be the same for the buyer price. Such divergences offer signs of technical deterioration during
and the seller, so naturally the amount of money flowing out of the market advances, and technical strength following declines. Through judicious
must equal that put in. The flow-of-funds approach is therefore concerned observation of these signs of latent strength and weakness, technically ori-
with the before-the-fact balance between supply and demand, known as the ented investors are alerted to the possibility of a reversal in the trend of
exante relationship. If at a given price there is a preponderance of buyers over the market itself.
sellers on an ex ante basis, it follows that the actual (ex post) price will have Since the technical approach is based on the theory that the price is a
to rise to bring buyers and sellers into balance. reflection of mass psychology, or the crowd in action, it attempts to forecast
Flow-of-funds analysis is concerned, for example, with trends in mutual future price movements on the assumption that crowd psychology moves
fund cash positions and those of other major institutions, such as pension between panic, fear, and pessimism on one hand and confidence, excessive
funds, insurance companies, foreign investors, bank trust accounts, and cus- optimism, and greed on the other. As discussed here, the art of technical
tomers' free balances, which are normally a source of cash on the buy side. analysis is concerned with identifying these changes at an early phase, since
On the supply side, flow-of-funds analysis is concerned with new equity offer- these swings in emotion take time to accomplish. Studying these market
ings, secondary offerings, and margin debt. trends enables technically oriented investors and traders to buy or sell with
This money flow analysis also suffers from disadvantages. Although the a degree of confidence in the principle that once a trend is set in motion,
data measure the availability of money for the stock market (for example, it will perpe'fiiite itself.
mutual fund c¥J:1 position or pension fund cash flow), they give no indica-
tion of the inclfilation of market participants to use this money for the pur-
chase of stocks, or of their elasticity or willingness to sell at a given price on
the sell side. The data for the major institutions and foreign investors are
not sufficiently detailed to be of much use, and in addition they are reported Classification of Price Movements
well after the fact. In spite of these drawbacks, flow-of-funds statistics may Price movements may be classed as primary, intermediate, and short term.
be used as background material. Major movements, sometimes called primary or cyclical, typically work

. ~, " ,,

J \ th~mse1ves out In a period of 1 to! yean and are a renecnon of investon'

. .... / aw tudes ~ward. the business cycle. J~ ~ usually develop
~..J . ever a ~n~ of 6 ~ks to as many months., eometimes lon~r. Although
y ~ ofpnme ~~~ce. they are nevenheless useful to identify. It is clearly
rmportant tc disunguuh between an intermediate reaction in a bull market
an~ the tim downleg of a bear market, (or example. SJacn...Ierm ~
which last less than ~ o r .. WttU, teed 00 be random in nature. Sttular or
~ry long term trends embr.tcing several primary trend movements and sures the after-tax profits of all the companies whose shares are
mtraday trends lasting a few minutes to a few houn round out the p0ssi- listed in the market. and it measures these cumulative profits 50 fat'
bilities for price movements, into the future one might as well say the horizon is infinite. Th is
cumulative mass of after-tax profits is then, as the economists will
say, "discounted back to present value" by the market, A man does
Discounting Mechanism the same thing when he pays more for one razor blade than
of the Market another, figuring he' ll get more or easier shaves in the future with
the hig her-priced on e, and figuring its presen t value on that basis.
~u priu movt1I'lmtsbave one thing in common : 1'kJ arrc rrfl«tion o/llu trmd This fu ture flow of earnings will ultima tely be affected by busi-
In l1uhopa,!ttm, Jrnowledgt, optimism, and greed oj marlrd parlidpanl.s. The sum ness conditions everywhe re on earth. Little bits of information are
total of these emotions is expressed in the price level, which is. as Garfield constantly flowin g into the market fro m around the world as well
Drew no ted. "never what th ey [stocks] aft wonh, but what people th ink th ey as throughout the Unite d States, an d the ma rket is mu ch more
an: worth,", efficient in reflecting these bits of news than are govern ment sta-
This process of market evaluation was well expressed. by an editorial in tisticians. The market relates this inform a tion to how mu ch ,
T7u WallStmet joumaJ.:1 Ameri can busi ness can earn in the future. Roughly speaking, the I~
general level of the market is the present value of the capital stock
!he stock .marke t consists of everyone who is "in the manet" buy- of the US. " -1~'
Ing or selling shares a t a given moment. plus everyone who is not cr- V"",,. - --
"in the market, b.ut might be if conditions were right, In this sense,
It This im'pttei that investo rs and traders are looking ahead and taking
the s.tod.. market IS po tentially everyone with any personal savings. action 50 that they can liquidate at a higher price when the anticipated news
It IS. this broad base. of participation and potential participation or development actually takes place. If expectatiqIlS concerning the devel.-
that gIves the martel ItS strength as an economic indicator and as opment are better or worse than origUuIJy thought, then investc rs sell either
:m allocator of scarce capital, Movements in and out of a stock. or sooner or later through the market mechanism, depending on the partic-
~ and out of the market. are made on the margin a.J each in~tor ular circumstances. Thus, the familiar maxim sdl on good ntUIS applies on
dlg~ts new information . Th is allows the market to incorporate all when the goodnews is right on or below the market's (that is, the investon ')
a~a~le .infonnation in a waX.tja6po one penon could hope to. expectations. If the news is good, but not as favorable as expected. a quick
Since Its Judgments are the c<1r&nsw of nearly everyone, it tends reassessment will take place, and the market (ether things being equal) will
to outperform any single penon or group. • . . [The market] mea- fall. Hthe news is better than anticipated, thepossibilities are obviously more ."
favorable . The reve~n.e ~1I , of course, be true in a declining market, This . ';
process explains thepal1ldox of equity markets peaking when eco nomic con- ~ 'I:
~rfield~ , Nru; MdJtodJft>r hufl1 ill 1M Stod MarlId, Me u:alfe Praa, Boston 1968, p. 18. ditions are strong, and fonning a bottom when the outlook is most gloomy.f -.. '-:" ~ ' :~I
r......J Wall s-t JOtDTI4l, OcL 20, 19n. Reprin ted by pennission o r th e Wall SInd jounJ4J. The principle of discounting is not confi ned to equities alo ne, bu t ca n be ( l,,:,.;. :.h.
~1'1 .. ght Dow J onel &. Co.• Inc. 19n . All rights rao:rved
applied to an] freely traded e ntity. 'i
v t~ 'f

. r.
,. " -'I
,'' ..,-,.,'
- .
:t .


The reaction of any market to news eeents PIll be mos t ins~ because
i, 9

if the market, as rdlec kd by price. ignoml sJ~1ybu llish news and sells
off. it is certain that the eYent was well discounted, that is, already built into
the price mechanism. and th e reaction should therefore be viewed bearishly.
If a market reacts mon;: ~:YQrably to bad n~ than migh t be expected. this
in tum should be inti-rf,retdt as a positive sign. There is a good deal of 'Nis-
dam in the saying, -A bear argument known isa bear argument understood,•

logical influences on price are ~ry important. Currencies do not fit well in to
business cycle analysis. Although data reported several months after th e fact
The Financial Markets
are ~ry good at explaining currency movements, technical has been
and the Business Cycle most usefu l for timel y forecasts and th e identification of e mergm g trends.
The maj o r movem ents in bond, SIOCk.. and co mmodity pri ces are cawed by
long-term trends in th e em o tions of me investing pu blic. Th ese em otio ns
reflect th e anticipated level and growth rate oCfuture econo mic activity,and
th e attitu de of lnYeSton toward that activity.
Technical Analysis and
For example. there is a definite link between pri mary moveme nts in the
Trend Determination
stoc k made l an d cyclical mceeme ms in th e economy because trends in co r. . ..; Since tech nical analysis involves a study of th e actio n of markets, it is no t
pon te profitability are an in tegral part of the bus iness cycle. I( b<l5ic eco- :/ concerned with th e d iffieult an d subjective tasks offore casting trends in the
nom ic fo rces alo ne influence th e stock market, th e task of determining th e economy, o r assessing the attitudes of investors toward th ose changes.
changes in primary movem en ts would be relatively simple. In practice , it is Technical analysis tries to identify turning points in th e marid ~ assessmen t
n ot, and this is due to several factors. of th ese factors.
Fint. cha nges in ~tj, direc~n of the economyca n take 5Om~ time to mate- Since technical anal ysis can be applied successfully to any freely.traded
rialize. M th e cyclelltrffo'rdi: other psychological considerations, such as entity such as stocks, market averag~s,co~modities, ~nds, I~ ~~a an d
political developments o r purely internal factors like a speculative buying so on, I will frequ~r,.tly use ~e .tenn s«Unty as a g~ne.~c one embracmg all
wave or selling pressure from margin calls, can affect the equity market and of these entities. ffie1eby avoid ing unnecessary repeuuon .
resul t in m i51~ading rallies and reactions of 5 to 10 percent or more. The appro ach taken here d iffers from that found in stan dard. presen ta-
Second, changes in th e mark et usuall y precede cha nges in the econo my tions of tech nical analysis. The various techniques wed to determm~ ~~ds
by 6 to 9 months, but th e lead time can someti mes be far shorter or lo nger. an d identify their reversals will be examined in Pan I. '7rend·Det~nrumng
In 1921 an d 1929. th e economy turned before th e market did. Techniques," which deals with price patterns, trendlmes, moving averages
Third. even when an economic recovery is in the middle ~ its cycle if''f. (MAs) , momentum, an d M) on . .
doubts about its durability often arise . when these d oubts c'oirlCid~ ~ Pan II. ·Mark~t Stru ctur e," is pri ncipally concerned with anal~1S of th e
political or o ther a~ne developme nts, sharp an d confwi ng co umer':-I"":' - U.S. equi ty market, although examples usin g other secu~ties are mdude~
cyclical price movements w uaUy develop. to dem onstrate that the p rin ciples are universally apphca ble. ~l th at 15
Fourth, profi ts may increase, but investors' attitudes toward th ose p rofits required are th e ;ppi6priate data. This section off~rs a mo re detailed expla-
may ch ange. For example, in th e spri ng of 1946 th e DJIA stood at 22 times nation of the M OW indicators and ind exes . It also sho ws how th ey can be
th e pric~/eami ngs rati o. By 1948, th e comparable rati o was 9.5 when mea- combined to build a framework for determinin g th e quality of th e Internal
sure d aga inst 1947 earnings. In thi s period, profits had almost doubled and struc tu re of the market. A study of market character is a co rnerstone of tech-
pri ce/ eamtngs ratios had fallen, but stock prices wer e lower. nical anal)'1is, since reversals of price trends in the major averages are alm ost ~n.d and commodity prices are linked m uch more directly alwa)'1 preceded by laten t strength or weakness in th e market struc ture .J ust
to econ omic acuviry than are stoc k marker prices, but even here, psycho- as a careful driver does not j ud ge th e performance of a car from th e


• 11
10 TechnicalAnalysis Explained The Market Cycle Model

tanding of the historical relationships between certain price ~verag~sd~d

speedometer alone, so technical analysis looks further than the price trends
of the popuIar averages. Trends of investor confidence are responsible for
~arket indicators can be used to identify turning poin~. ~o smgl.e m rca-
tor can ever be expected to signal all trend reve~s, so It IS essentIal to use
price movements, and this emotional aspect is examined from four view-
points or dimensions, namely, price, time, volume, and breadth. a number of them together to build up a consensus. .
This approach is by no means infallible, but a careful, patient, and objec-
Changes in prices reflect changes in investor attitude, and price, the first
tive use of the principles of technical analysis can put the odds of suc.ce~s
dimension, indicates the level of that change.
very much in favor of the investor or trader who incorporates these pnncl-
Time, the second dimension, measures the recurrence and length of cycles
in investor psychology. Changes in confidence go through distinct cycles, ples into an overall strategy.
some long and some short, as investors swing from excesses of optimism
toward deep pessimism. The degree ofprice movement in the market is usu-
ally a function of the time element. The longer it takes for investors to move
from a bullish to a bearish extreme,the greater the ensuing price change
is likely to be. The examples in the two chapters on time relate mainly to
the U.S. stock market, but much of this material is equally valid for com-
modities, bonds, or currencies.
Volume, the third dimension, reflects the intensity of changes in investor
attitudes. For example, the level of enthusiasm implied by a price rise on
low volume is not nearly as strong as that implied by a similar price advance
accompanied by very high volume.
The fourth dimension, breadth, measures the extent of the emotion. This
is important because as long as stocks are advancing on a broad front, the
trend in favorable emotion is dispersed among most stocks and industries,
indicating a healthy and broad economic recovery and a widely favorable
attitude toward stocks in particular. On the other hand, when interest has
narrowed to a few blue-chip stocks, the quality of the trend has deteriorated,
and a continuation of the bull market is highly suspect.
Technical analysis measures these psychological dimensions in a number
of ways. Most indicators monitor two or more aspects simultaneously; for
instance, a simple price chart measures both price (on the vertical axis) and
time (on the horizontal axis). Similarly, an advance/decline line measures
breadth and time.
Part III, "Other Aspects of Market Behavior," deals with more specialized
aspects. These include interest rates and the stock market, sentiment, auto-
mated trading systems, individual stock selection, and technical analysis as
applied to global markets.

Financial markets move in trends caused by the changing attitudes and
expectations of investors with regard to the business cycle. Since investors
continue to repeat the same type of behavior from cycle to cycle, an under-

J'I" for no
~1 .... It

The Market
Cycle Model
In the Introduction, technical analysis was defined as the art of identifying
a trend reversal at a relatively early stage and riding on that trend until the
weight of the evidence shows or proves that the trend has reversed. In order
to identify a reversal, we must first know what a trend is. This chapter
explains and categorizes the various trends, and concludes with a discussion
of one of the basic trend-determining techniques, peak-and-trough pro-
gression. It is one of the simplest, and perhaps the most effective, trend-iden-
tification techniques used in technical analysis and forms a building block
for many of the other techniques discussed later.

Three Important Trends

A trendis a time measurement of the direction in price levels covering dif-
ferent time spans. There are many trends, but the three that are most widely
followed are primary, intermediate, and short term.

The primary trend generally lasts between 9 months and 2 years and is a
reflection ofinvestors' attitudes toward unfolding fundamentals in the busi-
ness cycle. The business cycle extends statistically from trough to trough for
approximately 3.6 years, so it follows that rising and falling primary trends
(bull and bear markets) last for 1 to 2 years. Since building up takes longer
than tearing down, bull markets generally last longer than bear markets.


- - - - - ... --""----~~~~-.--<-- - - - -
The Mlll'ket Cycle Model 17

. The primary trend cycle is operative fOT bonds. equities. and commodi- It is important to have an idea of the di rection and maturity of the pri-
~es. Pm~~ trends also apply to currencies, but since currencies reflect mary trend. but an analysis ofintermediate trends is also helpful for improv-
inves to rs .attltudes toward the inte~lationshipof two different economies ing success rates in trading. as well as for detennining when the primary
an analysIS of currency rdationships does not fit neatly into the business movement may have run its course. '
cycle approach discussed in Chapter 2.
.The I?rim~ trend is illustrated in Fig. I-I by the thickest line. In an ide.
ah:zed snuauon, the primary uptrend (bull market) is the same size as the Short Tenn
p~ downtrend (bear market) , but in reality, ofcourse, their magnitudes
are different, Because .it is very ~ponant to position both (short-term) Short-term trends typically last from 2 to 4 weeks, sometimes sho rte r and
u;ades and (Iong-t~rm) m~tmentsIn th e direction of the main trend, a sig- sometimes longer. They interrupt the course of the intermediate eycle,jwt
nificant part of this bookIS concerned with identifying reversals in the pri-
as the intermediate-term trend interrupts primary pri ce movements. Short-
mary trend. term trends are shown in the market cycle model (see Fig. 1-1) as a dotted
line . They are usually influenced by random news events and are far more
difficult to identify than their intermediate or primary counterp arts.
~yone.who .has loo k.c d at a pri ce chan will notice that prices do Dot move
m a stnught line, A pnmary.upswing is interrupted byseveral reactions along The Market Cycle Model
the way. These countercyclical trends within the confines of a primary bull It is apparent by now that the pri ce level of any market is influenced simul-
6market are known as i~ r-r-« hft~ ~_ •• Th
" .,.."..,,~ ey 1asr anywh ere ,rrom taneously by several different trends, and it is important to understand which
weeks t~ as lo ng as 9 months, sometimes even longer, but rarely shoner. type is being monitored. For exam ple , if a reversal in a short-term trend has
Inte~cllate-.term trends of the stock market are examined in greater just taken place, a much smaller pri ce movem en t may be expected than if
detail In Chapter 4, and are shown as a thin solid line in Fig. 1-1.
the primary trend had reversed.
Long-term investors are principally concerned with the dire ction of th e
primary trend, and thus it is important for them to have some perspective
on the maturity of th e prevailing bul l or bear market, However, long-term
mwst!m must also be awafl' oj inle'mudiate ttrm and. 10 a less" extent, short-term
tmuls. This is because an important step in the analysis is an e xamination
and understanding of the relationship between sfrort- and intermediate-
term trends, and how th ey effect the primary trend. Also, if it is concluded
that th e long-term trend has just reversed to the upside, it may pay to wait
before committing cap ital because th e sho rt-term trend is overextende d o n
the upside . A lack of knowledge of the sho rt-te rm trend's position by an
Prtmary trend _ investor could th erefore prove costly at th e margin.
' _""1,..,. Slwrt-tmn tmdm are principally co nce rn ed with smalle r movem ents in
pric e, but they also nnd to bow the dirn:tion oj the inttrmediaU and primary
trmds. Thi s is because surp rises oc cur on the upside in a bull market and
Approldmately .. yellrs

A1JlUe 1·1 The market cyde model

~pted from an kiea firs! brought to my Illlention by Ian S. Notley of Vehon AsceI, RIdgefield,

-I >
, ,.
18 19
-on the downside in a bear market, In other words. ming short..(enn trends j
• •
~thin the confines of a bull market are likely to be muc h greater in mag-

nirude than short-term dcwntrends and vice versa, A trading loss usually hap-
pens ~we the trader is positioned in a countercyclical position against
~e mam trend. In effect. all Ma1td participanu rwd to Aaw SOMt tiM offIK11i•
•ng 1tMwltdgr of all tJrm hmds. although the emphasis wiU depend on
whether their orientation comes from an investment or a short-term trad-
ing perspective'.

Intraday Trends
In recent yean, computers and real-time trading have enabled traders to
identify h ourly and even tick-by-tick. movements. 1M principles of t«hniaU
an4lym applJ equ.a/lJ to tJuu vn"J jhorl-unn rrwvt7IImU and 1mjust as valid. There ,,
ace two main differences. Fint, reversals in the intraday charts have only a 'Cl,I:
very short term implication and ace not significant for lc nger-eerm price 8uI1 mel'_ a'I Iar'o9" In 8ear mel' ''''' a.. Iclnot' '"
Mal ia' dawn phalt
MeU!O, Ull phMt
reversals. Second, extremely short-term price movements are much mo re
influenced by psychology and instant reactions to news events than are Agun 1-2 The relatiooshlp between the secuJ"r end primary tTends.
longe.Ncnn o~es. Decisions therefore have a tendency to be emotional, i'
kne~erk reacucns. Intraday pri ce action is also more susceptible to manip-
ulatlon. ~ a consequence, price data used in very sho rt-term charts are
much more erratic and generally less reliable than those that appear in the Peak-and-Trough Progression
longer-term charu. Technical analysis, as pointed ou t before , is the an of ide ntifying a (pri ce)
. trend reversal based on the weight of the evide nce. As in a coun of law, a
trend is presumed innocent until prove n guilty. The evidenc e is the objec-
The Secular Trend tive element in technical analysis. It consists of a series of scien tifically
derived indicators or techniques that work well mosL.Or the time in the trend-
The primary trend consis ts of several intermediate cycles, but the secular, identificati on proc ess. The an consists of combi ning these indicators into
or ~ry long-term, tre nd is constructed from a numbe r of primary trends. an overall picture and recognizing when that picture rese mbles a market
Th is supe r.cycle. or lon g wave, extends over a substantially greater' period. peak or tro ugh.
wually lasting- well over 10 yean, and often as long all 25 yean. It is discussed Widespread \UC of compu ters has led to the dc:velopment of some very
more fully in Olapter 2. A diagram of the interrelatio nship between a 5«- sophisticated trend-ide ntifiation techniques in market analysis. Some of
ular and a primary trend is she wn in Fig. 1-2. these indicators work reason ably well. but most do noLThe continual search
It is cena.inlyvery he lpful to understand the direction of the secular trend for the -Holy Grail,- or perfect indicator, will undoubtedly continue , but it
Jwt as the primary trend influences the magnitude of the Imermedtate-rerm is unlikely that such a technique willever be discovered. Even ifit were , news
r.a.lly relative to the countercyclical reaction. so the secular trend influences of its discovery would soo n be disseminated and the indicator would grad-
the ~agni.~de and duration of a primary trend rally or reaction. For exam- ually be discounted. .
ple, 10 a rumg secular trend, primary bu ll markets will be of greater mag- In th e qUe5t for sophisticated mathematical te chniques, some of the am-
nit ude than primary bear markets. In a secular downtrend, bear markets will plest and most basic techniques of technical analysu are often overlooked.
be more powerful and will take longer to unfold than bull markets. One simple , but basic technique that has been underused is peak-and-


20 The ~ Cycle Model

. trough progressi on (see Ch an 1·1) . which rdalel to Charles Dow's o riginal

observation th at a rising market moves in a k rin of waves, eac h rally an d
reac tion bei ng higher than its predecessor. When th e series of rising peaks
an d troughs is interrupted. a trend reversal is signaled. To explain th is
approach. Dow used an analogywi th the ripple effect O ( \¥;IveI on a seasho re.
He po inted o ut thatjwt as it was possible for someone o n the beach to iden-
tify th e turning of the tide by a reversal of receding wave action at low tide,
so the same o~rctiYe could be achieved in th e market by observing th e price
action .
In Fig. }.g, the price has been advancing in a series of waves, with each
peak and trough reaching higher than its predecessors. Then, for the first
time, a rally fails to move to a new high . and the subsequent reaction pushes
it be low the previous trough. This occurs at point X and gives a signal that
the trend has reversed, Figure 1-4 shows a similar situation, but th is time
the trend reversal is from a downtrend to an uptrend.
The idea of the interruption of a series of peaks and troughs is the basic
building block for both Dow theory (see Chapter 3) and price pattern anal)'"
sis (see Chapter 5 ).

16 Moody's AM. Bond Ytelds t..
j .

. '''' ~

. . ks to com Jete each wave in a series of

For exam ple , ott takes 2 to S wee al 'U ~ an interm ediate one, since
rallie s and reac?ons, th e trend rev~" f WI ries ofsho rt-term (2- to 3-week)
interm ediate pnce movements co nsist 0 a sc
. . .\ I th ' terruption 0 r a senes
. or falling
fluctuauons. Simi ar y, e m . als 1fro m a primary bear to
peaks and troughs by a rising one Si gn a reverse
' .' • 1m IMlI a primary bun market.

22 Part I: Trend-OeterminIn T~

. A Peak-and-Trough DI1emma The dilemma is probably best deal t with by referring b-a to the second
half of the definition of technical analysis given at the beginning oC this
Occasionally, peak-and-trough progression b«omea more complicated than cha pter. "and riding that trend until the wight oftheeuidrnu shows or proves
~e ~p~el5h~ in Fi.~ I·! and l..-t. In Fig. 1-5(0), the market has been that it has been reversed,"
dvan ang ~ a ~n~ of rumg peaks and troughs, bu t fo llowing the hi hest In this case. if the ~ight oC the eviden ce- from other technical indica-
peak" ~e pri ce declines at point X to a leYellhat is below the Previo~ low tors. such as maving avenges (MAs). volume, momentum. and breadth (dis-
At ~J~~turc. the series or rising troughs has Moen broken but MI the cussed in later chapters), overwhelmingly indicates a trend reversal, it is
senesofnsmg-h In 0 lh er WO'lD,
...... at pomt
' X. on ly half a signal
• has be e
r---'" probablysafe to an ticipate a chan ge in trend. even though peak-and-tro ugh
generated. The complete signal of a reversal of both ris . 1. _ nd progression has not com pletely confirmed the situation. It is still a wise pol-
teo hs ari . mg pea..,. an
ug ames, at po m t Y, when the price slips below th e level previo ly icy, though. to view this signal with some degree of skepticism un til the rever·
reached at po int X W
sal is confirmed by an interruption in both series of rising peaks as well as
swAt point ": ~ere is quite a dilemma because the trend should still be clas-
be ed:u posiuve, ",-?d ~t the very fact that the series of rising troughs has
en interrupted mdlca~es underlying technical weakness. On the one
Figure I .-5 (b) &bows this type oCsituation for a reversal from a bear to bull
trend. The same principles of interpretation apply at point X. as in Fig.
~~. W; arc ~resented wah half a bearish signal, while on the other hand 1-5(6). Occasionally. the determination of what constitutes a rally or reac-
~ang~rdpo~nt Ywould mean giving up a substantial amount of lh e prof. tion becomes a subjective process. One way around this problem is to choose
Its eamec unng the bull market,
an ol:!jective measure such as categoriring rallies greater than. say. 5 percent.
This can be a tedi ous process. but som e software programs (such as
MetaStoek. with its zigzag tool ) enable the user to establ ish such benchmarks
almost instantly in a graphic format.

What Constitutes a Legitimate

Peak and Trough?
Most of the time , the vario us rallies an d reactions are self-evide nt, so it is
easy to determine that these turning points are: legitimate peaks an d tro ughs.
Technical lore has it that a reac tion to the prevailing tre nd should retrace
one-third to two-thirds of the previous move. Thus. in Fig. 1-6the first rally
from the trough low to the subsequent peak. is 100 percent. The ensuing
reaction appea.n to be just over a half or a 50 percent reuacemem of the
previous move. Occasionally. the reuacemen t ca n reach 100 percen t.
Technical anal ysis is far from precise , bu t if a retracem em move is a good
deal lese than the minimum one-third , then the peak or trough in qu estion
is held to be suspect.
You can appreciate that a line is a fairly controlled period of profit tak-
ing or digestion of losses. The depth of th e trading range can fall short of
the minimum "approximate one-third retracem ent" requirement and , in
such ins tances, the correction qualifies mo re on the basis of time than mag-
nitude . A rule of thumb might be for the corre ction to last at least one-third
to two-thirds of the time taken to achieve the previous advance or decline.
f'igunl 1·5 HaIf-slgoal ~. In Fig. 1·7 the time distance between the low and the high for the move


24 Par1 I: Trend-Dete:rmlnlng Tec:hnlques 25

. represents 100 percent. The co nsolida tio n prio r 10 th e brea kout should con- These are only rough guid elines. and in m e final an alysis it is ajudgmen t
stitu te at least o ne-third to two-thi rds of th e tim e taken to ac hieve the call based on expe rience. common sense, a bit of intuitio n, and. perhaps
ad~ce. am ple time t~ co~sotidate gains, and move on to a new hig h . It's most important ofall. a review ofo th er facto rs s!Jch 35 volum e, support, rem-
possible for the consolidauon to con stitute mo re than 100 percent of the ranee principles. and so o n. We have main ly be en studying th ~ conce~ts
p receding price movement, In fact, the larger th e co nso lidation, th e greater in a rising trend. H~ r. th e principles work exactly the same In a decli n-
the hustle between buyers an d sellers an d th e mo re significant the upper ing trend in that ral lies should re trace o ne-third to tw<Hhirds of th e ~~
and lower boundaries become.
ow decline. Also. lines o r consolidations should take at least one-third of
th e duration of the previous decline . .
It is also im portant to categorize which kind of trend is be ing mo nitored.
Obviowly. a reversal derived fro m a series of ral lies and reactions ea ch last-
ing, say. 2 to S ~b would be an in termed iate reversal. This is because th e
swings would be short te rm in nature. On the o ther hand. peak-and-trough
resersals that develop in intraday charts are likely to have significance ?vc r
a much sho rter period. How short would depend on whether the SWIngs
were a reflection of ho urly or. say. 5-minute ban.

• A number of different trends simultaneously influence th e price level of
any security.
• The th ree most important trends are primary. intermediate, an d sho rt
• The p rinciples o f technical analysis apply to intraday trends, but since they
are more random in nature. th e analysis is generally I~ reliable than for
longer-term trends.
Ret..-cemenl is r.r • Very long term. o r-secular; trends influe nce th e magnitude of p rimary bull
and be ar trends.
less tIlMl one-thinl
to ~thiUs of the • • Peak.-an d-uough progression is the most basic trend-identific:atio n tec h-
prevlous tlIly
nique and is a basic building block of technical analysis.
er-t h~ or mont • As a general ru le. in order to qualify as a new legitimate peak o r trough,
of the lime is Iaklln
10 form tile fely the p rice should retrace between one-th ird an d two-thirds of th e p revi-
ous move.
• lines o r consolida tions also qualify as pew and troughs where th ey form
between o ne-th ird and two-th irds of the time taken to p roduce th e pre--
vious advance or d eclin e.

Figure 1·7 IdentlMng peaks and troughs (time).

L t
Financial Markets and the Business Cycle 27

I ate assets, with the result that a market normally reaches a major turning
point well ahead of the actual development;
An expanding level of economic activity is usually favorable for stock

2 prices, a weak economy is bullish for bond prices, and a tight economy is
favorable for industrial commodity prices. These three markets often move
in different directions simultaneously because they are discounting differ-
ent things.

Financial An economy is rarely stable; generally, it is either expanding or con-

tracting. As a result, financial markets are also in a continual state of flux.
A hypothetical economy, as shown in Fig. 2-1, revolves around a point ofbal-

Markets and the ance known as equilibrium. Roughly speaking, equilibrium can be thought
of as a period of zero growth in which the economy is neither expanding
nor contracting. In practice, this state of affairs is rarely, if ever, attained,

Business Cycle since an economy as a whole possesses tremendous momentum in either

the expansionary or the contractionary phase, so that the turnaround rarely
occurs at an equilibrium level.
In any event, the "economy" consists of a host of individual sectors, many
ofwhich are operating in different directions at the same time. Thus, at the
beginning of the business cycle, leading economic indicators, such as hous-
ing starts, might be rising, while lagging indicators, such as capital spend-
Introduction ing or employment levels, could be falling. Investors in financial markets
are not concerned with periods of extended stability or equilibrium, for such
The basic concern of this book is the technical approach, but it is also impor- periods do not produce volatile price swings and opportunities to make
:nt to und.erstand that primary trends of stocks, bonds, and commodities quick profits. The ever-changing character of the economic cycle creates
b .determmed by the attitude of investors toward unfolding events in the tremendous opportunities for investors and traders because it means that
u:'mess c~cle. Each ~arket has a tendency to peak and trough at different different industries are experiencing different economic conditions simul-
pomts dun~g the business cycle in a consistent, chronological manner An taneously. Since housing leads the economy, housing stocks do well at the
understandI~g of the interrelationship of debt, equity, and comm~dity start of the recovery, when capital-intensive stocks such as steel are still under
markkets provides a useful framework for identifying major reversals in each pressure. Later in the cycle, the tables are turned and it's housing that peaks
mar et.

Economy is growing

The Discounting Mechanism

of Financial Markets
The tre~d of all financial markets is essentially determined b investors'
expectations of movements in th h Y
Iik I h . e economy, t e effect those changes are
~ elY to ave on the pnce of the asset in which a specific financial market
eacs, and the psychological attitude of investors toward these fundamen-
tal lac tors. Market participants typo all .. Economy is contracting
fi . Id IC Y anucipata future economic and
lDanCIa evelopments and take action by buying or selling the appropri- Figure 2-1 The Idealized business cycle.

Financla1 Markets and the Business Cycle
28 Part II Trend-Determining Techniques

first. This situation gives rise to the group rotation process, which is dis-
cussed at length in Chapter 19.
Since the financial markets lead the economy, it follows that the greatest
profits can be madejust before the point of maximum economic distortion,
or disequilibrium. Once investors realize that an economy is changing direc-
tion and returning toward the equilibrium level, they discount this devel-

til opment by buying or selling the appropriate asset. Obviously, the more
dislocated and volatile an economy becomes, the greater is the potential for
a return toward the equilibrium level, and also for a strong swing well
~ 1s
beyond it to the other extreme. The risks are also greater ifyou are too early. B
Under such conditions, the possibilities for making money in financial mar-
kets are greater because they too will normally become subject to wider price
Figure 2-2 The idealized business cycle and fmancial market turning points. (B - Bonds;
S _ Stocks; C - Commodities)

Market Movements and Bonds

the Business Cycle Commodities
The major movements of interest rates, equities, and commodity prices are
related to changes in the level of business activity. Please note that the term
commodity prices refers to industrial prices that are sensitive to business con-
ditions, as opposed to weather-driven commodities such as the grains.
Figure 2-2 represents a business cycle, which ranges from 3 to 5 years
between troughs. The horizontal line reflects a level of zero growth, above

II which are periods of expansion, and below which are periods of contrac-
tion. After the peak is experienced, the economy continues to grow, but at
a declining rate, until the line crosses below the equilibrium level and con-

traction in economic activity takes place. The arrows in Fig. 2-2 show the
idealized peaks and troughs of the financial markets as they relate to the
business cycle. Figure 2·3 Idealized sine curves for three markets.

Periods of expansion generally last longer than periods of contraction,
because it takes longer to build something up than to tear it down. For this
reason, bull markets for equities generally last longer than bear markets. Referring back to Fig. 2-2, we can see that the bond market is the first
The same could be said for interest rates and commodities, but in this case financial market to begin a bull phase. This usually occurs after the growth

the magnitude and duration of a primary trend depend on the direction of rate in the economy has slowed down considerably from its peak rate and
the secular trend, as discussed in Chapter 1. quite often is delayed until the initial stages of the recession. Gener~ly
Figure 2-3 shows how the three markets of short-term interest rates, com- speaking, the sharper the economic contraction, the greater the ~otenUal
modities, and equities also relate to the typical business cycle. In the exam- for a rise in bond prices (that is, a fall in interest rates). Alternauve~y, the
ple, interest rates have been plotted inversely to correspond with bond stronger the period of expansion, the smaller the amount of econormc ~nd
prices. A bull market for bonds is marked by a rising line and a bear mar- financial slack, and the greater the potential for a decline in bond pnces
ket by a descending one. (and a rise in interest rates) .

• L ...
30 31

Following the bear market low in bo nd prices. economic activity begins these the m stQgf:S and they can be used as reference points for determin-
to contract more sharply. At this poi nt. participan ts in th e eq uity market are ing the current phase of the business cycle . The six stages are indicated in
able to "'ook th rough- the valley in corpora te profits. which are now declin- ~. U . .
ing sharply because of the recessio n. and begi n accumulating stocks . When identifying a stage. it is important to look at the long-term techm-
Generally speaking, the Jonger the lea d between the low in bonds and that cal position of all three markets 50 that they can act as a cross-check on each
of stocks, the greater the potential for the stock market to rally. This is other. The stages are also useful in that specific groups do well at parti cu-
because the lag implies a panicularfy wak recession in which extreme co r- lar times. For example, liquidity-driven or early-cycle leaden tend to out-
porate bel l tightening is ~Ie to drop bre ak-even levels to a very low level. perform the market in Stages 1 and 2 when bond prices arc rising and
During the recovery, increases in revenue are therefore able to q uickly move interest rates falling. On the other hand, eamtngs-dnven or late-cycle lead -
to the bottom line. en pe rfo rm well in Stages 4 and 5 when commodity prices are rallying.
Mter the recovery has been under way for some time . capacity starts to These aspects are covere d more fully in Chapter 19 on group rotation.
tighten, resource-based companies fed some pricing power return, and
commodity prices bottom. OcC3.5ionally after a co mmod ity boom of unusual
magnitude, industrial commodity prices reach their actual bottom during
the recession due to severe margin liquidation on behalf of speculaton. Longer Cycles
H~r, this bottom is ofte n subsequently tested.; a sustainable rally only Some expansions encompass m uch longer periods, and they usual ly include
begins after th e recovery has been un derway fora few months. At this point, at least o ne slowdown in the growth ra te followed by a seco nd round of eco-
I all th ree fina ncia l marke t! are in a rising tre nd .
Grad ually. the economic and financial slack, which developed as a result
nomic expansion. This has the effect of splitting the overall expansion into
two or three parts. each of which results in a complete cycle in the finan -
of the recession, is substantially absorbed, putting upward pressure on the cial markets. I call this a tWuble CJCk. An example of this phenomenon is illus-
pric e of ~redit. that is. interest rates. Since rising interest rates mean falling trated in Fig. 2-5. j' .
bond pnces. the bond market peaks out and begins its bear phase. Because
~~e excess pant and labor capacity still exists, rising business activity remits
ID improved productivity and a continued positive outlook. The stock mar-
~et disco un ts trends in corporate profits , so it re mains in an uptrend until

investors sense that th e economy is becoming overheated and the potential Idealized Busiless Cycle for the Six Stages d the Business Cycle
for an improvement in profits is very low. At this po int, there is less reason
...,., ..... ' ..... ' ...... . ...... ......
to hold equities, and they in tum enter into a bear phase. Later. the rise in
interest roues takes its toU on the economy and commodity prices begin to
slip. Bood.

O nce this j uncture has been reac hed. all th ree financial markets begin
to fall. They will conti nue to declin e un til the credit markets bottom OUL
This final stage , which develops around the same time as the beginning of
the recession, is usually associated with a free fall in prices in at least one of .\ R..-~
the flnanciaJ markets. If a panic is to develop. this is one of the most likely
po ints for it to take place.
l' Cf
Introducing the Six Stages
Since the re are three financial markets and each has two turning points, it
follows that there are conceptually six turning points in a typical.cycle. I caJl

1 Commodities

32 Put I: Trend-Determining Techniques Fmancial Markets and the BusinessCycle 33

"'lie S C Market Experience, 1966-2001

Chart 2-1 shows how peaks and troughs developed for the various m~rkets
between 1966 and 1977. Please note that inversely plotted short-term mter-
est rates have been substituted for bond prices. This is because there is a
much closer link between equity prices and short-term rates than wi~
longer-term rates, due in part to the fact that corporations do more of their
borrowing in the money markets than the bond markets. Also, market par-
ticipants buy stocks on margin, the Cost of which is determined by the level
of short-term rates. Short-term rates are also more volatile than those at the
end of the yield spectrum.
The peaks and troughs in Chart 2-1 turned out very much as expected.
B t Although the chronological sequence was more or less pe:fect, the leads
s and lags in each cycle varied considerably, because of the different charac-
teristics in each cycle. In 1966, for instance, bonds and stocks bottomed
Figure 2-5 Financial market peaks and troughs in a double cycle. more or less simultaneously, whereas the lag for the commodity market bot-
tom was well over a year. .
Chart 2-2 shows the same markets, but this time we are looking at the
A double cycle developed in the 1980s and another in the 1990s. In the
1980s. The two small, upward-pointing arrows in 1982 and 1990 reflect reces-
mid-1980s, for example, commodity and industrial parts of the country were
sions. The series of three bottoms that developed between 1984 and 1986
very badly affected, but the east and west coasts continued their expansions
reflects the mid-1980s growth recession. Generally speaking, the chrono-
unabated. The strong areas more than offset the weaker ones and so the
logical sequence works satisfactorily until we get to the late 19808where the
country as a whole avoided a recession. It is certainly true that the six-stage
concept works well in many cycles, but it must be noted that there are excep-
tions. I believe, therefore, that it is best used as a conceptual framework
rather than an ironclad discipline. For example, some of the worst excep- Chart 2-1 Three financial markets, 1966-1977. (From Intermarket Review.)
tions in over 200 years of recorded history occurred in the 1990s when the
U.S. equity market was experiencing an unprecedented equity rally, which
fatally distorted the normal chronological sequence.

The Role of Technical Analysis 110

Technical analysis comes into play by helping to determine when the var- 90
~ous markets have turned in a primary-sense way. This is achieved by apply- 80
mg the various techniques outlined in subsequent chapters, moving average 10
crossovers, changes in the direction of long-term momentum, and so
forth. Each market can then be used as a cross-eheck against the other two. 200
For example, if the weight of the technical evidence suggests that bonds

have bottomed but commodity prices remain in a bear market, the next
thing to do would be to look for technical signs pointing to a stock mar-
ket bottom. 911
[~ ..

Plrt I: Trend-DeterminingTechniques Financial Markets and the Business Cycle 35

. Chart 2-2 Three financial markets, 1980-1992. (From Intermarket RevIew.)

Chart 2·3 Three financial markets, 198~2001. (From lntermarket RevIew.)

",,0'''''''0" ~~
···1 . · · '\ .. \.
10 6 Od),,' ':.


, .r:
.. ' ' .
' "• - -,
• • •
200 .. - .»

.. ' .~ .... - I'

1989 bottom in rates is juxtaposed with the stock market peak.

~nfortunately, these out-of-sequence events are a fact of life. Fonunately, • Some cycles experience a slowdown in the growth rate and not an actual
10 the 20 years I have studied these relationships, they represent the excep- recession. Even so, the chronological sequence between the markets still
tion rather than the rule. appears to operate.
Chan 2-3shows the closing years of the twentieth century. This is the most • The leads and lags vary from cycle to cycle and have little forecasting value.
difficult period I have encountered because of the record performance by • The chronological sequence of peaks and troughs in the various finan-
the stock market and the strong deflationary forces associated with the tech- cial markets can be used as' a framework for identifying the position of a
nological revolution. This had the effect ofreducing the normal cyclicalfluc- specific market within its bull or bear market cycle.
tuation in the equity market. Since the stock market boom was
unprecedented, it is unlikely that the normal chronological sequences have
been more than temporarily interrupted.

~ Summary

It! • A typical business cycle embraces three individual cycles for interest
rates, equities, and commodities. All are influenced by the same economic
and financial forces, but each responds differently.
• These markets undergo a chronological sequence, which repeats in most

---------~_ ..._-_._--._-_.,---_ ..__...- .L~.
Dow Theory
\ 37
( ...;"; ~
',...1.,-:,;;/ .
It should be recognized that the theory does not al~ys keep pace.WIth ___
events; it occasionally leaves the investor i~ 40l!.bt and 1S by ~o means l~al- ) J:
lible, since small losses are sometimes inc{(rft'd. These .pomts emphasize <--'.::
that, while mechanical devices can be useful for-forecasting .the stock .mar- -

3 1,' I
ket there is no substitute for obtaining additional supportive analysis on
which to base sound, balanced judgment. Remember, in our weight-of-the-
evidence approach Dow theory is one piece of evidence.
The Dow theory evolved from the work of Charles H. Dow,who published

Dow Theory a series of The Wall Stree: Journal editorials between 1900 and 1902 con-
cerning market action. Dow used the behavior of ~he stock ma:ket as a
barometer of business conditions rather than as a basis for forecastmg stock
prices themselves. His successor, William Peter Hamilton, .developed Dow's
principles and organized them into something approachmg the theory :u'
~h~ Dow ~eory is th~ oldest and by far the most ~~licized method of iden- we know it today. These principles were outlined rather loosely in
tifying m'll0r trends m the stock market. An extensive account will not be Hamilton's book The Stoek Market Barometer, published in 1922. It was not
necessary here, as there are many excellent books on the subject. A brief until Robert Rhea published Dow Theory in 1932 that a more complete and
explanation, however, is in order because the basic principles of the Dow formalized account of the principles finally became available. .
theory are used in other branches of technical analysis. The theory assumes that the majority of stocks follow the under?,m g trend
The goal of the theory is to determine changes in the primary or major of the market most of the time. In order to measure "the market, Dow con-
movement of the market. Once a trend has been established, it is assumed structed two indexes, which are now called the DowJones Industrial Average
to exist until a reversal is proved. Dow theory is concerned with the dime- and the DowJones Transportation Averag&5. The Industrials were originally con-
tion of a trend and has no forecasting value as to its ultimate duration or size. structed from 12 (but now include 30) blue-chip stocks and the Dow Jones
Starting in 1897, an investor who purchased the stocks in the DowJones Rail Average, comprising 12 railroad stocks. Since th e Ra1'l A:.:v~~ge~_. '7 v'\.r"~)'
Industrial Average (DJIA) following each Dow theory buy signal, liquidated intended as a proxy for transportation stocks, the evolution of aviatiort an?
the position on sell signals, and reinvested the money on the next buy sig- other forms of transportation has necessitated modifying the old Rail
nal would have had his or her original investment of $44 in 1897 grow to Average in order to incorporate additions to this industry. Consequently, the
about $51,268 byJanuary 1990. 1 If, instead, the investor had held onto the name of this index has been changed to Transportation Average.
original $44 investment throughout that period, the investment would also
have grown, but only to about $2500. In reality, the substantial profit eamed
by following the Dow theory would have been trimmed by transaction costs ..J'--'U
and capital gains taxes. Even if a wide margin for error were allowed, and Interp'reting the Theory
errors can certainly arise in the interpretation of the theory, the investment
In order to interpret the theory correctly, it is necessary to have ~ record of
performance using this approach would still have been far superior to the
the daily closing" prices of the two averages and the total of daily transac-
results of a buy-and-hold strategy. The theory will still have done well in the
tions on the New York Stock Exchange (NYSE). The six basic tenets of the
period between 1990 and 2001. However, the spectacular bull market of the
1990s would not have made the comparison between the theory and the buy- theory are as follows.
and-hold approach particularly inspiring for Dow theorists.

. 'Thi~ assumes that the averages were available in 1897. Actually, Dow theory was first pub- 'It is important to use closing prices, since intraday fluctuations are more subject to
hshed m 1900. manipulation.

J 36
,£... 'W

38 Part I: Trend-Determining Techniques Dow Theory

1. The Averages Discount Everything since the termination of the last preceding secondary reactio~.".8 (My own

-, Changes in the daily closing prices reflect the aggregatejudgment and emo-
tions of all stock market participants, both current and potential. It is there-
fore assumed that this process discounts everything known and predictable
that can affect the demand/supply relationship of stocks. Although acts of
view is that a secondary or intermediate move should l~t a mrmmum of. 4
weeks.) This relationship is shown in Fig. 3-1(e) for a nsmg market and m
Fig. 3-1(b) for a declining one. .
Occasionally, a secondary reaction can retrace the whole of the preVl~us
rimary movement, but normally the move falls in the one-half to t,wo-thlr~
God are obviously unpredictable, their occurrence is quickly appraised and ~rea, often at the 50 percent mark. As discussed in greater detail later l?
-J their implications are discounted. the chapter, the correct differentiation betw~en the ~~t leg of a new,Pn-
mary trend and a secondary movement within the existing trend proVldes
Dow theorists with their most difficult problem.
2. The Market Has Three Movements
1 There are simultaneously three movements in the stock market.
Minor Movements The minor movement lasts from a week or two.up to as
long as 6 weeks. It is important only in that it forms part of ~e pnmary ~r
Primary Movement The most important is the primary or major trend, secondary moves; it has no forecasting value for longer-term mv~stors. ~t~
more generally known as a bull (rising) or bear (falling) market. Such move- is especially important since short-term ~ovements can be manipulate
ments last from less than 1 year to several years. some extent, unlike the secondary or pnmary trends.
A primary bearmarketis a long decline interrupted by important rallies. It
1 begins as the hopes on which the stocks were first purchased are abandoned.
The second phase evolves as the levelg of business activity and profits
decline. The bear market reaches a cl:ti?,J when stocks are liquidated

1 regardless of their underlying value (because of the depressing state of the

news or because of forced liquidation caused, for example, by margin
calls). This represents the third stage of the bear market.


A primarybull marketis a broad upward movement, normally averaging at

least 18 months, which is interrupted by secondary reactions. The bull mar-
ket begins when the averages have di~9UI!~d the worst possible news, and
confidence about the future begins to revive. The second stage of the bull
market is the response of equities to known improvements in business con-
ditions, while the third and final phase evolves from overconfidence and
speculation when stocks are advanced on projections that usually prove to
be unfounded. (s) (b)

Figures 3·1 Secondary retracements.

Secondary Reactions A secondary or intermediate reaction is defined as "an
important decline in a bull market or advance in a bear market, usually last-
ing from three weeks to as many months, during which interval, the move-
ment generally retraces from 33 to 66 percent of the primary price change "Rhea, Robe~. DowTheury. NewYork:Barron's, 1932.

40 PBJ1 1: Trmd-DetennlnJng T~ 41

3. Una Indicate Movement

Rhea defined a line as -3. price mo~ment 2 to 3 weeks or lon ger, during
which period., the pri ce variation of both averages moves within a range of '00 x
app roxima tely 5 pe rcent (of their mean average ). Such a movement in di- "'I'
~ ,'


cates eithe r accum ulati on [stod motIing imo Jtnmg and Irnowledgm& hands ;
and tkrrforr bull:ishJ or distribution (stod: 1Mf!ing into Il¥al hands and tMrrforr :'f
I bearirh)....
An advance above- th e limits of the line indicates accumulation and pre--

diets higher prices, a nd vice 'Ve1U. When a line OC CU I'$ in the mid dle of a , III ... III " " ... SO" D
" Fill ... III " "
primary advance, it is really forming a horizontal secondary movement and
should be' treated as such. . 11' tf'.:.:.- (0) (1))
Myown view is that the forma tion DCa le8itima~ line should p roba bly
take longer than 2 to 3 weeks with, say. a minimum of 4. Mer all, a line is
l"C'ally a substitute for an Intermediate price trend and 2 to 3 Wttb is the
' 00
, .
time for a short-term or mino r price movement..
", ''''
li ~
4 . PriceNolume Rela tionships Provide Background '" ' 00

Th e nonnal rd ationship is for volume to expand on rallies an d contract on

dechnes, If volume becomes dull on a price advance and expands on a
decli ne . a warni ng is given tha t the prevailing trend may soon be reversed.
, .. .. III " "

••e • 0
,, .
• •• , , • so ..

Thil principle should be used as background information onJy. since the
ronclU5ive evidence of trend reve rsals can be gWen on ly by the price of the PrYnary trend~ .
respective averages,

decline tales the average below Its low point, confirming a .bear market as
5. I'rlcc ActIon Determines the Trend it does so, at point X In Fig. 3-2( b), following the third pea k IRthe ~un mar-
Bullish ind ications are given when successive rallies penetrate pe aks while ket, a bear market is indicated as the avera~e falls below tJ:te pTeVlOUli ':~i
ondary trough. In this instance, the preceding secondary 15 part of a
the trough of an Inte rvenin g decline is above the precedin g trough.
market, no t the first trough in a bear market, as shown in Fig. 3-2(a) . Many
Conversely, bearish ind ications come fro m a series of declining pe aks and
troughs. Dow theorists do not cons ider pe netranon at poin t Xi n Fig. 3-2(b) to be a
sufficient indica tion of a bea r market. They prefer to take a .mor~:mse rv­
Figu res 3-2(a)-(d) show a theoretical bull trend int errupted by a sec-
ative position by waiting for a rallyand a subseque nt pe netration 0 at pre-
onda ry reac tion. In Fig. 3-2( a), the in dex makes a series of th ree pew and
vious tro ugh marked as point Yin Fig. 3-2(6). . ..
troughs, each higher than iu respective predecessor. Th e index rallies fol-
. In such cases, it is wise to approac h the int erpretation with additional cau-
lowing the I.hird declin e, but is unable to surpass its third peak. The next
tion If a bearish in dication is given from the volume patterns and a .cl~arly
. ~/
I" ~ .-
ide~tifiable speculative stage for the bull market has alrea~y ma~enahzed,
' Ib id. it is pro bably safe to assume that the bearish indication JS valid. In the
absence of such charac teristics, it is wiser to give the bull market the bene-

42 Part I: Trend-Determining Techniques Dow Theory 43

~- /p-'i:-~
. fit of the doubt and adopt a more conservative position. Remember, tech- lived. On the other hand, the largest primary swings usually develop when
nical analysis is the art of identifying trend reversals based on the weight of the characteristics of a third phase are especially m~~d during th.e p~e­
the evidence. Dow theory is one piece of evidence, so if four or five other ceding primary movement. Hence, the excessive Dou6 of specuI~tIon m
indicators are pointing to a trend reversal, it is usually a good idea to treat 1919, 1929, 1968, and 2000 in the NASDAQ were followed by particularly
the half signal at point Xas an indication that the trend has reversed. Figures sharp setbacks. Intermediate-term ~ovements are discussed more exten-
3-2(c) and (d) represent similar instances at the bottom of a bear market. sivelyin Chapter 4.
The examples in Figs. 3-3(a) and (b) show how the primary reversal
would appear if the average had formed a line at its peak or trough. The
importance of being able to distinguish between a valid secondary correc- 6. The Averages Must Confirm
tion and the first leg of a new primary trend is now evident. This is perhaps
the most difficult part of the theory to interpret, and unquestionably the One of the most important principles of Dow theory is that the movement
most critical. of the Industrial Average and the Transportation Average should always be
It is essential to establish that. the secondary reaction has retraced at least considered together (that is, the two averages must confirm each other).
one-third of the preceding primary movement, as measured from the ter- The need for confirming action by both averages would seem fund~­
mination of the preceding secondary. The secondary should also extend for mentally logical, because if the market is truly a barometer o~ future bUSI-
at least 3 to 4 weeks. ness conditions, investors should be bidding up the pnces bo~ of
Vital clues can also be obtained from volume characteristics and from an companies that produce goods and of companies that transport th~m
assessment of the maturity of the prevailing primary trend. The odds of a expanding economy. It is not possible to have a he~lthy ~conomy in which
major reversal are much greater if the market has undergone its third phase, goods are being manufactured but not sold (that IS, shipped to market).
characterized by speculation and false hopes during a primary upswing or This principle of confirmation is shown in Figs. 3-4(a) and (b).
a bout of persistent liquidation and widespread pessimism during a major In Fig. 3-4(a), the Industrial Average is the first to signal a bear. trend
decline. A change in the primary trend can occur without a clearly identi- (point A), but the actual bear market is not indicated until t?e
fiable third phase, but generally such reversals prove to be relatively short- Transportation Average confirms at pointB. Figure 3-4(b) shows t?e begin-
ning of a new bull market. Following a sharp decli~e, the Industnals make
a new low.A rally then develops, but the next reaction holds above the pre-

300 300



(a) (b) ·'{b)
Figures 3·3 Lines being formed at a peak or trough.
Figure 3-4 Dow Theory requires both averages to confirm.


44 Part I: TnlnlI-Detennlni TedWq.Jes 45

viow low. Wh~n prices p ush above the preceding rally. a bull signal is given
Additional Considerations
by the lnd~trials at po mt A. In the meantime, the Transporta tion A~~
mUes a series of two succeeding lows. The question that arises is: Which Dow th eory does not specify a time period beyond which a co nfinnation of
average is correctly ~prnenting the prnatl ing U'ttId? Since it is alw;ays o ne average by the other becomes invalid. Generally. the closer th e confir-
~umed th.1.t a trend is in existence un til a reeersal is proved, the co nclu- ma tion, the stronger the follo wing move is like ly to be. For example, con-
SIO? sho uld be drawn at this point that th e Transportation Avt r.tge is indi- fmnatio n of the ] ~1 932 bear mark et was given by the Rail Average j ust
Clung th e correct outcome. ) day after the In d ustrial Avera ge. The sharp 1962 bre ak was confirmed on
It ~ on ly when this average exceeds th e peak of th e preceding secondary the same day. ::..'., ~ \
~t POint B that a new bull market is co nfinned by both averages, resulting One of the maj or criticisms of Dow th eory is that many of its signals have
In a Dow theory buy signal. Th e movement of one average unsupported by proved to be late, often 20 to 25 percent after a peak or trough in the aver-
~e o ther c,a n often lead to a false and misleading co nclusio n. which is well ages has occurred. One rule of thumb that has enabled Dow theorists to
Illustrated In Fig. 3-5 by th e follmYing example from 19W. ... :"an ticipat.e prob able reversals at an earlier date is to observe th e dividend
Th e 1 9~1 ~ in September 1929 and was co nfinned. t '" yield on th e Industrials. When th e yield o n th e Ind ustrial Average has fallen
by both 3ver.ages in late Octobe r, In J une 19M, both a.verages made a n ew to 3 ~rcen t or below, it has histori cally been a re liable indi cator at mark et
low an~ then ra llied and reacted in AUgtlSL Following this colTtt1ion. th e tops. Similarl y, a yield of 6 percent has been a re liable indi cator at market
t~dU5trials surpassed their prnio w peak. Many observers believed that this bottoms. Dow th eorists would not necessarily use these levels as actual buy-
S1gnaled ~ e end of a particu larly sharp bear market and that it was only a ing o r selling points, bu t would probably consider altering the ~rcentage
matter of u me before the Rails would follow miL }., it turned o ut" th e action of th eir eq uity exposure if a significan t no nco nfinnation developed between
of the Ind w tJ'Ws ~ totally misleading; the bear market still had another the Industrial Average and th e Transpo rtati o n Avenge when the yield o n ,.. .-
2 yt:'an to run. . th e Dow reac hed th ese ex tremes. This strategy would help to improve the •
r- •
investme n t return of th e Dow th eory, but would no t always result in a supe- " '.
rior performance. At the 1976 peal. for example, the yield on the Dow never
reached the magi c 3 percent level. and pri ces fell 20 pe rce nt be fore
a mechanical signal was confirmed by both averages: In addition , the g
percent top would have missed the mark by about 5 yean in th e late 19905.
Over the years , man y criticisms have been leveled at th e th eory o n the
bas is that from time to time (as in periods of war) th e Rails have been over-
re gulated or that th e new Transportatio n Average no longe r re flects
investors ' expec tations abo ut the future movement goods. Th e th eo ry has
stood the test of time. however, as Table g..J indicat es. Ind eed , cri ticism is
perfectly healthy. fo r if the theory gained widesprea d acceptance and its sig-
nals were purely mechanistic instead of re quiring experien ced judgme nt.
'" they would be instantly discounted. which would render the Dow theory use-
less for profitabl e investment,
Ol.aru g..l to 3-4 shp)V Dpw th eo ry signals between ) 953 and 2(0 ) , These
sho uld not be taken i1 g~1 since they re present my own interpretation of;
that many co uld legitimately disagre e with. They were also made with th e •
Figurw 3-5 1930 t!X.OJ11pIe. ben efit of hindsighL However, J tried to make th em as co nservative as

.;.. ;)"


n " n n n n ~ n n n n n P
u n n " " H P P fl R

Part I: Trend-Detennining Techniques Dow Theory 49

~~a::s~~ve=lt:~~~:~~'1iA Mil sIgna1 was given in October of 1989 (at Chart 3-4 Dow theory signals, 1997-2001. The 1990 bull signal remained in force until
formations). It may be pushing it a bit because ~~; down from double tops (line August 1998 when both series made new lows follOwing an intermediate rally. The sell signal
than.the first. Even though the Industrials went on to register a =~i~thewas slightly higher for the Industrials(at the dotted vertical line) is a bit controversial because this average actua1ly
confirmed. Had they broken above the mid-1990 tradl g, Transports never made a marginal new high as it had done in 1989. However, if the price action of the
confirmation, but this did not happen Th ensu ng range, this would have constituted a spring/summer of 1998 Is treated as a line formation, the 'signalis valid. The NYSE AID line
also triggered with a couple of lines in'the ~me ~buy ~g~1 (at ~ solid vertical line) was also crossed below its 20D-day moving average around this time. After this, there was no period
shows that the Industrials broke down from a line but er
so the bulltrend remained intact. In 1994 the Tran
it was the Industrialsthat refused to confirm. SPO ro
u: cern. 1990. The arrow at A
bS~~ contrmed by the Transports,
own rom a line, but thistime
in which both series traced out a series of rising intennediate peaks and troughs. The new all-
time high in the Industrialsset in January 2000 would have been a bullsignal, but the Transports
were unable to confirm. True, they did make a new high in the spring of 1999, but there was
no line or intermediate test of the 1998 low. After the start of 2000, the Industrialswere never
able to extend their rising peaks and troughs. The two horizontal lines in 2001 indicate the
points at which both series would confirm for a new bull market.
6000 Industrials 1200
3000 1000 Industrials


2600 A


1000 i

possible. Charts 3-3 and 3-4 feature both averages. The half-si als described
at the en? of Chapter 1 have not been allowed.
There IS no question that interpretation of the th
bi I
eory can, at times be
I !, Summary
er:r su ~ective. AI the more reason to treat the signals as one witn . 'th
weight of the evidence "pie." ess III e • Dow theory is concerned with determining the direction of the primary

III trend of the market, not the ultimate duration or size of the trend. Once
confirmed by both averages, the new trend is assumed to be in existence
until an offsetting confirmation by both averages takes place.
• Major bull and bear markets each have three distinct phases. Both the
III identification of these phases and the appearance of any divergence in
the normal volume/price relationship offer useful indications that a

reversal in- the major trend is about to take place. Such supplementary
evidence is particularly useful when the action of the price averages them-
selves is inconclusive.

•_l!'.:,~ !~", ; .·,.";-~':,".'"!:= :or'" '= ' >' :' F' ' ' ' ' :' -' , , , , , :, ,=~,_ ,~ , , , ~", ~_,. , , , , ~ ",~ i


Typical Parameters for Intermediate Trends 51

Intermediate Cycles Defined

A primary trend typically consists of five intermediate trends, three of

which form part of the prevailing trend, while the remaining two run
counter to that trend. In a bull market, the intermediate countertrends are
represented by price declines; in a bear market, they form rallies that sep-
arate the three intermediate downwaves, as shown in Fig. 4-1.
It is apparent from the previous discussion that there are essentially two
Typical types of intermediate price movements. The first, which goes in the direc-
tion of the primary trend, may be called a primaryintermediate price movement.
iI Parameters for The second is an important price movement that lasts from 4 weeks to 3
months, occasionally longer. It normally retraces between one-third and two-
thirds of the preceding primary intermediate trend. This price movement,

III Intermediate which runs counter to the main trend, is called a secondary movement or reac-
tion. Since a primary intermediate price movement operates in the same
direction as the primary or main market trend, it almost always lasts longer

II Trends than its secondary counterpart. Its price magnitude is normally much
greater as well. ./,.. .v./
These countertrends or reactions against the main trend are notoriously

difficult to forecast in terms of character, magnitude, and duration.
Therefore, they should generally be avoided from a trading point of view,
as they will almost invariably be subject to confusing whipsaws. By their
very nature, they tend to fool the majority and are usually extremely
Some Basic Observations \I--: treacherous. It is possible to design successful mechanized systems based on
~e two previous chapters discussed the main or primary trend (that is, the
pnce moveme?t that corresponds to changes in economic activity over the
course of a .typical 3- to 4-year business cycle). Though it is clearly important

II to have an Idea of the direction and maturity of the primary trend, it is also
hel~ful to ha,:,e some understanding of the typical character and duration of
the Inte~ediate ?"end for the purpose ofimproving success rates in trading,

and also ill assessmg ~hen the primary movement may have run its course.
Asucce~sful analysis ofintermediate trends for any market or stock offers Primor) inte"~ediDte
price movement
the followmg advantages: .

~ • ?hanges .in intermediate trends aid in the identification of turning points

In the pnmary trend.
• In.terme~iate-term trading involves fewer transactions than trading of

(Bear morket) "-
minor pn:e movements and therefore results in lower commission costs.
• ~ntermedIat~-trend reversal points OCcur several times a year and can, -,
if p.roperly mterpreted, enable a relatively high and quick return on

Figure 4-1 Intermediate cycles in a primary trend.
50 !

rI .
~,...,' ,'' ..;, ,' -.," '.""
II,,,,,,. z.'.•'. ,.'.',..•.•....•.•."
'" . . •. •••.. •. .•. .•.' '•..".. •" , .•" .. , , , " .. ""•• "" ...
_ ,""" ~~,
Pert I: Trend-DeterminingTechniques TypicalParameters for Intermediate Trends 53

intermediate price movements, but poor or losing signals usually come from
than its predecessor. In a bear market, the reverse conditions hold true (that
seconda~ market movements that occur against the main trend.
IntermedIate-term trends that move in the same direction as the primary is, declines are longer and greater while rallies are short and sharp, ~u.t .of
trend generally are easier to profit from. Traders who do not have the less magnitude). Accordingly, technicians are al~rted first. to the pO~lblhty
of a reversal in the primary trend when a third intermediate cycle IS n~ar­
patience ~o invest for the longer term will find that a successful analysis of
ing completion. It is also important to note whether the o~eral~ techmc~l
mte~edlat~ movements offers superior results, especially as the day-to-day
structure looks weak (strong in a bear market) as the preViOUS mte:med~­
or rmnor SWings are, to a large degree, random in nature and therefore even
ate low (high) is approached and, finally, to note whether that level is deci-
~ore difficult to ca~italize ?n. This tendency has been most pronounced
sively broken on the downside (upside).
m recent years when mcreasmgly sharp price movements have resulted from
This does not mean that primary movements can never encompass more
emotional knee-jerk reactions to the release of unexpected economic data.
or fewer than three primary intermediate price movements, for often they
A secondary reaction does not have to be a decline in a bull market or a
bec:r ~arket rally. It can also take the form of a sideways movement or con-
do. Expect three as a normal event. but do not be caught off guard if there
are fewer or more.
solidation, under the same idea as Charles Dow's line formation (see the
discussion in Chapter 3).

Causes of Secondary Reactions

Intennediate-Term Cycles
Since the primary trend of stock prices is determined by ~e attitudes of
Int~rmediate movements can either go with or against the main trend, investors to the future flow of profits, which are in tum determmed to a large
which m~ans that.there is an intermediate cycle, just as there is a primary degree "by the course of the business cycle, it would seem illogical at first to
one. An mtermedian- cycle consists of a primary intermediate price move- expect longer-term movements to be interrupted by what often prove to be
ment and a secondary reaction. It extends from the low of one intermedi- very uncomfortable reactions (or in the case of a bear market. very decep-
ate trend to the low of the other, as shown in Fig. 4-2.
tive rallies). . _
In a bull market, the up phase of the cycle should be longer in time and History shows that secondary reactions occur because. o~ techmcal dIS-
greater in m~gnituJe. The low on the secondary reaction should be higher tortions, which arise in the market as a result of overoptmnsm (or excess
;r \_3 pessimism), and also because new factors emerge s~g~esting~a~ business
conditions are not going to be as extreme as was originally anticipated, or
even that they are going to materialize in the opposite direction. For.e:rnm-
ple, after the first intermediate-term rally in a bull market for eqUitieS, a
reaction may develop because investors, who had discounted a strong
recovery, now see some chinks appearing that might even forecast an
actual decline in business conditions. Such fears eventually prove
ungrounded, but are sufficient to cause a countercyclical intermedi~te
I reaction. Another possibility might be fear of rising interest rates, which
I could choke off the recovery. Since prices had discounted a strong.recov-
ery, this change in perception causes investors to pull back and pnce~ to
I I I fall accordingly. At the same time, many investors get carrie~ away durm.g

r- term ~cle
L _ Intermedllle__ -I-. _
term I
eye e
- -
-+I lnte,modi.te-
term cycle
+I I + .1.r--t-
term cycle the rally -phase and leverage themselves up. As prices begin to fall, this
causes their equity to shrink and forces them to liquidate, which adds fur-
Intermodl.te· ther fuel to the price decline.
term cycle
Figure 4-2 Intermediate-termcycles. A bear market rally for stocks generally takes place because of an
improved outlook for business conditions over what was anticipated. A bear
54 Part ), Trend-Detennlnlng Tecfvllqua TypbI Parameten; for Intmnedlale Trends 55

Relationship between PrimaJY

Intermediate Moves and
Subsequent Reactions

mark.e t ral ly for bo nds develops under the opposite set of conditions .
Corrections in commodity and currency markets all have their roo ts in a
In PrnjitJ in tJuStod MtuIrd, Ganley published a series of diagrams using th e
classification of interm ediate trends establishe d by Robert Rhea Gartley 's
co nclusion was that tJuS7NJihr in magnihuUtJuprimary i~ movt-
mmt was, tJu lmgrr tJu n1rau'nImt ten&d to W. and via wr:J4,. H~ no ted ~at
this was just as valid for bu ll marke t reactions as for bear market rallies,
Observations of the period since 1933 for vinuallyall mar kets appear to sup-
Changed but in correct. perce pncn of the underlyin g (primuy) economic
trend. The cataJyst for the rally is the rush by traders and investors to cover
their shon positi ons (for a definition and explanation of short seUing, see
po rt this hypothesis.
Fo r example, the rally off the 1962 stock market low was only ] 8 perc~~t
compared to the mean avenge of 30 percent between 1933 and 1982. Th~
the Glossary). h should be add ed that the appmmt motivating fone (or the
I co rrection need not neca.sarily be directly linked t o the outloo k. for bwi-
n~ o r int erest rates.
re presented part of a double bottom fonnation an d therefore the first pn-
maryintermediate rally. This relatively small adY.lllC~ was followed by a some-
Any of these in flue nces could be the excuse for a counttrcydical inter- what larger 71 pe rcen t retracement, However, th e ensuing rally from late
mediate price movement, It could be lin ked to the antidpatrd resolution 1962 unti l mid-l963 was 32 percent and was followed by a small 25 percent -l
or worsening or a political or military problem. fo r example. E1se'ntiaHy,
me change in anticipated condi tions com bined with the unwindi ng of the
reeacemem c r the gain. In terested readers maywish to be satisfied that what
goes up does not necessarily come down, and vice versa,
techni cal distorti ons of the previous primary in tennediate tre nd and its Th e 1976-1 980 gold buU market was very powerful, but the Inte rmediate
associa ted sh,up price movement is sufficient to confuse me majority. O nly corrections were quite brief On the other hand , th e rallies between 1982 c[
when busin ess conditions au" correctly expected to change from rttovery
to recession (Or vice vena) iJ the primary tre nd of equities likely to
and 1990 were far less strong, bu t were followed by corrections of much
greater magni tud e pro po rtionally.

-" I
In his ~xc~lI~rlt book ProjitJ in tJuStodtMarin. H. M . GartleY poi nted out
th a t in the 40 ~ ending in 1935. two-thirds of all bull IIW"k~t correc:tions

Using Intennedlate Cycles to
in the U.s. stoc k ma rket d e-..elopedin two waves of liquidation separated by Identify PrimaJY Reversals
a minor rally retraced between o ne-third and two-thinls of the first
de clin e. An observation of su ch correctio ns since 1935 also bea rs out the Number of Intermediate Cycles
find ing th at mOSt interm ediate corre ctions consis t o f two. rather than one A primary moveme nt may normally be expec ted to ~ncomp~ two and a
o r three, phas~s of liquidati on. Unfortunately, in termediate co rrections half interm edia te cycles (5« Fig. 4-1) . Unfortunately, na t all pnmary move-
within a bear market ca nno t be so ~asily categorized since some are rnen ts correspond to th e norm; an occas ional primary movement may con-
o ne- move affaira o r consist of a rally out of a small base, while still o the rs sist of one, two, three, or even four intermediate cycles. Furthermore, these
unfold as a very volatile sideways movement. Even tho ugh Gartley 's obser- intermediate cycles may be of very un equallengths o r magnitu de, making
vations are concerned with equities, this fo rm of correction applies to all
fin ancial mark~ b.
their classificatio n and identifi cation possible o nly after the event, Even so,
intermediate cycle analysis can still be used as a basis for id~ ntifying th e I,e
matu rity of th e primary tre nd in most cases.
Whenever prices are w~1I advanc ed in a pri mary inte rmediate trend fol-
!' v > lowing the completion of two in term ediate cycles, tech nicians should be
'Lamben Gann Pl.tbliahing, Pomeroy, Washington, 1981.
~ "talerted to the fact tha t a reversal of th e primary trend itself may be about

L "i

56 Part I: Trend-Determining Techniques Typical Parameters for Intermediate Trends

to take place. Again. if only one intermediate cycle has been completed, the ~P.Ok

chances of prices reaching higher levels (lower levels in a bear market) are ----,
quite high. I
Characteristics of a Final Intermediate Cycle in a Primary Trend I 200
In a~dition to actually counting the number of intermediate cycles, it is also
possible to compare the characteristics of a particular cycle with those of a
primary trend's typical pivotal or reversal cycle. These characteristics are dis-
cussed in the following section.

Reversal from Bull to Bear Market Since volume leads price, the fail-
ure ofvolume to increase above the levels of the previous intermediate cycle
up phase is a bearish sign. Alternatively, if over a period of 3 to 4 weeks vol-
ume expands on the intermediate rally close to the previous peak in vol-
ume but fails to move prices significantly. it represents churning and should
also be treated bearishly. Coincidence of either of these characteristics with
a .downwar~ cros~over of a 4Q-week moving average (see Chapter 9) or a
Bea' """kIt with 2 lfJ.
divergence m an intermediate-term momentum indicator (see Chapter 10) inbo,medilltl c:y~.
would be an additional reason for caution.
There are essentially two broad characteristics that suggest that the down- Agure 4-3 Intermediate trends and volume.
ward phase of an intermediate cycle could be the first downleg of a bear
market. The first is a substantial increase in volume during the price
decline. The second is a cancellation or retracement of 80 percent or more
ous bear market. Another sign of a basic reversal occurs when prices retrace
of the up phase of that same intermediate cycle. The greater the retrace-
at least 80 percent of the previous decline. Again, the.greater ~e propor-
ment, the greater the p~obability that the basic trend has reversed, especially
tion of retracement, the greater the odds of a reversal m the basic trend. If
~ecause a retracement m excess of 100 percent means that any series of ris-
the retracement is greater than 100 percent, the odds clearly indicat~ ~at
mg troughs has been broken. thereby placing the probabilities in favor of
a reversal in the downward trend is likely because the series of declining
a change in the primary trend.
Other si~s would include the observation of a mega-oversold or an peaks will have broken down. . '
Since volume normally expands substantially as the mtermediate down
extreme swmg (see Chapter lOon momentum for a full explanation of these
phase during a bear market reaches a low, a shrinkage in volume during an
intermediate decline could well be a warning that the bear market has run
its course. This is especially true if the price does not reach a new.1ow dU~­
ing this intermediate decline, since the series of declining intennediat~ cycli-
Reversal from Bear to Bull Market The first intermediate up phase
cal lows which is a characteristic of a bear market, may no longer be intact,
of a bull market is usually accompanied by a substantial expansion in vol-
An example of this is shown in Chart 5-10, where the overall peak in
ume that is ~ignificantly greater than those of previous intermediate up
volume was seen in the June 1962 decline rather than the August to October
ph~es (see Fig. 4-3). In other words, the first upleg in a bull market attracts
sell-off.Chart 5-7 shows no perceptible slackening ofvolume at the lows, but
noticeably more volume than any of the intermediate rallies in the previ-
Part I: Trend-Determining Techniques Typical Parameters fOT intermediate Trends 59

~he volume expansion during the January rally, combined with the better- The median average primary intermediate advance since 1897 appears to
109 of the October to November high (a 100 percent retracement), offered be around 20 to 22 percent. The median primary intermediate upmove in
a valuable clue that the bear market was over. The final decline in October the 1933-1982 period does not differ from that of the earlier period clas-
represe?ted the down phase of the third intennediate cycle in the bear mar- sified by Rhea. However, the median duration' appears to have increased
~et, which should have warned of its probable maturity. A final sign might considerably from 13 weeks in the 1897-1933 period to 24 weeks in the
include a mega-overbought condition or an extreme swing. Again, please 1933-1982 period.
refer to Chapter 10 for an explanation of these concepts.

Amplitude and Duration of Primary Intermediate Downmoves

Intermediate Trends in the U.S. Using Rhea's classification, 39 cases of a primary intermediate decline
Stock Market, 1897-1982 developed between 1900 and 1932, as summarized in Table 4-3.
Amplitude and Duration of Primary Intermediate Upmoves My research shows that between 1932 and 1982 there were 35 ~rimary
intermediate declines, with a median of 16 percent (the declme was
Be~een 18~7 and 1933, Robert Rhea, the author of Dota Theory, classified measured as a percentage from the high). The results are summarized in
?3 lOte~edlate-trend advances within a primary bull market, which ranged Table 4-4.
10 magnItude from 7 to 117 percent, as shown in Table 4-1. The results in the 1932-1982 period did not differ appreciably from those
I have classified 35 intermediate-term moves between 1933 and 1982 and in the 1897-1933 period. Rhea's median average swing was 18 percent, as
the median averaged 22 percent from low to high. The results are shown compared to the more recent 16 percent, whereas the median duration in
in Table 4-2. .
the earlier period was 13 weeks, as compared to 14 weeks in the 1932-1982

Table 4-1 Primary Intermediate Upmoves, 1897-1933

Porportion of intermediate moves Table 4-3 Primary Intermediate Downmoves, 1900-1932
Price magnitude
25 7-14 Proportion of intermediate moves Price magnitude
50 15-28
25 25 3-12
28-117 50 13-27
100 25 28-54
Median 20
Median 18

Table 4-2 Primary Intermediate Upmoves, 1933-1982 Table 4-4 Primary Intermediate Downmoves, 1932-1982
Swings from low to high, percent Duration, weeks i Swings from high Duration (weeks)
Mean average
Mean average




Amplitude and Duration of BuD Narkat Secondary Reactions

Between 1898 and 1933. Rhea classifiC'd 0 cues ofbull market secondaries. i.
In terms of retracemem oCthe previous primary Imermedtaee upm~.lhcy
ranged from 12.4 to ISO percent. with a median of 56 percent. This com-
pared with a range in the 193~1982 period from 25 to 148 percent, wilh a
median of5 1 percent. The duntion oCthe median in the earlier period was ~
5 weeks, as compared to 8 weeks between 1933 and J982. The median per.
centage lou from the previous primary Intermediate peak WOlS 12 percent
(the man average was 13 percent) be~en 1933 and 1982. .. I~
f •

". 1(·.... I
Amplitude and Duration of Bear Market Rallies
Rhea estimattd that the median bar market rally ~traced 52 percent of

111l intermediate KST
the previ o us decline. which is comparable to my own median estimate of
61 percent in the 1932-1 982 period. The two ranges were 30 and 116 per-
cent and 26 and 99 percent, ~pectively. Median dur.ations were 6 weeks
"., If~ ·
r, (" I
.. I
in 1898-1 933 and 7 weeks in 1932-1982. Rallies off the low averaged 12
... ow ... ... ... ... ... .~

and 10 percent for mean and median, respectively, for the 19S3-1982
perioo. . '" ... J
Intermediate Trends in the U.S.
0JaJt 4-2 S&:P Compostte, 199 1- 2001, and an KST.lFrom www.prf"g.rom .)

.'" ..lLL
Stock Market Since 1982 .. ..
v.;.::.. J
.. .. '
Charu 4-1 and 4-2 show the Standard & Poor's (S&P) Composite between .'
the end of 1982 and the opening of the twenty-fint century. The thick ver-
tical lines approximate interm ediate rally peals and the thin ones inter-
mediate troughs. The lower panel co ntains an intermediate oscillator, the .. .,-. ., . .' .
~. e: -::: . .'

Intermediate KST (see Chapter 12 for an explanation), which roughly
renece the tuming points. This period en compassro. the secular buU mar- ". lntermocllate KST
ket that began in 1982 and ended at the turn of the century. The classifI-
cation ofintennediate trends WilS particularly difficu lt compared to previous
,~ If
r-V .. . .. , f",
I'-jf ··I f
periods. I tried (a.s much as possible) to make the intermediate tre nds fit II ' 1/."
the swings in the oscillator. Because oscillaton have a tendency to lead in
buD markets, the actual peaks in the intermediate rallies usually lag those , '(" .'
in the &sT. The two charts show that the classification of these trends is far I 1
from a precise task and co nfinns earlier research between 1897 and 1982
that the range of intermediate trends varies tremendous ly. I n the period
~ ... ow ... ... '" ... ... !•
i ;

L - -------- -- -
62 Part I: Trend-Determining Techniques

covered by Chart 4-1,for instance, the first intermediate rally in the bull mar-
ket lasted well over a year from August 1982 to October 1983. Even if I had
taken theJuly 1983 top as a reference point, the rally would still have lasted
for almost a year. Moreover, the whole year of 1995 was consumed by one
complete intermediate advance.
• The typical primary trend can be divided into two one-half primary inter-
Price Patterns
mediate cycles, each consisting of an upmove and a downmove. In a bull
market, each successive upwave should reach a new cyclical high, and in
a bear market, each successive downwave of the intermediate cycle should
reach a new low. Breaking the pattern of rising lows and falling peaks is
an important, but not unequivocal, warning of a reversal in the primary
trend. For more conclusive proof, technicians should derive a similar con- Basic Concepts
clusion from a consensus of indicators. . d i Ch ters 6 through16 are
The techniques discussed in this chapter an 10 ap d . d by the
• A secondary movement or reaction is that part of an intermediate cycle concerned with analysis of any price trend that has been etenn10e
that runs counter to the main trend, a downward reaction in a bull mar-
ket or a rally in a bear market Secondary intermediate movements typi- interaction of buyers and sellers i?, a free marke~. F' 5-1 and 5-2. Fig-
The concept of price patterns IS demonstrate 10 Igs. d .
cally last from 4 weeks to 3 months and retrace between one-third and
ure 5-1 represents a typical price cycle in which ~ere are ~ree:nrS~~~
two-thirds of the previous primary intermediate price movement. 'deways and down. The sideways trend is essentially a honz?~ o.
Secondary price movements may also take the form of a line or horizon- SI , . k ments Sometlmes
sitional one, which separates the two major mar et move .
tal trading pattern.
• The character of intermediate cycles can be used to help identify primary
trend reversals. II t
• As an approximate rule, the stronger an intermediate rally, the less the
retracernent is likely to be, and vice versa for primary bear markets.

Figure 5·1 Normal reversal.


Figure 5-2 V-type reversal.

i. 63

_.......""""'""'_....,,-,---_._-_ _-_ __ __.._"--

.. .. .
64 Pat1 L Trmd-Detennlning T~ Price PattllllTlS p . :. ;i·
,{ J
a highly e-motional mark~ can change without warning. as in Fig. 5-2. but
I ;: o n AA. pnces
Following the unsuccessful assault . uu BB,
turn down unu ne
this rarel y bappees. Consider a fast-moving train, which takes il long time known as a support ltvtl, is reached.jusr as th e price level atAA reversed th e
to slow down and then go into reverse: the same is normally true of finan- balance in fa:vor of th e sellen, so the suppon level BE alten th e balance
cial markets, again. This time. th e tre nd mOY'eS in an upwnd ~on because at BB
To the market technician, the transitional phase has great rignificana p~~me re lativdy attractive for_b~n ~ho m~ th e boat on the
~~ it marks the tu~ing poilJ.1-~~tl il rising and a falling market. If . way up, while sellers who fed that th e pnce will agam reac~ ~ hold off.
pnq! ~yc...~n advancing , the of the bu~n has outweighed ~~ For a while. th e re is a ~~ff between buyen and sellers ~tllln the con-
the petsiiiilsm of sellers up to this point. and pri ces have risen accordingly. fines of th e area bounded by lin es M and BB. Finally, th e pnc e falls below
During th e tra nsitio n phase, the balance becomes more or less even until BB, and a majo r new (downward) tn' n~}s ~ignaled. .
finally, for one reason or ano the r. it lttitpid in a new direction as the rel- To help explain this co nce pt, th e co n:re'st be~ nJ'~ye n and sellers IS .a
ative weigh t of selling pushes th e trend (of prices) down. At th e termina- like a battle fought by two armies engaged in G'ench warfar e. In Fig. 5-4(a) . I
tion ofa bear market, the reverse prpc~.o.ccurs .
annies A and B are facing off. Line M represents anny A's defense, and BB
Th ese transition phases are almosr?d\ran~iy signaled by clearly defi nab le is army Ifs line of defense. The arrows indicate th e fo rays betwee n th e two
price patterns or formations. The successful completion of suc h patterns or lines. as both armies fight th eir way to th e opposing trench, but are unable
formations alerts the tec hnician to the fact that il reversal in trend has taken
place. This phenomenon is illustrated in Fig. 5-3, which shows th e price
to penetrate the line of defense. In Fig. 54 (6), arm y B finally pushes
through A's trench . Army A is th en forced to retreat and m~e a stan d at .
actio n a t th e end of a long rising trend. As soo n as th e price rises above line th e second line of defense (line A,AJ . In the stock market. line AA repre- -·'it.
BB, it is in th e transitional area. alth ough this is apparent o nly some time aenu selling res istan ce. which. once overcome, signifies a change e ~
after the picture has developed. ance between buren and sellers in favor of th e buyen, 10 that pnces WIll
Once into the area. the price rises to line AA, which is tec hnically term ed advance quickly un til new resistance is me t, The second line of defense. line
a mUUZflU' ana. The word mu-lana is used because at this point the index AtAJo represents resistance to a funher advance.
shows opposition to a further price rise. This term. along with support. is diJ.. On the oth e r hand. anny B migh t qu ite easily break th ro ugh A,AJo, but
cussed at le ngth in Chapter 15. When th e demand/supply relationship th e further it advances without time to co nsolidate its gaim, th e more hkely
comes into balance at AA, the market qui ckly tu rns in favor of th e sellers it is to become..srr; rHtended and th e greater is the probability of its suf-
because pri ces react, Th is temporary reversal may occ ur because buyen fering a seriow-;ettS'ick.. At some point, th erefore, it makes more sense fo r
refuse to pay up for a security, or beca use the higher price attracts more sell- this successful fo rce to wait and consolidate its gains.
ers, or for both of th ese two reasons. The important fact is that th e re la- If pri ces in finan cial markets ex tend too far without ti~e to diges t th eir
tionshi p between the two groups is temporarily reversed at this po int. gains. they.too are m ore likely to face a sharp and_seeml~g1y un expected
reversal. "V· e.

" At

l..:-- Q... .1,,;..lJ"':- ";'" _ ..J r '\-..- ·oJ \ A, ~ ~ ; .. ..

J ~ , , ;,. ..:.:,.. s \;.- \\.; ~ ~' ",.':" , ( ... -0
......' j J. ~ ...\..-o • • "J .,
~~.,,,. ; I
- ••
l..C.8-- \_ i ~ ($
..... ,.... v ~ , \ ;.....:. $ - / '
~ A (a)

Aaura 54 Trench warfare.

(b) <j
' "I

'j\' >~ "

FIgur. ,5.3 Tredlng range redilll!;le.

66 Part I: Trend-OetenninIn Techniques 67

Introducing the Rectangle

The transitional or hori zontal ph ase separating ruing and falling price
trends discussed previously is a pattern known a! a rnwnglt. This formation
corresponds to the line formation developed from th e Dow theory. The rec-
tangle in Fig. 5-3. marking the turni ng point between the bu ll and bear
phases, is termed a mJenal pattern. Reversal patterns at market to ps are
kn own as distribu.tUm areas or patterns (where the security is distributed from Thus, a pattern that shows up on a monthly chart is likely to be far more
strong informed participants to weak uninformed o nes) I and those at mar. significa nt than one on an Intraday chart, and so forth . It is just as impor-
ket bottoms are called accumulation patterns (where th e security passes from tant to build a strong base from which prices can rise as it is to build a large.
weak uninformed participants to strong info rmed o nes) Ieee Fig. 5-5(a» . strong. deep fou ndation upon whic h to construct a skyscraper, .
If the rectangle were completed with a victory for the buyers as the price In the case of financial market p~Y\:.e,!l. H!-..f fou!19ation is an accumulauon
pushed through line AA [see Fig. 5-5(b»), no reversal of the rising trend pattern that represents an area ofitia~'Givt" com6a't between buyers and sell-
would occur. The breakout above AA would therefore have reaffirmed the ers. The term accumulation is used because market bottoms always occ ur
underlying trend. In this case, the corrective phase associated with the for. when the news is bad. Suc h an environment stimulates sales by u ninform ed
mation of the rectangle would temporarily inte rrupt the bu ll market and investors who were not expecting developments to improve. During an accu-
become a consolidation pattern. Such formations are also referred to as con. mulation phase, more sophisticated investors and professionals wou ld be
soiidaJion or continuation patterns. positioning o r acc umulating the asset concerned in anticipation of
I?uring the pe riod of formation , there is no way of knowing in advance improved conditions 6 to 9 months ahead. During this period, it is moving ,.
which way the price will ultimately break; therefore. it should always be from weak. uninformed traders or investors to strong and knowledgeable ,..,
assumed that the prevailing trend is in existence until it is proved to have hands. The longer the period of accumulation. the greater the amount of :.j
been reversed. ,,
a security that moves from weak into strong hands and the larger is the base
fro m which prices can rise.
At market tops, the process is reversed, as th ose who were accumulating
Size and Depth at or near the bottom sell to less sophisticated market participants, who
become more and more attracted as prices rise and the un derlying funda-
The principles of price pattern construction and interpretation can be me ntal conditions improve. Where a substantial amount of distribution
applied to any time frame. from I-minute ban all the way through to develops, this is usually followed by a protracted period of price er~i?n or
mo~thly or even annual charts. However. th e significance of a price for-
mation or pattern is a direct function of its size and depth.
base building. • ± ,, ~ :I


/ .8
, .

Figure 5-6 Long rectangle.


(.) Ib)
flgur. 5-5 la) ReYersaI rectMgIe. (b) ConsoIldatlon rectangle.
68 Part I: Trend-Determining Techniques Price Patterns 69

I The time taken to complete a fonnation is important because of the

amount of an asset changing hands, and also because a movement in price
beyond the boundaries of a pattern means that the balance between buy-I""';f\., I
ers and sellers has altered. When the price action has been in a stalemate" . t
for a long time and investors have become used to buying at one price and \.
selling at the other, a move beyond either limit represents a fundamental I
change, which has great psychological significance. I
The dep,tP, 5}L a f9rmation also determi~~~}ts significance. Consider the i
trench waf'inal~ once more. If the ~posing trenches a~7very close
together, say, within 100 yards, this means that the victorious ~Sault, when
I it comes, will be less significant than if they are separated by several miles;
in such a case, the battles will have been much more intense and the vic-
Figure 5-7 Arithmetic scaling (same vertical distance = same point move).

tory. that much greater. The same is true in the financial markets. The
breeching of a wide trading range has far greater psychological significance Chart 5-1 compares an arithmetic scale with a ratio scale for the Standard
than that of a narrow one. & Poor's Composite. The distortions on the arithmetic scale are self-

I Measuring Implications
Ratio Scale
Most of the results obtained with technical analysis procedures do not indi-
cate the eventual duration of a trend. Price patterns are the exception, since Prices plotted on a ratio scale show identical distances for identical per-
their construction offers some limited forecasting possibilities. Before we centage moves. In Fig. 5-8, let's say that the vertical distance between.I and
examine this aspect, it is first necessary to draw the distinction between arith- 2 (a 2:1 ratio) is 1/2 in. Similarly, the 2:1 distance between 4 and 2 IS also
metic and logarithmic or ratio scaling, because the choice determines the represented on the chart as 1/2 in. A specific vertical distance on the chart
significance of the measuring implications. indicates the same percentage change in the price being measured, what-
ever the level. For example, if the scale in Fig. 5-8 was extended, 1/2 in. would
alwaysrepresent a doubling, from 1 to 2, 16 to 32, 50 to 100, and so on,just
Arithmetic Scale as 1/2 in. would indicate a rise of 50 percent and 1 in. would show a qua-
druplingofprices. Almost all computer software gives the user the option
Arithmetic charts consist of an arithmetic scale on the vertical or y axis, with of choosing between arithmetic and logarithmic scales. .
time shown on the horizontal or x axis, as illustrated in Fig. 5-7. All units of It is important to remember that market prices are a function of psy-
measure are plotted using the same vertical distance, so that the difference chological attitudes toward fundamental events. Since these attitudes have
in space between 2 and 4 is the same as that between 20 and 22. Arithmetic a tendency to move proportionately, it makes sense to plot them on a scale
scaling is not a good choice for long-term price movements, since a rise from that reflects proportionate moves equally. . .
2 to 4 represents a doubling of the price, whereas a rise from 20 to 22 rep- Figure 5-9 shows a rectangle that has formed and com~leted a (~Istnb~­
resents only a 10 percent increase. tion) top. The measuring implication of this formation IS the ver~cal dIS-
In the U.S. stock market, a price move in excess of 50 points for the Dow tance between its outer boundaries, that is, the distance between hnes AA
is normal. A 50 point move in 1932, though, when the average was traded and BB projected downward from line BB. If AA represents 100 and BB 50,
well below 100, would have represented a huge move. For this reason, long- then the downside objective will be 50 percent, using a ratio ~cal~. ~en
term movements should be plotted on a ratio or logarithmic scale. The projected downward from line BB, 50 percent gives a measunng implica-
choice of scale does not materially affect daily charts, in which price move- tion of 25. Although this measuring formula offers a rough guide, it is us~­
ments are relatively small in a proportionate sense. For periods over 1 year, ally a minimum expectation, and prices often go much further than their
in which fluctuations are much larger, I always prefer to use a ratio scale.
70 Part.: Trend-Determining Techniques Price Patterns 71

Cha~ 5-1 S&P Composite arIthlMtlc VI. I'ItIo ale. This chart contains the S&P plotted
on ~nth~c and ratio scales. Note how the.sthmetIc ICBIe totally obliterates any price fluc-
tuations in the early part of the century and ....,.. the price movements in the last 20
years or so. The 1929-1932 perlod, the InOII .,.... on record does not even appear
blip on this scale. This is not true of the IowIr panel, which fea~ a logarithmic or ratio scale.
(From Intermarket Review.) •


1000 S&P Composite (arithmetic scale)

Figure 5-9 Measuring implication from a rectangle.

o-t------ --------.r' Fig. 5-10(b) uses the same projection based on a logarithmic scale. In this
case, a more realistic objective of 50 is obtained.
If a rectangle appears as a bottom reversal pattern or as a consolidation
S&P Composite (ratio scale) pattern, the measuring rules remain consistent with the example given for
the distribution formation. This is shown in Figs. 5-11 (a)-( c).
If the minimal objective proves to be the ultimate extension of the new
trend, a substantial amount of accumulation or distribution, whichever is
appropriate, will usually occur before prices can move in their original direc-
tion. Thus, if a 2-year rectangle is completed and the downward price objec-
tive is reached, even though further price erosion does not take place, it is
still usually, though certainly not always, necessary for a base (accumulation)
to be formed of approximately the same size as the previous distribution (in

this case 2 years) before a valid uptrend can take place.'
In many cases, the price will move beyond the objective. In really strong
, moves, it will move in multiples of it, where the various multiples or the
objective itself becomes an important support and resistance area.

Confirmation of a Valid Breakout

Figure 5-8 Ratio scaling (same vertical distance = same proportionate move). Price
So far, it has been assumed that any move, however small, out of the price
pattern constitutes a valid signal of a trend reversal (or resumption, if the
im~lied objective. In a very high proportion of cases, the objective level pattern is one of consolidation). Quite often, misleading moves known as
denved from the ~easurin~ formula becomes an are~ of support or resis- whipsaws occur, so it is helpful to establish certain criteria to minimize the
tance ",:hen the pnce trend IS temporarily halted..c..Vj;'\
. ~e lIDporta~ce of using logarithmic scales wh~n~ possible is shown
III Fig. 5-10. In Fig. 5-10(a), the price has traced out and broken down from 'Note: It is very Important to remember that price objectives represent the minimumultimate
target and are not normally achieved in one move. Usually, a series of rallies and reactions in
a.rectangle: Pr?jecting the vertical distance between 200 and 100 downward an upside breakout is required. or reaction or retracements in a downside breakout, before
gives an obJecuve of 0, clearly a very unlikely possibility. On the other hand, the objective is reached.
72 Part I: Trend-Detennlnlng Techniques Prloe P!lttems 73 ',)'.;
.". ••

... possibility ofmisinterpretation. Co nventionalwisdo m h olds that you sh ould

wait for a ~ percent penetration of the boundaries before concluding that
the breakout is valid. This filters out a substan tial number of misleading
moves. even th ough the resulting signals are tess timely.
This approach was developed in the first part of the twentieth century
when holding periods for market participants were mu ch longer. Today, with ,
the po pularity of intraday charts, ~ percent could rep~nt the complete
move and then some. I have no basic objection to the 3 percent rule for
longer-term price m ovements in which th e fluctuations are much greater. ,.I
However, the best approach is a co mmo nse nse o ne based on experience and
judgment in each particular case. It would be very conven ient to be able to
say that anything over a specific percentage am ount rep~nts a valid break-
out. but unfortunately a lot depends on the time frame being considered
and the YOlatility of the specific 5eCUrity.
For example, electric utilities arc very stable in their price action com-
pared to mining stoc ks where the volatility is far gre ater. Applying the same
percentage breakout rul e to bo th o bviowly doesn 't make sense . What co n-
stitu tes a decisive breakout, where th e odds of a whipsaw are consid e rably
!lO _
reduced is then very much a Mauer of personal j udgme nt based on ex~ 1
n ence, trial, and error. Thisj udgment ehculd take in to considera tion such
factors as the type of trend be ing mo nito red, the volatility of the security,
volume , and momentum characte ristics.
Anomer factor that can hel p carlyon in deciding if a breakout is valid
lies in the fact th at a good breakout should hold for several periods. For ,,
example, you may observe a decisive upside breako ut from a rectangle on
a daily chan, but if it canno t hold for more than 1 day above th e breakout
1~1. th e signal is highly suspect, Often the tech nical position is worse after
such breakouts. This is because breakouts that cannot hold indica te exhaus-
tion. an d ~llim....m~ffi.QfIcnioUowj:_dJ?y.JU:ong pti~JQds.irHbe
~~~ di~~o~ ~ind i catedJ)'. the (false).breakouL An example of
a 5C brea kout is Featured in Chart 5-2 fo r June 2001 Gold.
On cntering any trade or investment based o n a pri ce pattern breakout,
tbl 101 it is important to decide ahead of time what type of pri ce action would cause
you to co nclude that the breakout was a whipsaw. An example might be a
break below a previous minor low, a decline below a predete rmi ned level
from the breakout poi nt, A stop sh ould th en be placed below such a pivotal
point. In way, you will have calcu lated the loss you are willing to under-
take and th e point wher e th e original prem ise for th e trade; that is. th e
breakout is no lon ger o perative. Failure to ma ke such a decision ahead of
time will mcan th at your decision to scll is more likely to be based on emo-

I tion and knee-jerk reactions to news events than on a logical prcset pl an . A

- -"". -
__ _ _ ..1:
.. ~
74 Part I: Trend-Detmnlnlng Technlques Price Patterns 75

0M0rt S-Z G:*lli.nns an:! an ahM.tlon bruko.lI:. This c:hal1 of .Aaoe (2001) Gold shows
a nice dedsNe bn!akouI. from. ~ heed~lH&Sl pettem, ~ . this
proued 10 be en exhaustion rT"Il:JW *'aI tt. bnW.out WII$ UfIllb&e to hold aoo..e tM neckbne. v
In ad\ sltulltlons, II: Is UilUBIIy bell: to ~te ~ tlls re.aBze;I tNt II. IBlse 'breako.Jt has dlrveJ..
oped. Then! an! three pIaoes ID taM IdIon when the price $Itps ~ below the breakc:U:
point, tI faDs belowthe line .Ic**l!I tiM: heed with the I'tgN: ~. Of tt. pnc:. slips bebN
the righl shedder beIf . The ~ r:l 0RICh of the- ben;hmarks beccmes plOg,uslue!y mort
reliable as Il signal. but II: en:ounten a ~ potentiIIllou. (From Inlermo t lret Rcl>leul.)

... 2 •

As th e pattern nears completion, disinterest prevails and volum e a1rn ~t
dries up. N if by magic. activity picb up noti ceably when the seCU nty

moves above its level of resistance (bounded by the upper line in the rec-
tangle). It is sometim es possible to draw a trend lin th~ lower .v?l-
ume peaks , as shown in Fig. 5-12. It is thi s upward su rge In trading acuVity
that confirms the validity of the breakout. A similar move on low volume
possible po int of no return for lhe false breakout in Clan 5-2 would be at would be suspect and would result in a failure of volume to move with the
the point, where the dashed up trendline joining the lows was violated.
Following the sharp price rise from the rectangle. enthusiasm dies down
as prices correct in a sideways movem ent and volume contracts (rectan gle
Volume BB). Th is is a perfectly normal relationship. since volum e is correcting
Volume wually goes with the tre nd; that is, volume advances with a rising (declining) with price. Eventually, volume and price expand togethe r; and
trend of prices and falls with a declining one. This is a normal relation ship, the primary upward trend is on ce again confirmed . Finally, the buyers
and anything that diverges fro m this characteristic should be considered a become exhausted, and the price forms yet another rectangle (rectangle
warning sign th at th e p revailing price trend may be in th e proceu of revers- cq charac terized, as be fore, by falling volume, but th is time destined to
ing. Figure 5-12 shows a typical volume/price reladorahip. become a reversal pattern .
Volume, the number of un iu ofan asset (such as shares or contracts) that It is worth notin g that while th e volume from the breakou t in rectangle
cha nges hands during a specific pe riod. is shown by the vertical lin es at the BBis high, it is relatively lower than that which accompanied the move fro m
bo ttom of' Fig. 5-12. Volume expands m arginally as th e price approach es its rectangle AA In relation to the overall cycle, this is a be arish factor.
low at the begi nning or rectangle AA. but as the accum ulation pattern is In this case, volu me makes its peak j ust before en te ring rectangle BB,
formed. activity recedes noticeably. While the pea k in prices is not reached until rectangle CC
j" ,' t./ '


Volume co ntracts throughout the formation of rtttangle CCand expands

as prices break o ut o n th e downsi de. This expanded level of activity associ-
ated with me violation of support at th e lower boundary of the rectangle
emphasizes the bearish nature of the breakout, although expanding volume . ~ ...
is not a prerequisite for a va.lid lignOl1 with downside breakouts , as it is for - :.1
an upside move. Following the downside breakout, more often than not,
pri ces will reverse and put on a small recovery or retracemem rally. This I> ~
advance is invariably accompanied by declini ng volume, which itsdf rei»-
forces the bearish indi cations. It is halted a l m e lower end of th e rec tangle.
wh ich now becomes an am of resistan ce. Chart 5-~ .mows an example of

the DowJones Rail Average at the 1946 bull market high.

, ,,
If , ,,
It is now time to expand our discussion LO some of the other price pat-
forma tion . 7
terns. stMting with one of'the m ost notorious: the (H &5)

H&S Patterns ~
H&S as Revttsal Patterns
At Tops The H&:S is probably the most reliable of all chan patterns. It

occurs at both market tops and market bottoms. Figure ~13 shows a typi- I
cal H8cSdislribution pattern. (See also Chart M .) . I
This typical H&S distribution pattern consists ora final rally (the head )
~parating two smaller, alth ough not necessarily identical, rallies (th e shout-
dtt$).lf th e two shoulders WI:'~ trends cr Intermedtaee duration, th e first . ~ I
I Ii .r
sho ulder would be th e penultimate advance in the bull market,. a nd the sec- I I I'. I • •
ond the first bear market rally. Th e head would , of course, represent the "a ~-
I z I
final intermediate rally in the bull market. Volume chara cteristic! an: ofcrit- ~ I
ical importance in ~ing the validity of these patterns. Activity is nonnalJy
t ~ I

heavies t dUring the rormation or the IefllhouJder and also tends to be quite •
u heavy as prices approach the peak. The real tipoff that an H&S pattern is
developing comes with the fonnation or the right shoulder; whic h is invari-
ably accompanied by distinc tly lower volume. Quite often, the level or vol- ~

ume contracts as the peak of the right shoulder is reached. The line joining
th e bo uonu of th e two shoulde rs is called th e n«AIine..
I ,. . ...
".. a 1"1 a

PrOS Wl~O<lS j.O

1"1 a
~ 0 N

. -

o..rt 5-4 Th<! New York nmes a~ . 1928 . ThIs chart oI Th<! ~ York n ma alief-
" age of 50 nlllroad and Industria1 llocb ~ the fonnallon of an UPWll.rd-sIoplng H&S dur-
Ing Much. April, and May 1928. The mlnlmum dcM.mide objectiYe of about 182 \lIaS
achlewd fairly ~, b.Jt a 3-month period of basoI butIdtng commensurate with ~ H&S
pattern lUllS still nec:essI!IT)I before the effect of the di5l:riU:Ion was ~ M aM pnca wen!
able to resune their primary adveoce. Note the he.wy \OObne on the lEft ~ ard head .
end the meUrooely low vobne on the right shoulder. AI!o. activity dedined ~ dumg
the ionnatlon 01 the trlangIe . tu began to expand lbing the breako.Jt in September.


Agure 5--13 0a.Wc H&S lop.

• ~
If you loo k. carefully at fig . 5-13. you will ap preciate th at th e violation o f
th e neckline also re presents a sign al that me previous series of rising peaks
and troughs has now given way to At least one declining peak an d trough.
The riK!!L~outd.5r_~n:sents the fint lower ~...aD.dJhe...bouoao[ the
l!!.ove iollowinK~e hreak.d qwn ~r~ !}lJ a lo~L~gl) . " 0
Th e measuring formula for this price form atio n is th e distance between
the head and the ne ckline projected downward from the neckli ne. as
shown in Fig. 5-13. I,!..follows th at th e dee per the pa tte rn , th e greater itsbear-
ish significa nce o nce it has been co mpleted. Sometimes an H&S co mple-
ti~n will be. followed by a fairly extensive downtrend; at o the rs, th e negadse
effect of the pa ttern will be qui ckly cancelled by the completion of a base. ".
Oft en, traders will observe th e formation of an H&S l Op and anticipate
a b re akdown . This is an incorrect tactic based o n this evidence alone
because it is not known un til later wheth er the p~l ing trend will co ntin ue '" ,,
or if a reversal signal will be given with a decisive break below th e nec kline .
Over the yean . I h ave see n many analystJ, who should know beuer, forecast
a bearish trend based o n an incomple te H&S lop. Remember; in tiomiuJl
arI4iJm, IN fnroaiIin.g tTrnd is auu-J 10 be in foru unli! IN ~ of
nJid nia jmTrJI!J otherwisL An incomple te H&S is not evidence. just a possible '"
sce nario.
H &S patterns can be formed in 10 to 15 m inutes or tale decades to e
develop. Generallyspeaking, th e lo nge r the period. the greater the amount
of disuibution that has taken pl ace. and therefore the longer the ensuing

bear trend is likely to be. Th e larger H&S formations are often ~I'}' com-
plex and comprise several smaller ones, as shown in Fig. 5-)4.
Th e H &S patterns illustrated in Figs. 5-13 and 5-14 have a horizontal neck-
line, but there are many o ther varieties [see Fig. 5-) 5( 0)- ( c» ). all of which
~ ~ V ~ » ~ n n
possess the same bearish impl icationJ as the horizontal variety once they
have been co mpleted.
, . ..
25 •
..". MoJ
H 1 ft
B 25

-"4 '.lIt.

._c_ ' ;"

80 Part I: TrendpetermwngTechnlques 81

0IUt 5-S DJlA, 1898. ThI$~lnverse H&S JlIlttemdeYeJoped In thesprtng

ol l898. Notli! thet the April raDy devdope:I on very klw wIume. The ~t rli!actIon suc-
cessfully tl!Sled the March klw and the III'I5Uing In.akout rally wm IICalITIpBnled by a b.dlish
expansion ol vobnli!. By August. the Dow Jaws IniustrIllJ Average (DJlA) had rli!aChed 60.97
and by April 1899 It rose to 77.28.

"' - ' - ' - - ' - ' - --'- - ' - ,---:71

F1gun 5- 14 Complex H&5.
s ..
" .. I

(e) Feb. . ~.

." ...." '"

At Bottoms Figure 5-16 indica~ ho w an H&:S pattern would look at a

market bottom: this is o.sually called an iFlvmt H&s, a JrWnt 11&5, or an "
H&s 6ou<>m.
Normally. volume is relatively high at the bottom ot' the left shoulder and
during the formation of the head. The maj or factor to watch for is activity
on the righ t shoulder. which should contract during the decl ine to the
trough a nd expand substantially on the breakout (see Chart 5-5). Like the
H&S distribution patterns, th e inverse (accum ulation) H&S can have a num-
ber of variatio ns in tre nd line 5Io~. number of shoul ders, and so on. ..

Usual ly. the more com plex the formation, th e gre~ter its significance. Th is
goes back to the idea that pri ce formations re present battles be tween buy- H&S Format ions liS Con tinuation Patterns
en and sellers: the more battl es that go on, the greate r the complexity and
H&s and reverse H&S formations occaslcnallyshow up on th e charts as con-
me more sign ifican t the new trend is once' the battle has be en resolved.
tinuation patterns. Measuri ng impli catio ns and volume charac teristics are
Some of th ese reverse H&S variations are sho wn in Figs. 5-17 (a)-(c) .
!he same as for th e reversa l type. Th e o nly difference is that these pattern s
H&S patterns are extreme ly reliable formations, and th eir successful co m-
deVf:lop during a trend ra ther than at th e end. Chari 5-6 shows a n example
pletion usually gives an exce llent indicatio n of a trend reversal.
of an H&S consolidatio n d uring a decline .

82 Part I: Trend-Oetermlning T~ 83

" ."
Ag.ue 5-17 fWwrw H&S Ioelialion$. "'"
H&S Failures
Some times th e pri ce ac tion ex hi bits all the characteristics of an H&S dis- ~:
tribution pa ttern, but it eithe r refuses to penetrate the necklin e OT pe ne-
trates it temponuily and th en starts to rally, This represents an H&S failu re
a nd is ~ually followed by a n explosive rally. !Lis proJ!aE.!rE~~~_mi,sp.~£eE """ ('

p!:ss~~_rt!. On ce the real fundamentals ar e perceived, not only do new buy-

e ra rus h in, but also traders h olding short positions are fo rced to cover. Since " ,
I, I
, •
fear is a stronger motivator than greed. these bean bi d up th e pri ce very
," bw

Chan 5-7, wh ich shows th e d aily close for th e DjlA in 1975. illustrates th is
p he nome no n very clearly. as th e failure of an H&S pa ttern results in a fairly
worth while rally. Nevertheless, an H&S th at don; not work indicates tha t
while there is still some life Id"t in the situati on, the end may not be far off.
In Otart 5-6. the nlly ended rather abruptly in July.
Unfortunately. the pattern i~lf~ no indication that it is goin g to fail
Sometimes such evidence can be gleaned from oth er ttehnical factors. For
exam ple, if a countertrend signal looks as if it may be taking place, th is is
just as lilcly to result in a failure . For instance, the H&cS failure in 01art 5-7
was a sell signa l in a bull market, Clan 5-8 shows another example of an
H&S failure for Alberto Cul~r. Double Tops and Bottom.
Failures wed to be fairly rare, but now seem to be mo re common, which A double top co nsists of two peaks separated by a reaction o.r vall~. in prices.
indicates the necessity of waitin g for a decisive breakout on th e downside. Its main characteristic is th at the second top is form ed with disnnctly leu
If any action is contempla ted, it shoul d be take n when th e price breaks volume than th e fin t (see Chan 5-9) . It is no nnal for both peajs to fo nn at
above- th e right shoulde r on beavyvclume (see Fig. 5-18) . Usually. such sig- the same pri ce level, bu t it is also possible for the second peak. t~ slightly
nals offer subs tantial profits in a very sho rt period of time and arc ~U worth exceed th e first o r to to p out j ust a little below it. Rem e mber, this IS not an
acting on. In verse H &s patterns can also fail. Again. the failu re is usually exact scie nce . but a commo nsense interpretatio n of a battl e betwee n buy-
~oll~wed by an extreme ly sharp sell-off, as participa nu who bought in anti c- ers and sellers. ,
rpa uon of an upward breakout are flush ed o ut when the new bearish fun- Minimum downside m easuring implications for double tops. as shown In
dam en tals beco me more widely known . Fig. 5-19. ar e similar to H&S patterns.

~ •
84 Part I: Trend-Determining Tecm lques PrIce Patterns 85

Otart 5- 7 DJ1A 1974--1975 double bottom. Notice the lower volume established Ol"l the sec-
ond low in December 1974. Also, in the spring of 1975 , we see an exa mple of 1Iiailed H&S
Chart 5-8 Alberto Culver and 1I failed H&S top . ThIs cha rt of Albmo Culver shows II des-
top. (From fnlermorket RevIew.) sic H&S faUure. The neckline was violated In mld-1993. A retraeement followed, but the yel-
low flags began to come out when the early 1994 dec6ne failed to ma ke II new low. If the H&$
'" was going to work, this shouJcI have happened. Later, the price rellied back flbove the neck-
line. AI thai polot, the w.biUtyof the pattern was very m uch open to question. The real lip-
off carTlll the prlce ralliedeboce the do\l.m trendline joining the heed and the right shoulder
and finally when It rallied ebcce the right shoulder at arT'OW X. (From Inte rma,* et Relllew.)
t .. o

1~ o

." H
. h

'" (h) I, •

. r ' see 19m ".
... "" "

,111 11111 11111111111 11111 II

Nedlfine violated Figure 5· 19 DoOOk! top .

;(,j A double bottom is shown in Fig. ~20. This type of pattern is typicallyaccom-
panied by high volume on th e first bottom, very ligh t volume on the second,
111I 1I1I11I1I11111 11111 11 111111 111111111111 1111111 Il lII lI ll l,d ll llll ll1 an d very heavyvolum e o n the breakout, Usually, the second bottom is fonned
Rgure 5· 18 H&S failure.
above the first, but these fonnations are equally valid whether o r not th e sec-
ond rea ction reaches (or even slighlJ:y exceeds) the level of its predecessor.

! l
.. , •. . . .....-
e.. l" l" , I :• •


<j' , I

» -=j,
~ ~, I
-( ,, ,,
r Figure 5-20 Double bottom.

\ ,, I
, r
• Doub le patterns ma y extend to form trip le tops or bottoms, or sometimes
even quadruple or other complex formations. Some variations are shown
( I in Figs. 5-21(a)- (c) .
( , Th e measuring implications of all these palterns are derived by calcula t-
ing the distance betwee n the peak (trough) and lower (upper ) end of the
pattern and proj ectin g this distance from the neckline. O1art5-7, shown ear-

• lier, and Chan 5-10 show two classic double bottoms in the DJIA in 1974
and 1962, respe ctively.
..-; I
• Broadening Formations
,,, , Right Angled
Broadening formations occur when a series of th ree or more pri ce fluctu -
,, I
ations widen out in size so that peaks and troughs can be connected
with two diverging trendlines. The easiest types of broadening formati on s

, to detect are those with a "Ilanened" bottom or top. as shown in Figs.
5-22 (12) and ( b).

The pattern in Fig. 5-22(a ) is some times referred to as a righl-angUd broad-
mingjrmntJtitm. Since the whole concept of widening pri ce swings suggests
,I I
highly emotional activity, volume patte rns are difficult to characterize ,
although at market tops, volume is usually heavyduring the nlly phases . The 1,
I patterns at both bottoms and tops are similar lOthe H&S variety, except that

"\ 6, • the head in the broadening formation is always the last to be formed. A bear
~gnal comes with a decisive.J'~ide breakout. Volume can be heavy or
I light, but additional bearish-emphasis arises if activity exp ands at this point, "

Since a broad ening formatio n with a flatte ned top is an accumulation pat.
I ! ; ~ : : e .. . .... . ,
tern, volume expansion on the breakout is an important requ ire ment, as
shown in Fig. 5-22(b) . Examples of broadening formati ons are shown in
O1ans 5-11 and 5-12. Th ese two types of broadening formations can also
s. I !

~~ -=h "
f:: 11 "'e a r') :!l
."... Cil3i:::: ::r :!l
'_+ c;: ~ U'
'-' ..- <> ~ en "'
_ .;J>
c: CO
l: 00 tIl lil ~
2 ~ 't:J ::J 2:"~ til
-< _ il", __ o
_ _ rn "'S-& N
!:L::;.g o
~ Ql!a. ::J c:...
~ . ~ :i>
a ;! e
{3 ~ a. ..... '"a. '0
ro :::. c: \0
!l '" :J . 0"1 '"
~. 0'
lli8.g .~ li!
,.. 3
0"1 e
,,~ . ro ~
0"l? Z
3 5 Q. ~
s ~
-' a.. (t) o 3
a", S- 3
~ a.. CJ •

3 g- - ;:l
.., ::1
::1. S- _.

~"" :€ s
~ "' g :4-
~ --
~ Rj g -g.
l_ \l
SlA... ~~ '" ~ ..
nl QT g ~ ~

~ g0 ola. g. :J
S-3a. i:"'~ .. I ~

'" '0
c ~ '"
_. 0
2. Q:

r" :§; '---I C':l
~ ~ ~. z s
'-- n""::J
s: a :To :J
.c ... '" e

--- --- - - -- - - - -

Ch art 5-11 DJl.f\, 1938, right-a ngled triangle. Th is excellent example of a riqht-anqled triangle occurred at the bottom of the 1937-1938 bear
mark et. Note the substan tial volume that acco mpa nied the upside breakout. Following the breakout, the average traced out a right-angled broad -
ening formation with a nat to p. Usually, breakouts from these co nsolidation patterns are followed by a dram atic rise. In this case, however, the
158 level in Novem ber was destined to become the high for the 1938-1939 bull market. (From Intermark et Rev iew.)

I nd.
16 0




12 0

11 0


Jan . Feb. Mar. Apr. May Ju ne July Aug . Sept. Oct . Nay. Dec .

Part I: Trend-Determining T~ Price Patterns 91

Oaart 5-12 'WW ~, a right-angled broedenIng Ionnatlon. Thischart of WW Gninger

shows a right-angled brwdenlng bottom. Note thIIll II Is 001 ~ ~bk to dnIw the outer
Ix:udaries of the pattern 50 that they conned aD the peaks or lmughs exactty. The most Impor-
1M! thing Is to make sure that the bottoms dlverge arK! the lops lann roughly at the same Iewl.
The ooncept Is one of growing Instability on the downslde that Is unexpectedly reversed to the

upside. (from lntermorkef Review.)


'0 .,
figure 5·23 Consolidation broadening formations .

.' Chart 5-13 T-BIDs. 90 days, various price patterns. (From Inferma, ket Review.)
i H
develop as consolidation patterns, as shown in Figs. 5-23( a) a nd (b). Charts
5-13 and 5-14 show several more examples of right.angled broadening
• •
formatting. f
Broa dening formations occasionally fail to work. Some possibilities are I
shown in Figs. 5-24(a) and (b). Unfortunately. there does not appear to be
a reliable poin t beyond which it is safe to say that the pattern has failed to
operate. Th e best de fense in such cases is to extend the di~rging trend.
lines, that is, the dashe d lines in the figures, and await a decisive penetra-

tion by the price as confirmation.
When completed, righi-angled broadening formations of both the rever-
sal an d the continuation type usually result in a particularly dynamic move.
It is almost as if they are aborted H&:S fonnations in which the move is so
powerful that there is not time to ccmplere the right sho ulder.
'Y ,
n · . ~V
Orthodo; L. ..J ~'

The final type of bro adening formation. known as an rmhotk»: broadening top.
is shown in Fig. 5-25. This pattern comprises th ree rallies. with each
' V ,
·';Ir. • '
;"." ' ,~ , I
92 Part I, Trend-Determining Techniques 93 5- 14 IBM-broadening formations. This chart of IBM shows II form of !ignl.... ngled
brO<tdening top . Tbere is no doubt that the chart rel1ccts 1I bull market thai Is becoming more
~ more unstable. However, It Is more bkely thet the price would cxperien(;e a sharp seD-ilff
lITI~lCIllely eher the breakout. In the case 01IBM , howewr, it tllmg llround in II Il<Iding range
for quite II while before ITlO\Iing significantly lower. No te 4150 ano lhcT broadening formation In
1991 to 1992. Thisone is not exactly "righl-.angled"'~ the lower line hasa slight upIWllm
trll}eclory. Also, II eeesc pat tern con tains three com plete price swlngs, a nd this one only has
~. HO\WIIeT, the rallies lire progresslwly higher, thereby injecting the feeling of instability that
IS required lor any broadening forJl'l(lt ion . (From /n fennarkel Re l/leu.'.j


" Figure 5-25 Orthodox broadeningformation.


" emotional nature of both price and volume swings furth er compounds th e
" confusion and in creases the complexi ty of defining these situations.
Obviously, a breakout is difficult to pinpoint under such conditions, but if
" the formation is reasonably symmetrical, a decisive move below the de scend-
ing trendline joining the two bottom s, or even a decisive move below th e
10 second bottom, usuall y serves as a timely warni ng that an even greater
decl ine is in store.
""" . Measuring impli cations are similarly difficult to determine , but normally
... ... ... ... ll81 1WS 1119!l 1990 i£l1
th e volatile chara cter of a broadening top formation implies the co mple-
tion of a su bstantial amount of di stribution. Consequen tly, price declines
of considerable proportions usually follow the successful completion of such

-- ~-
Triang les, perhaps th e most common of all the pri ce patterns discussed in
('1 (bl th is chapter, are unfonunately th e least reliable . Trian gles may be co nsoli-
dati on or reversal form atio ns, and th ey fall into two categories: symme tri-
Figul"4!' 5·2 4 Broade ning formalion.fllilutes . cal and right-angled .

succeed ing peak high er th an irs pred ecessor, and eac h peak separated by Symmetrical Triangles
two bottom s, with the seco nd bottom lowe r than the first, O rt hodox broad- A symmetrical triangle is com posed of a series of two or mor e rallies and
en ing fo rm a tion s are associa ted with marke t peaks rathe r th an marker reactions in which each succeed ing peak is lower than its predecessor, and
tro ug hs . the bottom from ea ch succeeding reaction is hig her than its pr ed ecessor
These pa tte rns arc ex tre mely d ifficul t to detect un til some lime alter the (see Fig, 5-26). A triangle is therefore th e o pposite of a broad ening forma-
final to p has been formed since the re is no clearly definabl e- level of sup- tion since th e trendlines joining peaks and tro ughs COtllle7gl', unlike th e
port, th e violatio n of whic h co uld serve- as a be nchma rk. 11w violent and (orthodox) broad ening formati on in which they dille7gl' .


1_ '/z _ J 1 _'/Z~

AgunI 5-27 CassIe symmetrlcaI tnakolt. •

These patlerns are also known as toils, because the fluctuation in price
and volume dim inishes as the pattern is completed. Finally, bo th pri ce and
(usually) volume react sharply, as if a coi l spring h ad been wo und tighter
and tighter and then snapped free as prices broke out of me triangle.
Generally speaking, triangles seem to won. best when the breakout occurs
somewhere between one-half and th ree-fourths of the distance be tween the

I widest peak and rally and the apex (as in Fig. 5-27). Th e volume rul es used
for other patterns an: also appropriate for triangles. .
~ "IW<ld
RJght·Angl ed Triangl es: 7 (8) (b) ,'.
Right-angled triangles an: really a special fonn of the symmetrical type. in
f9rn; 5-29 Triangle faikn.
that one of the two boundaries is formed at an angle of90 degrees, that is. ,
horizon tal to the vertical axis [see Figs. 5-28( a) and ( b» ). (An example is "

illustrated in Chan 5-11.) The symmetrical mangle does not give an indi-
cation of the direction in which it is ultimately likely to break. but the right-
angle d triangle does. with its implied level of support or resistance and
contracting price fluctuations. An example of a bearish right-angled trian-
gle is shown in Chan 5-13.
One difficulty in interpreting these formations is that many rectangles
begin as right-angled triangles . Co.Jl!«Iuently, a great deal of cauti on should
\zsJsLB / --' B
B -

B:-t I ? D t :

be wed when evaluating these e1JiSive pa tterns. An example is shown in Fig. .;
5-29(0) , whe re a potential do wnward-slop ing right-angled triangle develops
into a rectangle in Fig. 5-29(b).
Traditionally. measuring objectives for triangles are obtained by drawing
a line parallel to the base of the triangle through the peak of the first rally.
B~B B~
-- B
, .

Th is line (BBin Fig. 5-30) rep resents the price objective that pri ces may be
expe cted to reach or exceed. (b) (d)
Figure 5-30(a) and (b) show this concept for a right-angled and S}-metri-
cal triangle at bottom.
96 Part): Trend-Determining Techniques Price Patterns 97

• Price patterns can be formed over any time period. The longer the time
required to form a pattern and the greater the price fluctuations within
it, the more substantial the ensuing price movement is likely to be.
• Measuring formulas can be derived for most types of patterns, but these
are generally minimum objectives. Prices usually extend much further.
• Price objectives are not normally achieved in one move, but are often
reached after a series of rallies and reactions have materialized.

Measuring _ _-+ __r.-_+-


Figure 5-31 Alternative triangle measuring implications.

The reverse procedure at market tops is shown in Figs. 5-30(c) and (d).
The same technique is used to project prices when triangles are of the con-
solidation variety. However, in my own experience, I have not found this
method to be particularly useful. I prefer instead to treat the triangle as any
other pattern by calculating its maximum depth and then projecting this
distance at the breakout. An example of this alternative method is shown
in Fig. 5-31.

• Prices in financial markets move in trends. A reversal is characterized by
a temporary period in which the e~iiliasm of buyers and sellers is
roughly in balance. This transitional process can usually be identified by
clearly definable price patterns that, when completed, offer good and reli-
able indications that a reversal in trend has taken place.
• Until a pattern has been formed and completed, the assumption should
be that the prevailing trend is still operative, that is, that the pattern is
one of consolidation or continuation. This principle is more important
when the trend has been in existence for only a relatively short period,
because the more mature it is, the greater the probability of an impor-
tant reversal.

. \
Smaller Price Patterns 99

Smaller Price (a) (b)

Patterns Figure 6-1 Rags.

. ']
pleted formation. Flags can form in a period as short as 5 days or as long
Most of the price patterns described in Chapter 5 can be observed in both as 3 to 5 weeks . Essentially, they represent a period of controlled profit tak-
reversal and continuation formations. The majority of those discussed in this ing in a rising market.
chapter m~alize during the course of a price trend and.are therefore of The formation of the flag in a downtrend is also accompanied by declin-
the continuation variety. Since many of them are reflections of controlled ing volume. This type of flag represents a formation with an upward bias in
profit taking during an advance and controlled digestion of losses during price, so the volume implication is bearish in nature, that is, rising price with
a decline, these patterns, for the most part, take a much smaller time to form declining volume. When the price breaks down from the flag, the sharp slide
than those described in the previous chapter. They most commonly appear continues. Volume tends to pick up as the price breaks below the flag's lower
in the daily charts. boundary, but it need not be explosive. Only upside b reakouts in bull mar-
kets require this characteristic.
It is important to make sure that the price and volume characteristics
)J.. . r ;
Flags ~ f ; agree . For example, the price may consolidate following a sharp rise in what
ajJpears to be a flag formation, but volume may fail to contract appreciably.
/ /
A flag, as the name implies, looks like a flag on the chart. It represents a In such cases, great care should be taken before coming to a bullish con-
quiet pause accompanied by a trend of declining volume that interrupts a clusion since the price may well react on the downside. A flag that takes
sharp, almost vertical rise or decline. As the flag is completed, prices break more than 4 weeks to develop should also be treated with caution , b ecause
out in the same direction that they were moving in prior to its formation. these formations are, by definition, temporary in terruptions). , '"
of a sharp
Flags for both an up and a down market are shown in Figs. 6-1 (a) and (b). uptrend. A period in excess of 4 weeks represent~ an unatily long tim e for
Essentially, they take the form of a parallelogram in which the rally peaks profit taking and therefore holds a lower probability of being a true flag.
and reaction lows can be connected by two parallel lines. The lines move Flag formations are usually reliable patterns from a forecasting point of
in a countercyclical direction. In the case of a rising market, the flag is usu- view because not only is the direction of ultimate breakout indicated, but
ally formed with a slight downtrend, but in a falling market, it has a slight the ensuing move is usually well worthwhile from a trading point of view.
upward bias . Flags may also be horizontal. Flags seem to form at the halfway point of a move. Once the b reakout has
In a rising market, this type of pattern usually separates two halves of an taken place, a useful method for setting a price objective is to estimate the
almost vertical rise. Volume is normally extremely heavyjust before the point size of the price move in the period immediately before the flag formation
at which the flag formation begins. As it develops, volume gradually dries began and then to project this move in the direction of th e breakout. In
to almost nothing, only to explode as th e price works its way out of th e com- technical jargon, flags in this sense are said to fly at half-mast, that is, halfway
. ./ , '
'\ 01' • ', . ~
'i-ot'> ..... . . . \ .
c .)"\' \ '; -'
100 Part I: Trend-DeterminingTechniques Smaller Price Patterns

up the move. Since flags take a relatively ahort period to develop, they do Chart 6·1 Alcoa pennant. (From Intermarket Review.)
not show up on weekly or monthly charta.

A pennant develops under exactly the same circumstances as a flag and
has similar characteristics. The difference is that this type of consolidation
formation is constructed from two converging trendlines, as shown in Figs.
6-2(a) and (b). In a sense, the flag corresponds to a rectangle, and the pen-
nant to a triangle, because a pennant is in effect a very small triangle. If any-
thing, volume tends to contract even more during the formation of a
pennant than during that of a flag. In every other way, however, pennants
are identical to flags in terms of measuring implication, the time taken to
develop, volume characteristics, and so on.
Chart 6-1 features a pennant for Alcoa in a down market. Note how the
volume shrinks during the formation of the pattern. It then expands on the

A wedge is very similar to a triangle in that two converging lines can be con-
structed from a series of peaks and troughs, as shown in Figs. 6-3(a) and
(b). However, whereas a triangle consists of one rising and one falling line,
or one horizontal line, the converging lines in a wedge both move in the
same direction. A falling wedge represents a temporary interruption of a ris-
ing trend, and a rising wedge is a temporary interruption of a falling trend.


Figure 6·3 Wedges.

(a) (b)
It is normal for volume to contract during the formation of both wedges.
Since wedges can take anywhere from 2 to 8 weeks to complete, they some-
times occur on weekly charts but are too brief to appear on monthly charts.
Rising wedges are fairly common as bear market rallies. Following their
completion, prices usually break very sharply, especially ifvolume picks up
Figure 6-2 Pennants. i noticeably on the downside.

' I.'"

102 Pmt I: T~ Tedri:p..s 103

I Saucers and Rounding Tops Gaps


Figures 6-4 (4) an d ( b) show the Iormauon of a saucer and a rounding top. A gap occurs when the lowest pri ce of a specific tradi ng pe riod is above the
A saucer pa ttern occurs at a market bottom. while a rounding top develops highest level of the previous trading period [see Fig. 6-5(a) ] or when the
at a market peak. A sa ucer is constructed by dr;?owing a circular line under highest pri ce for a specific trading period is below the lowest pri ce of the
the lOW!, which roughly approximates an elongited or saucer-shaped Jette r previous trading period [see Fig. 6-5(b)] . On a dai ly bar chart, the tra ding
U. & the price drifts toward the low point of the sauce r and investors lose period is regarded as a day, whereas on a weekly chan, it is a week, and so ,
, .:
inte rest, d ownward m omentum dissipms.This lack ofin te~st is also ch ar- on. "~
acterized by the volume level, which alm ost dries up a t the time the price Bydefinition, gaps can occur on ly on bar charts on which intraday,,
is reaching its low point, GradualJy, bo th p rice and volume pid up until or monthly pri ces are plotted. A gap is represented by an empty vertical
eventually each explodes into an almos t exponential pattern. space between one trading period and an other. Gaps form in overnight trad-
The pri ce action of the rounded top is exactly opposite to that of the ing as good or bad news is disseminated by the market. Daily g'< are far
sau cer pattern, but the volume characteristics an: the same. & a result, if more common than weekly ones became a gap on a weekly chart can fall
volume is plotted be low the price, it is almos t possible to draw a compl ete only between Friday's price ra nge and Monday's pri ce range; that is, it has
ci rcle. as shown in Fig. 6-4(b) . The tip-off to th e be arish implication of th e a 1 in 5 cha nce relative to a daily cha rt. Month ly gaps are even more rare
ro un d ed top is th e fact that volume shrinks as prices reach th eir h ighes t lev- since such "holes" on the chan can develop only between monthly price
els and then expands as they fall. Both these:characteristics are bearish and ranges. Th e most common place to find g.;tps is on in traday charts at the
are discussed in greater detail in Otapter 22. open. I will ha ve more to Ayon that po int later.
Rounding tops and bottoms are fine examples of a gradual changeover A gap is closed or "filled" when the pri ce comes back and retraces the
in the demand/supply balance that slowlypicb up mome ntum in the direc- . whole nnge of the gap . For daily charts. this process som etimes takes a few
tion opposite to that of the previous trend. Quite clearly, it is difficult to days, and at other times it takes a few wee ks or mo nth s. O n even more rare
obtain breakout poin ts for these patterns since they develo p slowly and do occasions, the process is mwr completed.
not offer any clear support or resistance levels on which to establish a pote n- It is certainly true that almost all gaps are eventually filled, but this is
tial benchmark, Even so, it is worth trying to identify them since thc:y arc Dot t1lwap the case. Because it can take months or even yean to fill a gap,
usually followed by substantial moves. Rounding and saucer fonnations can
also be observed as consolidation as well as reversal pheno mena an d can
take as little as 5 weeks to as long as several years to complete .

(bl {al (bl
104 PIlrt I: Trend-DdennInIng Techniq.Ies

I tra ding strategies should not hie implemented solely on th e assumption th at

the gap will be filled in the im mediate future . In almost all castS, senne lirul
of attempt is matU 10fill the gap. but q uite often a partial fJ.J.ling on a subse-
quent test is sufficient before the price ag;LiD reserts to th e direction of the
prevailing trend. The reason why most g;iPS are closed is that th ey arc emo-
tional affain and refl ect traders who have stron g psychological m o tivatio ns.
We coul d say excess Iear o r greed, depending o n th e d irection of th e tre nd
Decisions to buy or sell Gl 07I'J cost are net objective ones, which means th e
odds of people having seco nd th oughts when thinK' have cooled down are
pretty high. The second th oughts in this case are represented by the clos- (0) (b)
ing of the gap o r at least a good attempt at closing it,
Caps sho uld be treated with respect, but their importance should no t be
overe m phasized. Those thai occur during th e formation of a pri ce pa tte rn.

I known as commongaps o r ana gaps, are usually closed fairly quickly and do
not have mu ch technical significan ce. Another ~ of gap . which has little
significance. is the one that results from a stock going ex-dividend. There
the move. For this reason , co nti n uation gaps are sometimes called measur-
ing gaps [see Figs. 6-6(a) and ( b) J,

are th re e o ther types of gaps that are worthy of conside ration: breakaway,
I runaway. and exhaus tion gaps .
/Y ~S ..J. - ", ,,J "," "j
Exhaustion Gaps
A pri ce move sometimes contains more ~an one runaway gap. This indi-
cates mat a very powerful trend is in monon. but th e prese:nce of a seco nd
( J ,. ..h"""'5 ~ d-f · _ v - 5 O",\ ';\.l,
I Br eakaway Gaps< __
A breakaway gap is created when a price breaks o ut of a pri ce pattern (as
v-. u J \, ((.... , VoL
or third gap sho uld also alen the technicisn to ~e.~t tha t the move is lik~ly
to ru n out of steam soo n. H ence, th ere is' possibility that a second o r third
in Figs. 6-5(a) and ( b). Gene rally speaking. th e presence of th e gap e mpha- runaway gap wiD be the final o ne. An exh1ustion gap. is the~ore associ~ted
sizes the bullishness o r bearishness of the break out, depending o n which with the terminal phase of a rapid advance o r decline and IS the last m a
direction it tak es. Even so , it is still important for an upsi de breakout to be series of runaway gaps [see Fig. 6-6( b)) . _.
accompanied by a relatMly high level oholume. It should not be co ncluded One clue that an exhaustion ga p may be fonnmg IS a level of volume that
that every gap breakou t will be valid because the "sun: th ing" does no t exist is unusually heavy in relation to th e p,tj,ctPlange Qfthat day. In ~uch a case.
in technical analysis. However, a gap assodated w hh a breakout is more likely volume usually works u p to a ci~do well above p revious levels.
to be valid than o ne that is noL Gap breakouts th at occur o n the downside Sometimes th e price will close near th e VJCUum (or gap ) and we Daway ~m
are not required to be accompanied by heavy volume. its extreme read ing. If the next day's tnd ing creates an · isl~d" o n ......hich
the gap day is co mpletely isolated by a vacuum fro m the p~ous day s u:ad-
ing. this is usual ly an exce llent sign th at the ~p day was m fact tM turnmg
Continuation o r Runaway Gaps ( _ ('~ S IJ( '.I ~ ~ (> f ' ) point, This ind icates only temporary eshaw uo n. but sho.uld. be a red nag
Runaway gaps occur during a stra ight-line adva nce o r decl ine when price that signals to highly leveraged traders that they should liquid ate o r cover
quotatio ns are moving rapid ly and em otions are runnin g high . Eith er the y their positions. . . .
are closed ~ry quickly, suc h a.!I within a day or so, o r they tend to remain lf th e gap is th e first o ne during a moe•.n IS h~el~ to be a runaway ra~er
o pe n fo r much longer periods and are not generally closed until th e mar- than a breakaway type. especi ally if the po ce obJec,uve calle d for by a pnce
ket mak es a major or in termediate swing in th e opposite di recti on to the pattern has not yet been achieved. }JI exhaustion . gap sh o uld not be
price movement tha t wa. responsible for th e gap, This type of gap often regarde d as a sign of a majo r reversal . but merel y as a Signa l that, at the very
occurs halfway between II previous br eakout a nd th e ultimate duration of least, some fonn of consolidation sho uld be expected.

106 Part I: Trend-Determining Techniques
Smaller Price Patterns 10 7

The Importance of Gaps as Emotional Points

will see that th e trading range of the previous day (co n tained with in th e box
~ w~ shall le~ later, the places where gaps start or terminate are poten- on th e left) was not exceeded at th e opening price; thus, on a dail y chart,
tial PIVOta!.g:p{?~ on a chart because they represent high emotion. If you there would have be en no gap.
ha~e an arg~ment with a friend and one of you really shouts loudly at one If you are a trader with a 2- to 3-week time horizon using intraday char ts,
pomt, y~u Will both remember that particular moment because it represents you should approach gaps in a different way than if you h ave a 1- or 2-day
a.n ~motIonal extreme. The same principle can be applied to technical analy- time horizon .
SIS since charts are really a reflection of psychological attitudes. This means
The first ca te go ry should try to avoid initiating trades at the time the g-ap
that gaps have the potential to become important support-resistance levels is created. This is because almost all gaps are eventually closed. Sometimes
that have the power to reverse short-term trends. We will ·have more to say this happens within a couple of hours, and at others it can take 2 or 3 weeks .
on this ma tte r in Chapter 15.
Consequently, if you buy on an opening gap on the upside, as in Chart 6-3,
you run the risk that it will soon be closed. The problem is you do n ot know
whether it will be in 2 days or 4 weeks .
Ga p s on Intraday Cha rts
Intraday traders are also advised to step aside when the market opens
There are really two types of opening gaps in intraday charts. The first devel - sharply higher or lower. In the case of stocks, this is caused by an order
op~ as prices open beyond the trading parameters of the previous session, imbalance. That means that the market makers are fo rced to go sh ort so
as 10 Chart 6-2. I'll call these classic gaps since these are the ones that also that they can satisfy the unfilled demand. They naturally try to get the price
appear on th e daily charts. a little higher at the opening so that it will come down a littl e, enabling th em
Th: sec0.nd, more common gap develops only on intraday charts as the to cover all or part of the short position. The process will be reversed in th e
open1Og pnce of a new day gaps well away from the previous sessions' clos- case of a lower opening. The key then is to watch what h appens to th e pri ce
ing bar. I'll call these gaps intrabar gaps because they only fall between two after the opening range . Normally, if prices work th eir way hi gher afte r an
ba:s calculated on an intraday time frame. For example, in Chart 6-3, the upsi de gap and opening trading range, thi s sets th e tone of th e m arket fo r
pnce open ed up higher and created a gap. However, if you look back you at least the next few hours, often longer.

Chart 6 -2 March 199 7 bonds' IS-minutebar. (From Martin Pring's Introduction to Daytrodlng.)
Chart 6 -3 March 1997 bonds' I S-minute bar. (From Martin Pri ng's Introd uction to Day trading.)

114 113 .
jl tttt
Ii/1/11111'1111I1/1j ~I
113 112
III/11 /1
112 Jj
I J II I/i
112 112
(III/Hi 1 !}
111 112 Open ing

ill/ j
/ [.... ! 112 gap

~*V'\~ trA(~J

Classic gap 112
"'I\ I 111 jjl
; 111
::./ « ! 10 'hI '112 'h3
108 Part I: Trend-Det ermlnlng Tedmiques

On the: oilier h a n d , if th e pri ce starts to close the' gap after a fe w ban.

then the tone becomes a negati ve one. In Chan 6-4 feiilluri ng Merri ll
Lynch , th ere is an opening g<lp o n Wednesda y. Afte r a bit of backing an d
fillin g. the pri ce gradually works its way J?~r th roughou t th e day. The sig-
nal th ar th e opening could be an aber¥.ulOn deve lops afte r th e price slips •-• ••• •• •• •• •• •••• •••• •••• •• ••
below th e tr endline. Note how th e tre ndline proves to be res istance for
th e rest of th e session . Thursda y again sees an o pening gotp. bu t I.his time
•• •• •• •• •• •• ••
V the re is very' little in th e way of a trading range since th e price co nti nues
r ' . toclim b. Agai n. the ra lly away from the opening bar sets th e to ne fo r the
;.; i, - rest o f the day. On Friday another goap appears, b u t this tim e the opening
\,. Indin g 1110ge is r esolved o n th e d ownside as th e price b rea ks below the
$86 leve l flagg ed by th e line. Once again, this p roves to be resistance for
the rest of the day.

Island Reve rsa ls

An island reversal is a compact trading ran ge created at the end of a sus-
tained move and isolated from previous price behavior by an exhaustion gap ••

and a breakaway gap. A typical island reversal is shoVl"n in fig. 6-7 and in I
Chan 6-5.
Cha rt 6-4 ~ Lynch's 7· to ~ bar. (TI!/escu o.)

}t91~S t S7~ 1 MEF:P.Iu..llNe H.COi'K;J5 D3:'S La;

88 . - . • . - •. . - . •• - -· ·G8p3. -- • . -- - - ••• -B· · · -_• . . -..•. - · ··;.iJl 88
" -- - - - - - -- -- - - - - - - -- - - -~ ~- - - -- -- - - --J-,- " .,;
" -G~p-,- -------- -----------J- -- -- -'::~~ \,..."1'......;----"
'2 -- T ------------1 ,:.;:1<.-..... 1--- -------------------------- 1
eo ..:.:..:. t·.,~~ · - A· · - -· ·(- · .. --- ..-- -...--...-- ....-... -.. .. -.. eo
19 _ u~ __ - _ . '- ' '-<,-J ' C - -- __ m u mu m - 7'
76 •. . . ~ ~ ~ .Gap
- 2 •
lolu11e- mac 10Je
- - . . . . . . . . . . . . . •. . . .. . . .. 76 ••
103 ---------------- ------- --------------------- ----------fOJ
15 - - .- - - - - - - -- - - - - i------------------------------------ 50 ••
,• W~" ,1,,LdII, ~ . 1 , U,,'..' • •1 I. ~ II • •• • •

• •• •• •• •••• •••• •• •••• •••• •• g
l ue Wed Thu FI1 "' ilion 'I Tue •- • • •• •• ••



7 •
One- and Two-Bar
Price Patterns
The island itself is not usually a pattern denoting a major reversal. ~ ., I
However, islands often appear at the end of an intermediate or even a major
move and fonn part or an overall price pattern such as the top (o r bottom)
ofa head-and-shoulders (H&5) pattern (or an Inverse H&S pattern) . Islands
occasi onally occu r as l-day phenomena. Background
The price patterns we have considered :!O far ~ke some ti~e to. co m plete ,
wua1ly at least 15 bars. Th ey all reflect changes In the relationship betw een
'. '"
Summary buyen an d sellers, which tells us th at th ere has been a reversal in p"yehology.
Historically th e patterns described in this chapte r were called J- and 2-
• Ftags. pennants, and wedges an: short-term price pattern s that usually da y patterns ~r 1· and 2-week. patterns. With ~e advent of intraday charts,
develop halfway along a sharpprice m ovement. Their development is nor- the title "inside days, outside da ys," and so on IS no lon ger a genenc. tt:~ .
mally complete with in !I weeks, and they represent periods of quiet price Therefore, I have chosen to use th e term "bar" with th ese patte rns since It
movement and co ntra cting volume. They are almost always continuation applies to all charu from I -minute ban up to mon thly ones.
• Saucer formations and rounding tops are usually reversal patterns and are
typically followed by substantial price movements. In both formations, vol-
ume contracts toward the center and is at its highest at eithe r extre mity.
• A gap is essentially a vacuum or hol e in a bar chart. Ex-dividend and area
gaps have little significance. Breakaway gaps develop at th e beginning of
a move. runaway gap5 in the middle of a move, and exha ustion gaps at We have alre ady established th at a key factor influe ncin g th e significance
th e end. of a pattern is its size. Since th ese one- and two-bar patterns do not take very
• Island reversa ls are small price patt erns o r co ngestio n ar eas isolated from long to fonn th ey are , by definition, only of short- term 5ignifican~e. For
th e m ain price trend by two gaps . They often signal th e termination of exa mple a l~ay pattern would o nly be expected under nonnal circum-
an inte rm ediate move. stances to affect the pri ce over a 5- to 15-<1ay period. A two-bar pauern ere-
ated from H).minute ban would influence th e tre nd over th e co urse of th e
next ~ minutes to an hour o r so. Even so, th e mo re 1 study th ese p a tteDU,
the more impressed 1 beco me with th eir ca pability to reliably signa l sho rt-
term trend reversals.

• ';'r

112 Part I: Trend-Detennlning Techniques One - and Two-Bar Price Patterns

reversal is
represent a strong signal of exhaustion. An example of a top d (b)
shown in Figs. 7-1 (a) and (b) and a bottom reversal in Figs. 7-2 (p,) an th~
For those not familiar with bar charts, the, top of the bar rep(esents . '
high, and the bottom, the low. The tick . to th e Ie ft 'ill diicates ...

e opemng
price, and that to the right, the closing price.

r:'I '(..
Not all one- and two-bClfpatterns are therefore created equal. What we are
doing is hunting for clues as to the degree of exhaustion being signaled by
a particular reversal phenomenon, I could saythe word "help," for example,
but if I SMut it from the rooftop, you will get the message that I need help
far more clearly. The same principle op'e~tes)~themarketplace. For exam-
ple, ifan outside bar (discussed later) e'6.compassesthe trading range of three
or four bars, it is likely, other things being equal, to be more significant than
if it barely encompasses the trading range of one, and so forth.
There are a couple of ground rules to bear in mind when interpreting
one- and two-bar patterns:

• They generally signal an exhaustion point. In the case of an uptrend, such

formations develop when buyers have temporarily pushed prices up too
far and need a rest. In the case of downtrends, there is little if any surr
ply, because sellers have completed their liquidation. Such patterns are
almost always associated with a reversal in the prevailing trend.
• In order for these formations to be effective, there must be something to (a)

reverse. This means that top reversals should be preceded by a meaningful
rally, and bottom formations should be preceded by a sharp sell-off.

• It is important to interpret these patterns not so much as black and white,
but as shades of gray, because not all patterns are created equal. Some
show all of the characteristics I will be describing later in a very strong
way. Others will reflect just a few characteristics in a mild way. What we
Enro;"p""", tho "'O;"IJ
tho previous "',"00 then some! --;:--:I:: I i-
11 i-i-
might call a five-star pattern, with all the characteristics, is more likely to
result in a strong reversal than, say, a two-star pattern that has mild char-
acteristics. It is therefore necessary to apply a certain degree of common I
sense to their interpretation rather than jumping to an immediate con- I
clusion that the presence of one of these patterns guarantees a quick, prof-
itable price reversal.
Outside Bars !I
Outside bars are those in which the trading range totally encompasses that
of the previous bar. They develop after both down trends and uptrends, and I
Figure 7·1 Outside bars at tops.
. 1 ,

114 Part I: Trend-Determining Techniques One- and TIAIO-Bar Price Patterns 115

• The more bars encompassed, the better the signal.

• The greater the volume accompanying the outside bar relative to previ-
ous bars, the stronger the signal.
• The closer the price closes to the extreme point of the bar away from the
direction of the previous trend, the better. For example, if the previous
trend is down and the price closes very near to the high of the outside

i -r/ i-i- bar, this is more favorable than if it closes near the low and vice versa.

tf .-r. i-4 1---- Outside bar

Some comparative examples of strong and weak outside bars are featured
in Fig. 7-3.
When considering outside bars or any of the other 1- and 2-bar price pat-
terns, it is important to ask yourself the question: What is the price action
of this bar telling me about the underlying psychology? Wide bars, sharp
preceding rallies or reactions, and high volume all suggest a change in the
(a) I previous trend of sentimenL
Chart 7-1 features an example of an outside bar for Merrill Lynch. It rep-
~: '

resents what I would call a five-star signal since it has pretty well all of the
characteristics of a strong reversal.
The price was in a persistent downtrend for the afternoon of the 21st.
Then, a strong bar develops, totally encompassing the trading range of the

previous bar. This is a pretty strong statement because the bar opens close
to its low and closes almost at its high. Note also the very high volume that

i i-i-
accompanies this outside bar.
Several outside bars are apparent in Chart 7-2. Example A is a good one

hi -r
because it is preceded by a relatively strong rally and is reasonably wide.

Encompasses the trading range of
the previous barandthensomel
1.The wider the barrelative to
preceding bars, the better
Figure 7-2 Outside bars at bottoms ;
(b) -r~ _~
i Weak

e j ', -
There are several guidelines for deciding on the potential significance of " ',

::. 1~' >" ;

an outside bar. They are as follows:

• 1?e wider the outside bar relative to the preceding ones, the stronger the
• T~e sharper the rally (reaction) preceding the outside bar, the more sig-
nificant the bar. Figure 7-3 Assessing the significance of outside bars.
116 Part I: Trend -Determining Techniques One- and Two-Bar PrIce Patterns 117

Chart 7-1 Merrill Lynch's ID-mlnute outside bar. (From Pring Research.)
Example B completely fails. The reasons lie in the fact th a t the close is right
..,1, 1
on the low, the bar is not much larger than 'th e previous one, and is not pre-
ceded by much of a decline. Thus, while in a,strict technical sense it is an
outside bar, it is not signaling much in the way of a change in sentiment.
Even though Example C is not really successful and it is preceded by a
decline, it is a reasonable size larger than its predecessor, and closes at its
high. I purposely put in this example to show that even when a substantial
number of the requirements are present, this does not guarantee the suc-
cess of a pattern.
Finally, Chart 7-3 shows two more examples of outside bars for the Dow
Jones Industrial Average (DJIA) in March 2001. The first represents a rever-
2. Good volume
sal from a downtrend to an uptrend. The second marks the top of a very
sharp rally. Note that it is very wide, encompasses the three previous bars,


Chart 7-2 S&P Composite 5-minute outside bars . (From Pring Research .)
Chart 7-3 DJIA 6D-minute outside bars. (From Pr ing Research.)

A 1006
483 940
483 936
482 930
482 926
481 920
481 916
:30 :45 1:00 :15 1:30 :45 1:00 :15 . 1:30 :45 1:00 1:15 I::
, .13:00 14:00 ,.. . , . . 1~:OO ..
One- and Two-Bar Price Patterns
118 Part I: Trend-Determining Techniques

and also violates a good up trendline. Add to this the fact that it also expe-
riences a whipsaw break to the upside and you have all the ingr ed ien ts of a
dramatic change in sentiment.
1+-- Inside bar

Inside Bars
Inside bars are the opposite of outside bars in that they form totally within
the trading range of the preceding bar. An outside bar indicates a strong
reversal in sentiment, but an inside one reflects a balance between buyers
and sellers following a sharp up- or down-move, which is usually later
resolved by a change in trend. During the period preceding the second bar
of this formation, buyers or sellers have everything going their way depend-
ing on the direction of the trend. Then a more even balance sets in during
the formation of the inside bar, which then gives way to a trend in the oppo-
site direction. Examples of inside bars are shown in Figs. 7-4(a) and (b) and
Figs. 7-5(a) and (b). Second bar is
The guidelines for determining the significance of an inside bar are as encompassed
follows : by the first

• The sharper the trend preceding the pattern, the better.

• The wider the first bar and its immediate predecessors in relation to pre-
vious bars, the better. This brings the strong underlying momentum of
the prevailing trend to a sort of climax. - ;, J ' j.\ \
• The smaller the inside bar relative to the outside bar, the more dramatic the
change in the buyer/seller balance and therefore the stronger the signal.
• Volume on the inside bar should be noticeably smaller than that of the
preceding bar since it indicates a more balanced situation.

Chart 7-4 shows two examples of inside bars. The first marks the end of
<:: the sharp Septe~.!>,..e~/Novemberdecline. Note the substantial width of the
i -: ~first barand th!:,'paltr:y range of the actual inside bar. During the formation Figure 7 -4 Inside bars at tops.
of the first bar, volume is very heavy and the price declines sharply.No doubt
here that sentiment is strongly on the bearish side. Then, on the second day.
volume dries up appreciably and the trading range is dramatically reduced,
Chart 7-5 shows some more inside bars, this time on an in~raday ch~t.
thereby indicating a fine balance between buyers and sellers. Although this
Example A is followed by a pretty good rally, though the quabty of the SIg-
marks the bottom of the move , the next short-term trend is essentially a side-
nal is not that great since there is not much ~f a con~t be~een ~e actual
ways one. Quite often we find that with inside bars there is a change as
inside bar and its predecessor. Example B IS an OUtright faIlure ~mce the
opposed to a reversal in trend.
price continues to advance. This shows that when a strong trend IS under-
The second inside pattern develops just under halfway up the rally and
way, price patterns can and do fail. Indeed, their very failure can often be
is followed by a sideways trading range. The move is topped off by a two-bar j,
reversal, which is discussed in the next section. I' a clue to the strength of that trend.

Part l Tnmd-Determ!nlng T~ 121

Cwrt 1 -4 Odonl: Industries' ~200 1 f\M:t-bat nM!nlII and nlde ben. (From PrIng

22 ..tlrlt~·h,
" •n
TWC/-ba r rell9f'
I I v ~r....."...
" I 1-'" .
~-." ...... ,,-"""<, !~f'
I~,,..) -",,-,1"1: / \1

.. Inside blI r t' IIIIt'-l ,I



':t S.Pt, ftIbeI'
• '"
Otart 7·5 1he S&P Composi te's 5-mlnute Inslde blll"l.lFrom PrI ng Research.)

figure 7·5 Inside bars a l bottoms.
Finally. example C is a classic. The two final ban expand coruiderably in
size. Th e actual inside bar is ulatiyety small. Note also how the open and ."
c1<»e develop at almost identical pri ces, thereby confirm ing the idea of a ;30 :46 :00 :11 :30 046 :00 :16 :30
~ry fine balance between supply and dem and. u :oo 16:00

12 3

Two-Bar Reversal
A ~bar reversal is a classic ",..a y in which cham signal ex haustion. These
F ""ems7d~lo7P after a pro longed advan ce or d eclin e . Examp les are shown
In r igs. ....... to -S.
Th ~~rst
bar oflhe fannation develope strongly in th e direction of th e the n
prevailing trend. For a five-star signal in an up tre nd • we neen
-_ • to see th e close

Arst baropens near

llle high and closes
near t he \OW ........
t'/ t
se cond bar open s near
ltle low 800 dosM nea t
!he pteYio LlSopening

of th e bar at. or very close to, its high . At th e openi ng of me next peri od .
bu yers come in expecting more o f the same. This means the price sho uld
opc1l very close to m e high of the previ ous bar. However. the whole point of
the tw&bar reversal is lIlat a change in psychology takn place as th e ba r
closes slightly above o r sligh tly below th e low of the p~ous bar. Hence . the
high expectations of participants at th e opening of the bar are totally d3shed ,-
Figure 7-6 Two-bar teYefMls 1I1 tops. at th e en d of th e peri od, ind icating a reversal in sen timent, To be effective,
th is has to be a climac tic experience. This means th at th e two-bar reversal
should conwn as many of the following elemen ts as possible :
Both bers are of
roughly equal he9hL • It needs lO be preced e d by a penisknl trend- ·tlu=-sharper th e better.
1_-- • Both ban should stand o u t as having exceptionall y wide trading ranges
relative to previ ous bars (see Fig. 7-7 for an uptrend).
• The openings and closings of both bars should be d ose to the extreme
poi nts of the bar.
• An expansion o f volume o n both ban en hances th e co nce pt of a change
in sen time n t,

Chan 7-6 shows a two-bar Te\~na1 fo rJune gold in M.arch of2001. 1n man y
cases, such pattern s will be followed by an immediate advance . In th is case,
~ .• nd llsually the advance was delayed a d ay as an additional piece of evide nce ind icated
Iargerthl n those a trend reversal. T his ca me in th e fo rm of an inside bar that develope d o n
enre pteced mg
the dayafter th e twa-bar r eversal. I have noticed that such "d ouble" patterns
Figure 7 -7 Two-bar ~ 1I1 the rally peek,
are olte n q uite effective in signaling reversals. Note aha that th e second bar
124 Part I: Trend-Determining Techniques One- and Two-Bar Price Patterns 125

Chart 7-6 June 2001 Gold Futures' two-bar reversal. (From Pring Research.) Chart 7-7 US Bancorp 2000 two-bar reversal. (From Pring Research.)

274 26'"
273 24-
269 23'"
268 22"
266 22"
266 21-
264 21-
262 20"
261 20"
259 EM"
257 1000

• The price opens strongly in the direction of the prevailing trend.

in the two-bar reversal encompasses the first bar. In effect, this is an outside • The trading range is very wide relative to the preceding bars.
bar. The two-bar reversal pattern rule does not require an outside bar, but • The price closes near or below the previous close (or near or above the
its very presence reinforces the idea that sentiment has definitely reversed. previous close in a downtrend reversal).
If we are looking at the quality of a signal, I would certainly add points for • Volume, if available, should be climactic on the key reversal bar.
this and the inside bar.
Chart 7-7 shows a two-bar reversal at the climax ofa rally in US Bancorp An example of a key reversal bar is featured in Fig. 7-9. In many cases, a
in the fall of 2000. Note how the volume expands dramatically. Also, volume retracement move (see Fig. 7-10) follows a key reversal bar (see Fig. 7-11),
on the second day is slightly higher than that on the first, thereby provid- especially if the initial reversal in trend is unduly sharp. Normally, the
ing an additional clue that the tide has turned in favor of the sellers. extreme point of the reversal bar is not exceeded. Figure 7-11 shows a key
These one- and two-bar price patterns generally have an effect for a very reversal bar at a bottom.
short period of time. They are not therefore suitable for long-term investors. Chart 7-8 shows a classic example beginning in September 1999, as the
However, for traders who are looking for clear-cut entry and exit points, they short-term rally is climaxed by an explosion of volume and a wide key rever-
can be of immense value. sal bar. Quite often a key reversal is followed by a sharp change in trend and
e f I I
a subsequent retracement. That is exactly what happens in this case as the
/ ~ price rallies in the fourth and fifth sessions following the key reversal. Note
that the termination of this brief collective advance is signaled by an out-
Key Reversal Bars side day. Even though these one- and two-bar patterns only have short-term
A key reversal bar is one that develops after a prolonged rally or reaction. significance as a general rule, quite often they can prove to be the first
Often the trend will be accelerating by the time the price experiences the domino in a major trend reversal. It all depends on the maturity of the trend
key reversal bar. The classic pattern has the following characteristics: in question together with the position of the longer-term indicators.
126 Part I: Trend-DeterminingTechniques One- and Two-Bar Price Patterns 127

Open~ well up on
---.1f ~ Wide trading range

Must be
by a strong advance
r1 "-
Closes <:lose to or below
the previous close

Figure 7 -9 Key reversal bar at tops. FIgUre 7·11 Key reversal bar at bottoms.

Chart 7-8 Barrick Gold, 1999-2000 two-bar reversals. (From Pring Research.)

Figure 7-10 Key reversal bar and retracement moves.

Chart 7-9 features the 1998 bottom for Merrill Lynch. Note that the actual
The second example of a key reversal to the right is also a good one in day of the bottom experiences a classic key bar reversal. The volume also
that volume expands along with the trading range. However, it is not pre- Cooperatesby expanding as the stock bottoms. It is also possible to construct
ceded by much of a rally and would not therefore earn as many stars as the a small down trendline, the violation of which confirms the signal being
first one. given by the reversal bar.
128 Part I: Trend-Determining Techniques One- and Two-Bar Price Patterns 129

Chart 7-9 Merrill Lynch 1998 key reversal bar. (From Pring Research.) Chart 7-10 S&P Composite 5-mlnute plnocchio bar. (From Pring Research.)


• The closing price should be both above the opening and in the top ~alf
The intraday example in Chart 7-10 almost meets all of the qualifications, of the bar in a downtrend and in the lower half and below the openmg
except that the opening is only slightly above the previous close. Even so, in an uptrend.
this bar does indicate exhaustion. One tip-off is the fact that the upper end • The bar is completed with a gap to the left still in place.
of the bar sticks out like a sore thumb above the previous two sessions. In
other words, the price broke out strongly to the upside, but was unable to Examples of exhaustion bars for both a bottom and top are. featured in
hold its gains, and by the close it had given up ground over the previous Figs. 7-12 and 7-13. They differ from the one-bar island reversal m that there
period. is no gap between the exhaustion bar and its successor. Examples of one-
bar island reversals are shown in Fig. 7-14.
What we are looking for here is for an extreme movement in the p~ce
Exhaustion Bars that is preceded by an already strong move. The idea that the bar opens WIth
a huge gap and closes in the opposite direction reflects the conce~t of a
Exhaustion bars develop after a really sharp up- or down-move. They are a reversal in psychology. The large gap and wide trading range also pomt out
form of key reversal, but differ sufficiently enough to warrant their own cat- the kind of frenzied activity associated with a tum.
egory. Chart 7-11 shows an example of the daily price action ofKellwood. It was
The requirements for an exhaustion bar are as follows: followed by a rally, a stark reminder that all these one- and two-bar
formations have short-term significance.
• The price opens with a large gap in the direction of the then-prevailing The exhaustion bar in Chart 7-12 does prove to be the actual bottom.
trend. Note how the termination of the test in mid-October was signaled by an
• The bar is extremely wide relative to previous bars. inside day.
• The opening price develops in the lower half of the bar in a downtrend You will find that gaps almost always develop on the intraday charts at
and in the upper half in an uptrend. the Open due to some overnight change in psychology. This means that
130 Part I: Trend-Determining Techniques One- and Two-Bar Price Patterns 131

One-bar island

r 1 {

-i Closes higher ~~
than the opening ~~
+----- One-bar island reversal

Rgure 7-12 Emotional bar at bottoms. Figure 7-14 One-bar island reversals.

than tne close

Opening is higher - - - -.. i Close is more than
~ ......... h.',"" dow. tho bar
Chart 7-11 Kenwood 2001 exhaustion bar. (From Pring Research.)


/' i-i-i--1

lJi-ji-j'j{i-I 20"

Figure 7-13 Emotional bar at tops.

exhaustion bars tend to be more prevalent in these very short term charts.
"I Chart 7-13 shows an exhaustion bar for the NYSE Composite. It has all the
Pinocchio Bars
characteristics: a large gap, a close higher than the opening, a wide range,
and so on. It is also followed by an inside bar that adds a further piece of Exhaustion also shows itself in other forms that are different from the char-
evidence that the trend had changed. acteristics we have so far been looking at. I call these Pinocchio barsbecause

.-- I
132 Part I: Trend-Determining Techniques One- and Two-Bar Price Patterns 133

Chart 7-12 Wamaco exhaustion bar. (From Pring Research.) breakout The character Pinocchio cannot hide that he is lying because his
nose gets bigger. In the case of our Pinocchio bar, it is the isolated part of
the bar above (below in a down trend) the open and close that is the big
nose, which signals a probable false move by the end of the bar. Figures 7-15
and 7-16 offer two examples offalse upside Pinocchio breaks. Figure 7-16
offers the idea that when a false break develops above a down trendline, this
is indicative of exhaustion since the price cannot hold above the strong resis-
tance reflected by the trendline.

False impression - - - - .
of strength

Chart 7-13 NYSE Composite 5-minute 1997 exhaustion bar. (From Pring Research.)

612 i-
Figure 7·15 Pinocchio bars.
507 Exhaustion bar

:30 :46 :00 :16 :30 :46 :00 :16

they temporarily give us a false sense of what is really going on. They are
bars in which the bulk of the trading takes place outside the previous and
subsequent trading ranges and that therefore give a false impression of a Figure 7-16 Pinocchio bars and down trendlines.
134 Part I: Trend-Determining Techniques
One- and Two-Bar Price Patterns

Marketplace examples ofPinocchio bars are featured in Charts 7-14 and . As so often happens following whipsaws, the price moves in the direction
7-15. The first shows a break above the trading range, which was nullified opposite to that indicated by the break, .
by the time the bar closed. The second shows a false break to the downside. One important fact about exhaustion is that the extremity of .the move
often proves to be an important support or resistance point. In this respe~t,
it is often a good idea to place a stop loss a lit~e ~it beyon~ the extremity
Chart 7·14 S&P Composite 10-minute 1997 Pinocchio bar. (From Pring Research.) of the Pinocchio bar, provided. of course, that It still results 10 a reasonable
risk reward.
813 Summary
812 • One- and two-bar reversals reflect exhaustion and signal a change. usu-
811 ally a reversal in trend.
810 • To be effective, they must be preceded by a worthwhile m?ve.
809 • Their trend reversal significance is only of short-term duration. What co~­
808 stitutes as short-term will depend on the time span of the .ba~ or bars In
807 question. Daily or weekly reversal patterns will be far more Significant than
806 10-minute bars.
• Reversal bars that contain more of the required characteristics normally
provide stronger signals than those that only have a few.

15h 08:40 OSh 10h 11h 12h 1311


Chart 7-15 S&P Composite lO-minute 1997 P"mocchio bar. (From Pring Research.)

t IH+C 4

782 I

778 I
Whipsaw break
778 i
f.VI¥~;' PO 1:20 :50:00 :20 1:40 1:00 1:20 -T:40 1:00 1:20 1:40 1:00 1:20 1:40 1:00 l::
?;..• .•. y••.. J>hJ20 ..... 09h 10h 11h 12h 0·- • • _ , '

A review of any chan will quickly revea l that p rices usually move in trends.
Quite o fte n , a series of asce n di ng bottoms in a rising market can be j oined
together by a straight line, and eo can the tops of a d escending se ries of r.lIly
peaks . These lines, known as tTmdlina, are a simple bu t invaluable a ddition Not. Iegitimlte t~ ndIine because
to th e technical arsenal [Itt Figs. S-l(G) and (b)]. • only toudJ es one trough

f\vun 8-2 A nontn!ndIine.

How to Draw Trendlines
A proper tre ndline has to connect two or more peak! or trough s; otherwise,
it will be d rawn in space and will h ave no significance. I often see people
constructing lines that only to uch o ne point. as in Fig. S-2. See b ow th e line
misses the second trough. This is a fundamentally important point because
whenever you draw or interpret a trendline, never eser forget that a true
trendline is a graphic 'Nay of representing the underlying trend. Conse-
quently, ifit only touches on e point. it is no t a true tTmdli'IL
Ideally, an up trendline is constructed by connecting m e final low with
th e first bottom in the rally. as line A-D in Fig. 8-3. In the case of a primary
trend . thi s wou ld be th e bear ma rket low a nd th e first intermediate bottom,

,- and vice versa. Fo r line A-D in Fig. S-S, th e fint low d evelops fairly close to
the fin al bottom, which results in a fairly shal low lin e. However, th e price
Flguu 8-1 Up and dov.n treodIines . rallies sharply. wh ich mean s th at th e violation of th e line will d evelo p well
13. 1

" ..
<.;.; •. •
138 Pare I, Trend-Determirin TechnIcpo;s 139

the penetration of rising or falling irend hnes, In .effect, these .1i?CS repre-
sent points of support (rising trendline] a nd resistance (declining trend-
line). . .
The completio n of a rectangle pattern can signify either (1) a ~nallD
the previous trend. in which case it becomes known as a rrvmal ~ .or
(2) a resumption of the previous trend, when it is defined ~ a w.m0lidatl~
after th e final peak. In such situations. it is better to re construct th e trend- or continuation pautm- Similarly, the pen etration of a trendh n~ w1l1 result ~
lin e as the pri ce moves up. In Fig. g.S. this is lin e B-C, which is o bvio usly a eithe r a reversal of that trend or iu co ntinuation. Figure 8-4 illustra tes t;hlS
better reflection of the underlying trend. Down trendlines ace co nstructed point from th e aspect ora rising pri ce trend. In r ig. 8-4(1.1), the t~ ndh ne
using the same principles, but in reverse. j oining th e series of troughs is eventually penetrated o n th e downside. The
fourth peal re presents th e highest poin t in the bull mar~el, so th e down-
ward violation of the uendline signals that a bear market 15 underway.
The upward price trend and (solid) trendline penetrati~n in :-ig. 8-456)
Bar Versus Une or
are identical to th ose in Fig. 8-4(1.1) , but the ac tion follOWIng th~ warnin g
Close-Qnly Charts lignal is entirely different because the trendline violation mere lySign als that
Yo u probably already have n o ticed from previous chapte rs that som e ch arts the advance will con tinue, but at a greatly reduced pa ce. A new (dashed)
are plotted with bars and others as line cham. The q uestion naturally arises trendline is subseque ntly co nstructed. . ...
as to which form of cha rt sh ould be used for th e purposes of trend and Unfortunately; there is no way of tetung at th e orne of th e violation which
trendline anaJys~ In most cases, bar cham offe r qu icker signals, whether poslibility will prove to be th e outcom e.
it is a peak-trough progression . price pattern completion, or trendline vio- Also. valuable dues can be gleaned by applying oth er techniques
lation. Th e problem in technical anal ysis is that speed comes wi th a pri ce. described in subseque nt chapters and by C\'aluatin g th e stale of health of
and the price in th is case is more whipsaws. It is impo rtant to rem ember the market's overall technical structure (examined in Parts II through IV) .
that with traditional daily or Wttkly cham. th e closing pri ce is very impor-
tant because it separates th ose who are willing to take hom e a position
ovemigb t or over a weeken d from those who an: net, Closingprius hauetAm-
Jurt t.rotIiIionaU] btm amsidnedto be~ importanl dum points thtm higlu or lows..
Also, since there is much excit em ent during the day as unexpected news
breaks, highs and lows ofte n represent random points o n the cha rt. For th is
re ason. it is often a better idea to construc t trendlinea using closing data. I
am not going to say that is alwaY' the case because: som e bar trendllnes have
greater significance than close-onlyones based o n th e roles outlin ed for sig-
nificance and described later in th is chapter. Th us. it is alwaJs crucial to applJ
a1l7unDn sm.w as much as strict tech.niaa1 nJa.

Trendline Breaks Can Signal

Reversa1s or Consolidations
Some trench can be sidewaY',th at is, a trading range, so it follows that trend-
lines QI1 also be drawn horizontally. Th e neckline of a head-and-shoulders
(H&S) pattern or the up~r or lower bo undary of a re ctangle is really a
trendline. Th e penetration of th ese lines warns of a chan ge in tre nd, as does
140 Part I: Trend-Determining Techniques Trendlines 141

Using the techniques discussed in Chapter 5 also help. For example, in a Clart 8-1 lnvesco Energy Fund and simultaneous trendline price pattern violations. ThIs chart
shows two examples of a trendline break and pattern completion developing close together. At
rising market, a trendline penetration may occur at the time of, or just the end of 1990, the Invesco Energy Fund violates a nice up trendline and shortly after com-
before, the successful completion of a reversal pattern. Some possibilities pletes a right-angledbroadening top. The end of the decline is signaled with another down trend-
are shown in Fig. 8-5 and Chart 8-1. In Fig. 8-5(a) , the rising trendlinejoins line break and the completion of a reverse H&S pattern. (From
a series of bottoms, but the last two troughs represent reactions from a right
shoulder and head that are part ofan ascending H&S pattern. Figures 8-5 (b) 10.0
and (c) represent a similar situation for a rectangle and a broadening top. 9.6
Figure 8-6 illustrates the same phenomenon from the aspect of a bear 9.0
market reversal. If the violation occurs simultaneously with orjust after the
completion of a reversal pattern, the two breaks have the effect of rein-
forcing each other. Sometimes, however, as in Fig. 8-7, the trendline viola- 8.0

tion occurs before the completion of the pattern. In such cases, the break 7.5
should be regarded as a sign of an interruption of the prevailing movement
rather than as a sign of reversal, because a trend is assumed to continue until
the weight of the evidence indicates otherwise.
During an advance, a setback below the previous trough should develop.
This would confirm an actual reversal. The opposite will occur in the case
of a declining market [see Fig. 8-7(b)]. Further clues as to the significance
of a specific trendline violation can be gleaned from volume characteristics,
as described in Chapters 5, 6, 7, and 22.
For example, if a series of ascending peaks and troughs is accompanied
by progressively lower volume, it is a sign that the advance is running out
of steam (since volume is no longer going with the trend). In this instance,
a trendline violation is likely to be of greater significance than ifvolume had
continued to expand with each successive rally.It is not necessary for a down-
side penetration to be accompanied by high volume, but a violation that
occurs as activity expands emphasizes the bearish undertone because of the
obvious switch in the demand/supply balance in favor of sellers.

(a) (b) (e)

Figure 8-6 Simultaneous trendline and price pattern breaks at bottoms.

Extended Trendlines
(al (bl (el t Most people observe the violation of a trendline and then assume that the
trend has changed and forget about the line. This is a mistake because an
Figure 8-5 Simultaneous trendline and price pattern breaks at tops. I extended line can become just as important as the previous violated line.
142 P.-tl:T~T~ T.......... 143

Flgun 8-1
Delayed pria! pIlttem bn!aks at tops and bottoms.
FIgure 8-8 Extended trendIine::5 'I tops and bottoms.

a..rt 8-2 MSO Eastern fuope: Indo. thovN1g the Irnpc:lrtan::Jr 01 mended lioes. (From

Just as a return move often happens following a breakout from a price
pattern, a similar move, known as a throwhacl:. sometimes develops follow-
ing a tre ndli ne penetratio n. Figu re 8-8(a) shows a trendline reversing its pre-
vious role as support. while the throwback move tunu it into an area of
resistance. Figure 8-8(6) shows the same situation for a declining market.
Chart 6-2. for instance, shows an up tre nd line break for th e Morgan
Stanley Capital Intern ational (MSCI) Easte rn European (Stock) Index.
The penetration of this relativelysteep line resulted in a co nsolidation break-
& the price worked its W<ly higher. it found resistance on two occasions.
Chan 6-3 shOW! the same idea but for a down trendline for the Euro. In
this insta n ce , the line continually reversed its role. starting out as res istance,
th e n support, and finally resistance again.
... .
Logarithmic (Ratio) Versus
Arithmetic Scales
The importance of plotting charts on a logarith mic as opposed to an arith- In a bull market, prices rise slowly after an initial bunt, and then advance
metic scale w:u discussed in Otapter 5. The choice of scale is even more crit- at a steeper and steeper angle as they approach the ultimate peak, looking
ical for a timely and accurate we of tre ndliae analysis. because at the end rather like th e left-hand cross section of a moun tain.
ofa major moveme nt, prices te nd to accelerate in th e direction of th e pre- Chart 8-4 shows that this expo ne ntial movem en t in late 1999 for the
vailing tre nd; that is, they rise faster at the end of a rising tre nd an d decline Athens Oeneral SE Index too k the price well away from the tren dl ine in th e
more sharply at the termination ofa bear market [see Figs. 8-9(4) and (b»). upper. arithmetically based chart. Consequently. the price had to fall that
144 T......... 145

0Md't 8-4 Athen5 Geoe:ral shoowing the dlffeT'el'lan betwem arithmetic and TIltio SCllIing.

(From lIIUIUl.·1

IlIlO Athens SE Atthmelie scale

......"'"""'" . . . ..
. .. . . .

• • • ""' . •
• ...
Significance of Trendlines
It has bee n established that a break in trend caused by the pene tration of a
trendline results in eithe r an actual trend reversal or a slowing in the pace of
the trend. Although it may no t always be possible to aMeSS which of these alter-
natives will develop , it is still important 00 undentand the significance of a

,. ~l
trendl ine penetration; the following guideli nes should help in evaluation .

figure 8-9 Artthmetic and ratio tcPr breaks at lop.

much farther before a penetration of the tre ndlioe cou ld take place. As it
result, up trendlines arc violated mere quickly on a logari th mic than on an
arithmetic scale. This can be seen from the vertical arrow. ConYe~ly. down
trendlines are violated sooner on an arithmetic scale. Note that a t the end I.engtt, of the Uno
of the chart, the up trendline on the arith metic scale has bee n violated. The size or length of a trend is an important factor, as with pri ce pa tterns.
whereas the logarithmically based one has not, If a series of asce ndi ng bottoms occurs over a 3- to {-week span, the result-
Ce nerally speaki ng, penetration of a logari th mically based tre ndline i5 ingtrendline is only of minor importance. If the tre nd extends over a pe riod
more accurate in reflec ting trend reversals than the pe netration of an arith- ofl to 3 yean, howeve r, its violation marks a significam juncture poinLJust
metically based trendline. ~ember, big tren ds result in big signals, small trends in small signals.

- -- - --
146 Part): Trend-Determining Techniques Trendlines 147

Number of Times the Trendline trendline is measured during a rising trend (AI in Fig. 8-11). This distance
Has Been Touched or Approached is then projected down from the point at which the violation occurs (A 2) .
A trendline derives its authority from the number of times it has been The term price objective is perhaps misleading, Objectives are usually
touched or approached; that is, the larger the number, the greater the sig- reached when a trendline violation turns out to be a reversal, but because
nificance. This is true because a trendline represents a dynamic area ofsup- they are more often exceeded (as with price patterns), the objective
port or resistance. Each successive "test" of the line contributes to the becomes more of a minimum expectation. When prices move significantly
importance of this support or resistance role, and thus the authority of the through the objective, as in Fig. 8-12, this area often becomes one of resis-
line is a true reflection of the underlying trend. Just remember that a close tance to the next major rally or support for a subsequent reaction.
encounter with the line (an approach) is almost as important as an actual Figure 8-13 shows the same possibilities for an upside breakout.
touching of the line because it still reflects the line's importance as a sup-
port or resistance area.
Also, if a line gains significance from the fact that it has been touched or
approached, the extended line will become equally as important, but from
a reverse point of view since extended lines reverse their support/resistance

Angle of Ascent or Descent

A very sharp trend, as in Fig. 8-10, is difficult to maintain and is liable to be
Figure 8-11 Measuring implications.
broken rather easily,even by a short sideways movement. All trends are even-
tually violated, but the steeper ones are likely to be ruptured more quickly.
The violation of a particularly steep trend is not as significant as that of a
more gradual one. The penetration of a steep line usually results in a short
corrective movement, following which the trend resumes, but at a greatly
reduc~d and more sustainable pace. Usua!!y~lh~~~.!A~!!...~f~.s~!,;~l!_
trendhne J"eIJr.esents a continuation rather than a reversal break.
""-'-'~'--' ,-' ..,.-~",." -, "'<"'-"~- •• '-~ • • . . ,_." .• _,~_.,,>,.- ,', ".-.- '-"'.

Measuring Implications
Trendlines have measuring implications when they are broken,just as price Figure 8-12 Downside measuring objectives.
patterns do. The vertical distance between the peak in the price and the

Figure 8-10 Sharp-angled trendlines.

Figure 8-13 Upside measuring objectives.
148 Part I: Tnmd-OeterminIng T~

Ti me and again. th ese pri ce o bjective areas prove to be important sup-

port o r resistance points. Unfo nunately. th ere ill no way to determine where
T........ 149

dr.lw lines that are parallel to those basic tre ndlines, as shown in Figs. 8-15{a)
and (c) . In a rising market. the parallel line known as a ~ ~j oins
the actual juncture point will be' fo r any rally or reaction. This emphasizes the tops of rallies (AA in Fig. 8-15{a)) , and during declines the return line
a point made earlier th at there is no known way of consistently determin- joins the series of bottoms tBB in Fig. 8-I5(c»). The area ~tween these
ing th e duration ofa price moveme nt. It is only possible to speculate on the trend ex tre mities is kn own as a mnd channa. Figure 8-15( b) IS a rec tangle ,
probability that a specific area will prove to be an impo rtant tu rning po int. which is a hori zon tal fonn of a trend channe l. _. ----.
Th e return line is useful from two points of view.~~t; i~ .r.ep~!!ts ?:"
area o! s~ --!! o r ~~~, depending on th~ ~irection of the trend.
Corrective Fan Prindple Second, and perhaps more impo rtant, th~ne~bo~ .~ thereturn trend-
line re presents.a signal that either th e tre nd will accelera te or th at a reve r-
At th e begin nin g of a new primary bull m ark et. th e in itial int ermediate rally sal of at least a temporary proportion in th e basic trend is abo ut to take p1a~~ .
is often explosive. and so th e ra te of ascent is u nsustai nably steep. Th is hap- In Fig. 8-16(0), the violation of the return line signifies that the pnce
~ns because the advance is o fte n a tech ni cal reaction to the previous advance has begun to accelerate. Figure 8-16(b) shows the same idea for. a
overex tended decline. as speculato rs who were ca ught short rush to cover downside breakout. In effect, the channel in Fig. 8-16{a) represents a ns-
thei r positions. & a resu lt. the steep trendline constructed from the first ing rectangle a nd th e tre ndline violation rep resen t! a breakou t.
minor reaction is quic kly violated . On the other hand, if th e angle of th e tre nd channel is much steepe r, as
This is represented as line AA in Fig. 8-14. A new trendliee is men con- in Fig. 8-17(0) , the violation of the return line reprcsenl5 an ex~austion
structed. wing the bottom of this first intermediate decline (A.B). The new move. The failure of the price to hold above (below) the ret urn hne. th e~
line rises at a less rapid rate than the in itial one. Finally, the process is signals an important reversal in tre nd. This is ofte n the case if the action IS
repeated, resulting in construction of a third line, AC These lines are known accompanied by high volume.
as Jan lirus. Th er e i.!I an established principle th at o nce the thi rd trendline Consider a situation in which a man is. srt~ng a thick piece of~ . At
has been violated, th e end of the bull market i.!I co nfi rmed. In some respects first, his sawing stro kes are slow but deli'6era\C.;,gradually. he re~lZeS that
these three rally po ints and trendlines can be compared to th e three stages his wi. is going to tak~ some time , bes;o1I1es fria.Strat.ed, and slowly mcre ases
ofa bull or bear market, as outlined in Chap ter 3. The fan pri nciple Isjust the speed of his strokes. finally, he ~ into a fran tic effort and is fo rce d
as valid for dcwntrends and can also be used for de term ining intermediate ." La give up his task for at least a tem porary pe riod because of complete
as well as primary trend movements. exhaustion, figure 8-17{b) shows an exhaustion move in a dedinin.g ma~­
ket. In this case, the expanding volume at the low represents a selling cb·
max. A5 a general rule, the steeper the channel. the more likely it is th at
Tre nd Channels the breakout will tu m out to be an exhaustion move.
So far, only th e possibilities of drawing trendLines j oining bottoms in rising
mark ets and tops in declin ing ones haw: been examined. It is also useful to

... . -,

, .- , .:.J-

_- -
• ~

7VV\L ......
. ..

• ,.-' . ...
Apre 8-14 CorrectNe Ian pr1ndPe- ~.. Fleur. 8- 15 Trend channels.
" ..:?_~ .
..1 ..
150 T......... 151

This temporary break o ften re p rese n ts a sign al that the prevailing tren d h as
much further to ru n, in th~ ,q me way as a 'whipsaw breakout from a price
pattern. It also mses a dil¥'m~ in the way in which a trendline should be
constructed. In Fig. 8-18. for instance. ~ see a false break above tre ndline
AB. Should An now be abandoned or should the peak of the exhaustio n
break be connec ted to the rall y high to form a new tre nd line? Again . it is a
maue r of common sense. O n the one hand, the whipsaw break is techni-
cally the correct place to draw the line , but commo n sense suggests that the
original line is a better reflection of the underlying trend. After all, at the
time of the wh ipsaw, it has bee n to uched three times . If the line is then
drawn to reflec t th e b reak, it will o n ly have been to uc he d twice, once at th e
outset and once at the whipsaw peak. In a sense, tJae whipsaw is adding fur·
&Mr cmlibilitJ 101M mmaJ lim b«awt thepria waf urwble ID hold abovt it. If we
had come upon this situation after the whipsaw broil. and tried to co nstru ct
a line. it would have been even more obvious that line ABwas far superior
to line AC beca use it has be en to uche d or approached on far more
occasio ns.

'" •
, ;i .."
152 Part I: Trend-Determining Techniques Trendlines 153

Summary Chart 8-6 S&P Composite 1966-1989 showing an exhaustion breakout. This chart shows
that a resistance trendline joining the 1974 low and 1978 highs was temporarily violated. ~is
proved to be an exhaustion move since the S&P Composite was unable to hol~ above the lm~.
• Trendlines are perhaps the easiest technical tool to understand, but con- This failure was foUowed by the 1987 crash. Not all exhaustion moves result m such dynamIc
siderable experimentation and practice are required before the art of consequences, but they certainly warn of potential trouble and should never be ignored. /From
interpreting them can be successfully mastered.
• Trendline violations signal either a temporary interruption or a reversal
in the prevailing trend. It is necessary to refer to other pieces of techni-
cal evidence to determine which is being signaled (see Chart 8-5).
• The significance of trendlines is a function of their length, the number
of times they have been touched or approached, and the steepness of the
angle of ascent or descent.
• A good trendline reflects the underlying trend and represents an impor-
tant support and resistance zone.
• Extended trendlines are an important concept and should not be over-
looked (see Chart 8-6).
• Exhaustion breaks often possess good predictive power.

Chart 8-5 Value Une Composite Index, 1989-1990. This represents an example of a major
trendline violation occurring in combination with a price pattern completion. In this case, the

formation was a broadening one with a slightly rising trendline (the almost flat bottom). Some
form of poetic license is often needed in interpreting charts since this particular one could not
strictly be interpreted as a broadening formation with a flat bottom, but the effects were cer-
tainly the same. (From

: : : : : :
ISO .00 : : : ; : : )lO.OO
: ~-=PJIL: : :
).0$.00 : .~: : :

...... : : :A

. . . .

2:t$ .00 : : : - UI.GO

ito .00

us .00




t:CO.O O

ISS .00

----.... -'-c-'---'-:--'
Figure 8-17 Exhaustion \•.
152 Part I: Trend-Determining Techniques Trendlines

-6 S&P Co osne 1%6-1989 showing an exhaustion breakout. 1?is .chart sh~1S

Summary ~~lstance trendlin: joining the 1974 low and 1978 highs was temporoldanly bo
: . -:n
to be an exhaustion move since the S&P Composite was unable to h a v e e m~.
• Trendlines are perhaps the easiest technical tool to understand, but con- ~aiIure was followed by the 1987 crash. Not ail ~austion moves result in such dynamiC
siderable experimentation and practice are required before the art of consequences, but they certainly warn of potential trouble and should never be ignored. (From
interpreting them can be successfully mastered.
• Trendline violations signal either a temporary interruption or a reversal
in the prevailing trend. It is necessary to refer to other pieces of techni- 151
cal evidence to determine which is being signaled (see Chart 8-5).
• The significance of trendlines is a function of their length, the number
of times they have been touched or approached, and the steepness of the
angle of ascent or descent. ZOG
• A good trendline reflects the underlying trend and represents an impor-
tant support and resistance zone.
• Extended trendlines are an important concept and should not be over-
looked (see Chart 8-6).
• Exhaustion breaks often possess good predictive power.

Chart 8-5 Value Une Composite Index, 1989-1990. This represents an example of a major
trendline violation occurring in combination with a price pattern completion. In this case, the
formation was a broadening one with a slightly rising trendline (the almost flat bottom). Some
form of poetic license is often needed in interpreting charts since this particular one could not
strictly be interpreted as a broadening formation with a flat bottom, but the effects were cer-
tainly the same. (From

JtO .00 :uo .•o

)05.00 )01.'0

JOO.'. )00.'0

2!1S .00 its .• 0

1500.00 Uo-.O'

tls.oo us."

271,00 t7S .10

270.0C> 1'0."

us .00


i us.oo
't!o tl'!'k III,.. hAY
-- •.......
__ :.--'---~_....' -...
fIIUG SCI' ec r "au
etc 1ft" FlU "all: 111'1
Moving Averages 155

Table 9-1 Simple MA Calculation

Date Index IO-weektotal MA

Jan. 8 101
15 100
22 103
29 99
Feb. 5 96
12 99

Moving Averages 19
Mar. 5
12 89 966 96.6
19 90 955 95.5
26 95 950 95.0
It is evident that trends in stock prices can be very volatile, almost haphaz- Apr. 2 103 950 95.0
ard at times. One technique for dealing with this phenomenon is the mov-
ing average (MA). An MA attempts to tone down the fluctuations of any
price series into a smoothed trend, so that distortions are reduced to a min-
Chart 9·1 Cash Wheat centered versus a noncentered average. (From
imum. There are many variations of MAs used in technical analysis. The
three most common are: simple, weighted, and exponential. The con-
struction and use of these averages are different; therefore, each type will 270
be dealt with in tum. 260
Simple MAs 210
A simple MA is by far the most widely used. It is constructed by totaling a 180
set of data and dividing the sum by the number of observations. The result- 270
ing number is known as the average, or meanaverage. In order to get the aver- 250
age to "move," a new item of data is added and the first item on the list 240
subtracted. The new total is then divided by the number of observations, 230
and the process is repeated. 220
For example, the calculation of a 10-week MA would follow the method 200
shown in Table 9-1. On March 12, the total of the 10 weeks ending on that
date is 966, and 966 divided by 10 results in an average of 96.6. On March
19, the number 90 is added, and the observation of 101 on January 8 is
deleted. The new total of 955 is then divided by 10. The calculation ofa 13-
week MA would require totaling 13 weeks of data and dividing by 13. This A comparison of the price index with its 13-week MA shows that the aver-
calculation is then repeated in order to get the average to move. A 13-week age changes direction well after the peak or trough in the price and is there-
MA is shown in the upper panel of Chart 9-1 by the dashed line. Generally fore "late" in changing direction. This is because the MA is plotted on the
speaking, a rising MA indicates market strength and a declining one denoteS thirteenth week, whereas the average price of 13 weeks of observations actu-
weakness. allyoccurs halfway through the 13-weektime span, that is, in the seventh week.

156 15 7 •

MAs. which are discussed later. The major technical principles of inter-
i p reting MAs ar e as fo llows:

• I\n MA is 6 smoothed lIt.'I":titm of 6 tTmd, ·and tAt tJwragt iudf is an arra oj sUJr

I port and misttJ7ICt. In a rising market. price rea ctio ns are ofte n reversed
as they find su pport in th e are a o f the MA. Similarly, a rally in a d eclin-
ing marke t often meets resistance at an MA an d turns down. The mo re
\ If it is to rdlecl the underlying trend correctly. the latest MA should be cen- times an MA has been touch ed. that is. when it acts as a support or resis-
I tered . or plotted, on the 5n'enth week, as shown in the lower pand ofOtan 9-1.
lfthe centering tec hnique had been......~~ \the example, it would have
tance area. th e greater th e significance when it is viola ted.
• 1\ carifuIJ.J ~ MA should nJled rJw undnlJing m:M; its violation lhnTforr
been necessary to wait 6 weeks before a3certammg whethe r the averag e had warns cMt 6 chongt in trmd rMJ abMdy Juwe ttJkm piau. If the MA is nat o r
changed di rection. Thai is why the blank area on Chart 9-1 e xists between has alre ad y changed di recti on . its violation is fairly ~n,l:hJ5ive p roof th at
I th e last plot for the average and th e, last plo t for the data point,
A time d elay, th ough it is an irrli!i:nt, is not particularly critical when ana-
the previous trend has reversed. ( ...." i'"
• If tAt violation occun whilt tAt .MA is still proe«ding sharply in tAt dirn:lUm oj
lyzi~g other tim e series such as economic data. H owever; give n the r ela tively tAt fImJoiling trtnd. tJais shouldbe tnaled as a prdimifU1.TJ warning tAal a 'nnd

I rapid movement of prices in th e financial mark ets and th e consequent loss

of profit potential. a delay or this nature is totally unacceptable. Technicians
mJl!nlJ1 has ttJkm plau. Confirmation should await a flattening in the angle
of ascent or descent, a change in direction in the MA itself, or alternative
have found that, for the purpose of identif)ing trend reversals, the best technical sources. The crossover of a movi ng a~rage with a sharp a ngle
results are achieved by plotting the MA on the final period. of asce nt or d escent is akin to th e violation of a trendline with a sharp
I A change from a rising to a declining mar ket is sign aled when th e p rice
m oves be low its MA. A bu llish signal is tri gge red when th e price rallies above
angle .
" - . ....... ~\l
. _ .1 -" .- v
• GtmrolJy sp«Jcing. tAt Jorig;tht 'imt span wvmd by an.MA. the grrou:r tAt rig-
th e average . Since th e usc of MAs gives clea r-cut buy and sell signals, they nifimna oJa ~ signaL For ins tance. the violation of an 18-month MA
help to eliminate some o f the subj ectivity associated with th e co nstructio n is substantially more important than th e crossover of a 3o-day MA.
and interpretation of trendlines . • Rrrxr:saIs in tAt dmdion of an MA an' wuaIly morr nliabk tMn 6 n MA
More often than not, it pays to take action based o n MA crossovers. pro- ao.uowT. In instances in which a change in d irection occurs close to a mar-
vided the parti cu lar time span in question has proved reliable in th e past, turning point. a very powerful and reliable signal is given . However,
The degree of accuracy depends substantially on the choice of MA, as diJ. in mos t instances, an average reverses wen afte r a new trend has begun
cussed b.te~, as well as the volatility of the security in question. The length and so is o nly useful as a confirmation .
of the. mOVIng average will also have an in fluence o n its accu racy, Ge ne rally
speak.l~g, th e lon ger th e tim e span, the more reliable th e average. In effect. In short , think of a n average as a type of moving trendline that ob tains
a mOVIng average for any given time span o n an in traday chart is likely to its significance fro m its length (time spa n) .lbe numbe r o f tim es it has been
be much less re liable than one co nstructed fro m mo nth-end data on a touched or ap proached, an d lts angle of asce nt o r descent.
monthly chart. Now though, we need [0 examine some of th e characteris-
tics of MAs in greater detail.
What Constitutes a Valid Crossover
A C71WoVtt' is any pe ne tration of an MA. However, close observation of any
Characteristics of chan featu ring an MA will usually reveal a numbe r of whipsaw, or false, elg-
S imple MAs nals. How can we tell which o nes are going to be valid ? Unfortunately, there
is no way of kn owing for certai n. Indee d , man y whipsaws ca nno t be avoid ed
Allboug h the examples in th is an d th e fo llowing sections featu re simple and should be regarded as a fact of Hfe. Howeve r, it is possible to avoid some
MAs, th e principles o utlined can be applied to weighted and exponential or these close calls by wing filte ring techniqu es. The type of filler to be used

~~.., ';,*" <.

'" ~
'~ . .
158 Part I: Trend-Detennlntng Techniques 159 ••

depends on the time span in question and is ~ry much a matter of indi- data, this approach would mean waiting fo r the second or third day before
vidual experimentation. concluding that the avenge had been violated. A more sensible method. is ,.
For example. we may decide to rake action on MA Cl'OSSOW:r5 fo r which to usc a combination of the period and percentage penetration for decid-
a :5 percent penetration takes place and to ignore aJl others. Violations of ing whether OJ. C'U55OYeT is valid.
a 4O-week MA might result in an a~~ price move of 15 to 20 percent, In An example is shown in Olart 9-3, where a joint trendline and MA vio-
this instance. a:5 percent penetradon would be a reasonable filter. An exam- lation develop for the Philadelphia Gold and Silver Share Index in August
ple is shown in Clan 9-2. On the other hand, since 3 percent would prob- 1994. Later, we see an MA violation and a head-end-shoulders (H&5) top
ably encompus the whole move signaled by a lo-day MA CI'055OYCT. this kind completion. both ofwhkh developed around the same time. Chart 9-4 shows
of filter would be of no we whatsoever. a similar situation .,.,;th Alberto Culver and a 5O-day MA.
Some analysts, recognizing that one-period whipsaws an quite common, MAs are usually constructed from dosing data, Oosing pri ces are more
require an MA ct'OSS<M:r to hold for at 1e2St one period. In the case of daily reliable than highs, lows, or openings because they rdlect positions that
investors are willing to carT)' overnight or, in the case of weekly charts, over
the weekend, lntraday trading can be subject to manipulation or dis toned
by an unwarranted emotional attitude to news events. For this reaso n, it is ,,
best to wait for the closing price to penetrate the average before conclud- .,
ing that a crossover has taken place. If intraperiod activity is used for MA
violations, it is usually best to calculate an MA based o n daily lows or highs.
O ne exception woul d occur when a bar cha n touc hes an MA o n num e r-
ow occas ions. In many instances, th e MA is clearly a sign ificant
support/resistance point and its violation should be treated with respect. •

~ 9- 3 PhlIadeIphIa GokIand SlM1:r Shan>. Index 1994. (From Mortln Prlng'slnlroduc-

flon to Tec:hnfeQl Anol).lflJ.)


'" .
'" ·
... n, .
0 ••

.. ,.


"" t llO

... ... '" ... ... . :/:... ,'" ,~ , .. Joint break

~epI8 mb.r O~ b .r

160 Part I: Trend-Determining Techniques Moving Averages 161

Chart 9-4 Alberto Culver and a simuItaneous MA trendline break. This chart features Alberto Chart 9-5 Canadian dollar and two MAs. (From
Culver together with a 50-day MA. Note how it was possible to construct a trendline that inter-
sected with the MA. Since both represent resistance, their joint violation represents two pieces
of evidence that prices were headed higher. A third came in the form of an expansion in v0l-
ume. (From



Joint break
50-day MA



In Chart 9-5, the dotted line shows a 25-day MA calculated from intraday
lows. The solid line represents the same average based on closing-only data.
If intraday crossovers of the former had been used as stop-loss signals on
the downside, considerably fewer whipsaws would have been generated than Even if the 48-month average were shortened to 24 months and the 5-day
if the MA had been based on closing data. average expanded to 4 weeks, for example, using the crossover signals would
still cause the 24-month average to give an agonizingly slow confirmation
of a change in trend. The 4-week average would be so sensitive that it would
Choice of Time Span continually give misleading or whipsaw signals [see Fig. 9-1 (a)]. Only an MA
MAs can be constructed for any time period, whether a few days, several that can catch the movement of the actual cycle will provide the optimum
weeks, many months, or even years. The choice oflength is very important trade-off between lateness and oversensitivity, such as the 1D-month MA in
For example, if it is assumed that a complete bull and bear cycle lasts for 4 Fig. 9-1 (b).
years, an MA constructed over a time span longer than 48 months will not The choice ofMA depends on the type of market trend that is to be iden-
reflect the cycle at all. This is because it smoothes out all the fluctuations tified: short, intermediate, or primary. Because different markets have dif-
that take place during the period and will appear more or less as a straight ferent characteristics and the same markets go through different cyclic
line crossing through the middle of the data. On the other hand, a 5-day phenomena, there is no such thing as a perfect MA. In recent years, exten-
MA will catch every minor move in the stock cycle and will be useless for sive computer research has been done on the optimum MA time span. The
the purpose of identifying the actual top and bottotn of the overall cycle. Conclusion from all sources is that there is no one perfect time span.
162 163 '.

Qart 9-6 St«1dllrd &: PoeJr'$(S&F'l Composite, ~ MSCIWorId Stock Index...t 4O-week
MAs. This chart shows • 4O-week MA et work In three different marl<ets. Crossowrs 01 each
Indo .....b;ed to whipsaws fn:lm time to time , but on balance this aventge Is stili f~ rdt-
,, ~ 10 · "'OIl'th
able. Note that the 4 (h...oeek llYeIllQe Is tontlnually being used. as f1 support or resstence !eYe!.
• ......,
•• ....
,t .,,,,,, .' ,• ..,
,,, , , • S&P"""""",.

• ',, ' " " ,<-- IN.....

2. ·~

' Q.. wt. o~
,, -; »

4 - ...,
...;"• • u r...
- , . . .' (:::I
".t>~' ~".. ~
(.j (bj

FlguN 9- 1 W 4-week MA. lb)lD-month venus 24-month MA.

What may work extremely well in one market OYer o n e specific period of .,
time is unlik.ely to be duplicated in the future . When we talk about choice ",·
of time span, we are really tryi ng to identify an MA that will work most of
the time with a specific time fram e, that is, sho rt, int erm ediate, or long.
Table 9-2 Suggested Time Frames
Ge-nerally speaking, lon g-term time spans are less infl u en ced by man ip ula-
tion and knec:jerk random reactions to unexpected news th an are short- Inle rmedi3llHenn
term on es. This is why lon g time spans usually give the best test results.
Examples of a 4()..wet:k spa n are featured in Chan 9-6. Research also shows
that simple averages generally ou tperform weighted and exponential o neL
12-mon lht
Recognizin g these limitations, the tim e spans in Table 9-2 are suggested. IS-mo n u.
The importa nt thing to re me mber is that an MA is om tech n ical 100 1in 24-mo mh
the technical arsenal that is wed with other tec hni ques as part of me arto!
· ~ommcned~ by Gordon, TI&t SId Marltd J!Wc4lon. In~tors Prna, P~isad~
identifying trend reversa ls. M.NJ.. l 968..
t~ by R*" W. CO""' ....d Thomas A. M~ in TI&t £~ fl{ Tdr&iaJl M4'NI
hodic.aIon. Dow JOneHrwin. HCltDeWOOd. m..1988, to bf: the belt ~e £01" the U.s. ,todr. IrW""
"" Uling weekly da ta.
Advancing Simple Moving Averages Ubid.; ~ to be the belt

A technique that has a Jo t of po te nti al. bu t iJ n ot widely use d , is to advance

Pe'rAgl: for the U.s. 110( " martel usin g monlhly data.
an MA. In the case of a 25-day MA. fOT example , th e actual p lo t wou ld not


~ •
164 Part I: Trend-Detenrtinn T~ 165
be made on the 25th day, but advanced to the 28th or SOth. and so forth. 0M0rt 9-7 The DJlA and Ifne MAs (H.M. Gartley, Profit. In tIw Stock Morh t, lambert
Gmm Publishing, Porroerov, WMhlngton , 1935 .)
The advantage of this approach is that it delays the crossover and filters out
occasion al whipsaws or false signals, In Proftls in the Slodr. MarltLt, H. M.
Gartley' calculated ma t du ring the period of 1919 to 1939, which covered
.~ ,.
almost all kinds of market situations. th e use of a simple 25-d2y MA ~ sign" on Iol'l;er
moving ........
neued 446 Dow points (slightly better than 433 points for the 3O-day MA
an d far better than the 516 and 216 for 40- and l !H:iay MAs, respectively) .
However, when the 25-day average was plotted on the 28th day, crossovers
" ,i \ -,-. . . ,
~ ,

resulted in an increase of 231 point! to 677. Th e 3o-day MA. when advanced

3 days, also produced superior results, with an additional gain of204 po ints 70 i
r. "-'/ ';"
• ,\\ ," "

./ : ' \\ \.
for a total of 637. Chart 9-7 represents three MAs. as calculated by Ganley. 'I .' :, ,
,. \
" '.\
Although these MAs arc plotted nonnally. the whipsaw signals would be
avoided by moving th e MA forward 3 days, as sh own o n the chart. I : ! ': ",-, Observ. f.I.. ';gn."
, • \' "J .:~. in tr.ding . r• ••.
Although the 25-day MA that advanced 3 days may not ultimately prove " . ;' I ;Il ..
to be the best combination . the technique of advancing a n MA is d early on e Ob.. rv. pltffec:l : :JO.dsy moying ,y" ' II' .
oper8lion 01moying , : plon ed on 30th day ..-
that could be usefully incorpora ted into the technical ap proach. It is always .... . \11. in . :- r 1s--d8V moving ~ .v.;..
difficult to kn ow how much to advance an MA. Experimentation is the '0 di.gon_ tr""c1. ;"_ ,' ' - plon ed on 1Slh dr(.
answer, One possibility is to advance the average by the square roo t of
the time span, For example, a !&day MA would be advan ced by 6 days (the
§ : ~ ~-dey mWi~ ......Qt.
i • I plotted on.lOth dIy. ..: .
.! ,f/'
square root of 36 = 6).1 See Chan 9-8 featuring the Madrid Ce neral lndex.
ee i •
: , I ,,
,, i d
Convergence o f , ,: 'I
Simple MAs •• f ,f
" ,
• Eu mplt. oIl....
A sharp price move is ofte n preceded by a gradually narrowing trading '0 · psnstr.tion. which

_uld M . limln. ' .d
range , In effect, decreasing pri ce fluctuations reflect a vel)' fine balance bv inc.N" in INd.
between buyel'$ and sellers , When the balance is tipped one wayor the other,
the price is then free to embark upon a major move.
v .. \ \
\-.~ i;
Th is kind of situation can often be identified by plotting several MAs and
observing them when they are all at approximately the same point, Otar1 ., "' \
\ .
I :' ,

9-9 shows the Spot Eureyen between 2000 and 2001 together with three MAs
.... 01 DowJ on
1&--, 3O nd 4O-dsymw lng _r~..
.,..11. of 30 ind" tt ri• • ~ocltt
plon.d wilh minimum lNd
. howinll
yllrious movi" g . Y"1gI phsnomeNI
' DJ Ind lalU'iab and three MAL (HM . GMtky'. ""fib ... Ilw SId M~ Lamben GanD
Publishing. p~.....,. WA. 1981.)
"Artllur Sbrew, Toruf- fI{ . P'IvJaAt-l e--iitJ a.-r A...Jpt, Commodity ~b
8w'eaIl, N~Yort, 1980.

166 Part I: Trend-Determining Techniques Moving Averages 167

Chart 9-8 Madrid General and an advanced MA. Here we see a 5<K1ay MAtogether with of differing time spans. Note how the three MAs almost completely converge
a 50~y average advanced by seven periods. Note how the advancedaverage filtered out sev-
eral whipsaw signals. This Is by no means a perfect approach, but it definitely comes into lis just before the price embarks on a sharp decline in November 2000. The
own after a sharp raDy or decline. (From convergence of the averages tells us that the balance between buyers and
sellers is very evenly matched and warns that a major move is likely. The
actual signal comes from the violation of the up trendline. A similar com-
bination develops in February 2001, only this time the 25-day average does
110O not participate in the convergence.


800 Multiple Simple MAs

750 Some techniques of trend determination involve more than one MA at a
time. Signals are given by a shorter-term MA crossing above or below a
700 longer one. This procedure has the advantage of smoothing the data twice,
which reduces the possibility of a whipsaw, yet it warns of trend changes fairly
quickly after they have taken place (see Chart 9-10). Two averages that have

Chart 9-10 MSCI World Stock Index and two MAs. In this chart of the Morgan Stanley
Capital International (MSCn WorldStock Index, the 10-week MAcrosses below the 3Q-week
MA in early 1998 and late 1999. However, the 3<K1ay series is still advancing, so no sellsig-
nel ts given.The late 1998 declineshows the weaknessof this systembecausethe sharpinter-
mediate correction resulted in a sell signal almost at the bottom. Because the bottom was
V-shaped, the counterveiling buy signal was not triggereduntil the pricehad experienced a good
rally. After that, the 10/30 system would have maintained a long positionuntil just after the
Chart 9·9 Spot Euroyen and three MAs. (From final peak. (From



~----- No signal

_----...------ t
3D-week MA
1997 1998 1999
168 169

been found reliable in detennining primary market moves are the 10- and Ov.rt 9-11 U.S. DoIM h\ex 1995. (From Mortln Pring's Introduction 10 TechnkGl
Ana lysis .)
3O-week MAs. when wed together. For the purpo~ of simp lifying th e cal-
culation . th e weekly d osing price is wed. rather than a ~y a~rage.
Signals an: given when th e }G-week avenge moses below th e 3O-wttk aver. fl JJ
age an d whe n the 3O-week aYC I1l~ itse lf is declining . This developmen t 96.5
warns that the major tre nd is down. It is not assumed to have reve rsed until 96JJ
both averages an: rising simul taneously, with th e lO-week. higher than the SU
3Q.week MA. A valid signal is n ot given if the to-week average rises above SU
......1 11
th e 3Q.week average while th e lo nger aY'l:TIlge is still d eclining (and vice vena B4JJ
fo r bull markets). By defini tion , th ese warn ing signals alwaysoccur after the 83.5
ultimate peak or trough of ltoci. prices and serve as C(JTljirm4lUm ora change 83JJ
in trend rather than as actual juncture points in the mselves. SO!
MAs should always be used in conjunc tion with o ther in d ica ton. This is B2JJ
because prices occasio nally fluctuate in a broad sideways pattern for an
exte nded peri od of tim e. resulti ng in a series of m isleading signab. Chart eo.5 4
9-11, feat uri ng th e Ij.S, Dollar Ind ex in 1995. sh ows an exam ple ofa pe riod eoJJ
in which many misleading crossovers wen: experienced, Usually. IUch frus-
trating trading-range action is followed by an ex eemely strong tre nd in
which th e losses in curred from the trendless period of whipsaw lignals are
more than mad e up for,
Whenever it is obvious that an MA has entered a trading range. and th is ~ 9-12 The EJ..rotop h\ex 1993-1995.
migh t take two or th ree whipsaws. it is bot to igno re it an d co ncentra te on
tre ndllnesjoining peaks and troughs oflhe range shown in Charu 9-11 and
9-12. using th eir breakouts as a basis for buying or selling,

Weighted MAs '"

An MA can correctly represent a tre nd fro m a statistical point of view only
if it is centered, but centering an average delays the signal. for th e reasons
discussed previow ly. O ne techn ique th at attempu to overcome th is prob-
lem is to weight th e data in favor of th e most r ecent observations . An MA
co nstru cted in th is manner can turn or reverse direction mu ch more quickly
than a simple MA, which is calculated by trea ting all th e data equally.
" .' _--l.:...__~JI.~_
5Ck1ay MA ....

Th en: are countless ways in whic h da ta can be weighted. but the m05t 110
widely used method is a technique where by the fi nt period of da ta is mul-
tiplied by 1. the seco nd by 2. th e third by~. and 10 o n until th e latest one
IN ...
170 Part 1: Trend-Determining Techniques Moving Averages 171

is multiplied. The calculations for each period are then totaled. The divi- If the difference between the new weekly observation and the previous
sor for a simple MA is the number of periods, but for this form of weighted week's EMA is negative, as in the reading 99.00 versus 99.64 for January 29,
average, the divisor is the sum of the weights; that is, 1 + 2 + 3 + 4 + 5 + the exponentially treated difference is subtracted from the previous
6 = 21. For a 10-week weighted MA, the sum of the weights would be 1 + week's EMA.
2 + 3 + 4 + 5 + 6 + 7 + 8 + 9 + 10 = 55. Table 9-3 illustrates how the The exponent used varies with the time span of the MA. The correct expo-
calculations are madeAnother method is to calculate a simple MA, but, in nents for various time spans are shown in Table 9-5, where the time periods
doing so, to use the most recent observation twice, which doubles its weight have been described as weekly. In effect, however, the exponent 0.1 can be
The interpretation of a weighted average is different from that of a sim- used for any measure of 20-houfs, days, weeks, months, years, or an even
ple average because the weighted average is more sensitive. Warning of a longer period.
trend reversal is given by a change in the direction of the average rather Exponents for time periods other than those shown in Table 9-5 can eas-
than by a crossover. ilybe calculated by dividing 2 by the time span. For example, a 5-week aver-
age will need to be twice as sensitive as a 100week average; thus, 2 divided
by 5 gives an exponent of 0.4. On the other hand, since a 2o-week average
should be halfas sensitive as for a IO-week period (0.2), its exponent is halved
Exponential MAs
to 0.1.
Weighted MAs are helpful for the purpose of identifying trend reversals. If an EMA proves to be too sensitive for the trend being monitored, one
However, the time-eonsuming calculations required to construct and main- solution is to extend its time period. Another is to smooth the EMA by
tain such averages prior to the Widespread use of computers greatly another EMA. This method uses an EMA, as calculated previously, and
detracted from their usefulness. An exponential moving average (EMA) is repeats the process using a further exponent. There is no reason why a third
a shortcut to obtaining a form of weighted MA. In order to construct a 20- or fourth smoothing could not be tried, but the resulting EMA, while
week EMA, it is necessary to calculate a simple 20-week MA first, that is, the smoother, would be far less sensitive. Remember, all forms of MAs repre-
total of 20 weeks of observations divided by 20. In sent a compromise between timeliness and sensitivity.
The 20-week average becomes the starting point for the EMA. It is trans- Bydefinition, EMA crossovers and reversals occur simultaneously. Buy and
ferred to column 2 for the following week. Next, the entry for the 21st week sell signals are therefore triggered in the same way as simple MA crossovers.
(January 8 in the previous example) is compared with the EMA, and the In their book The Encyclopedia of Technical Market Indicators,S Colby and
difference is added or subtracted and posted in column 3; 100 - 99 == 1.00. Meyers tested all the time spans between 1- and 75-week EMAs for the U.S.
This difference is then multiplied by the exponent, which for a 20-week EMA stock market between 1968 and 1987. They discovered that the 42-week EMA
is 0.1. This exponentially treated difference, 1.00 X 0.1, is then added to gave the best performance, offering an equity gain of97+ points, but lagged
the previous week's EMA, and the calculation is repeated each succeeding behind the 45-week simple MA, which experienced a gain of 111 + points.
week. In the example, the exponentially treated difference for January 8 is Chart 9-13 features a 55-week EMA for Albertson's.
0.1, which is added to the previous week's average, 99.00, to obtain an EMA
for January 8 of99.10. This figure in column 6 is then plotted.

'Robert W.Colby and Thomas A. Meyers, The Encyclopedia of Technical Mariut Indicators, Dow
Jones-Irwin, Homewood, Ill., 1988.
Moving Averages 173

Table 9-4 EMA Calculation

EMAfor Difference Col. 3 x Col. 2 +
previous (col. 1 - . col. 4 col. 5
+ Price week col. 2) Exponent =/- EMA
Date (1) (2) (3) (4) (5) (6)

Jan. 1 99.00
8 100.00 99.00 1.00 0.1 +0.10 99.10
15 103.00 99.10 3.90 0.1 +0.39 99.49
.... 'l< 22 102.00 99.49 2.51 0.1 +0.25 99.74
99.00 99.64 (0.64) 0.1 -0.06 99.68
a> a> 29

a> <0 Table 9-5 Expontential Factors for Various Time Frames
a> en

Number of weeks Exponent

5 0.4
10 0.2
15 0.13
20 0.1
40 0.05
80 0.25

Chart 9-13 Albertson's simultaneous price pattern completion and MA crossover. One of
my favorite averages is the 65-week EMA. As featured in this chart, we see an H&S top for
Albertson's at the end of 1999. Note how the EMA and neckline both intersect in approxi-
mately the same area. This type of action emphasizes the significance of the break. /from

60 '" - ' ,


30 Joint break after

several Whipsaws


"'00 1000


174 Part I: Trend-Determining Techniques Moving Averages

Envelopes Chart 9-14 Masco 1998-2001 using a ± 15% band of a lO-<lay simple MA.

It has already been established that MAs can act as importantjuncture points
in their roles as support and resistance areas. In this respect, the longer the 36 : ..__ ••.•••••• : '\~ ,+15% band
time span, the greater the significance of the average. This support and resis-
tance principle can be taken one step further by constructing symmetrical 30
lines parallel to an MA called envelopes (refer to Fig. 9-2). This technique is ............ -
based on the principle that stock prices fluctuate around a given trend in
cyclical movements of reasonably similar proportion. In other words, just 25
, ······t·····....... .
as the MA serves as an importantjuncture point, so do certain lines drawn
parallel to that MA. Looked at in this way, the MA is really the center of the -15% band
trend, and the envelope consists of the points of maximum and minimum 20
divergence from it.
There is no hard and fast rule about the exact position at which the enve-
lope should be drawn, since that can be discovered only on a trial-and-error
basis with regard to the volatility of the price being monitored and the time 15
span of the MA. This process can be expanded, as in Fig. 9-3, to include four
or more envelopes (that is, two above and two below the MA), each drawn
at an identical proportional distance from its predecessor. In this example,
the envelopes have been plotted at 10 percent intervals. If the MA is at 100,
for example, the envelopes should be plotted at 90,110, and so on. An exam-
ple featuring Masco appears in Chart 9-14. an MA, Bollinger bands are calculated as standard de.viations. above and
below an average based on closing prices. They are designed With the con-
cept that the bands widen and narrow as the price trend becomes more or
Bollinger Bands less volatile.

A useful addition to envelope analysis is a new approach devised by John

Bollinger.' Rather than being plotted as fixed percentages above and below Rules for Interpretation .
The deviation setting for the Bollinger bands determines the distance
between the outer bands and the center one. Chart 9-15 shows an average
set to 20 and a deviation to 6; all the other charts in this sec.tio~ have b~en
plotted with a 20 X 2 parameter. It is fairly evident that th~s ~Ider ~etung
has no significance because its bands are never touched. Slmilarly, If they
are set at too narrow a basis, there will be far too many whipsaws.
There are several rules for interpreting the bands:

Figure 9-2 Single envelopes. Figure 9-3 Multiple envelopes. • When the bands narrow, there is a tendency for sharp pric~ changes ~o
follow. This, of course, is another way of saying that when ~nces trade In
a narrow range and lose volatility, demand and suppl~ are In a fine. state
'ohn Bollinger, Bollinger Capital Management, P.O. Box 3358, Manhattan Beach, CA90266 of balance. In this context, a narrowing of the band IS always ~ela~ve to
( . the recent past, and that's where Bollinger bands can help In Visually
176 Part I: Trend-Determining Techniques Moving Averages 177

Chart 9-15 Northern States and the Bollinger band (20 X 6). (From Chart 9-16 Oneok and narrowing Bollingerbands. (From

34 Large deviations
33 /"\ /'o\never get touched <,
32 ;I, / r-. ~
! ~ \ I \
2S \'/

showing the narrowing process. They also give us some indication ofwhen Chart 9-17 Northern States and Bollingerbands (20 x 2) (From
a breakout might materialize because they start to diverge once the price
begins to take off. Two examples are shown in Chart 9-16, where it is also 30.5 Bollinger 20 x 2
possible to construct some trendlines marking the breakout point. 30.0
• If the price exceeds a band, the trend is expected to continue. This is '1"'--
really another way ofsaying that if the price moves above the band, upside ~ ~
momentum is strong enough to support higher ultimate prices and vice 29.0
versa. After both breakouts in Chart 9-16, we see the price immediately 28.6
move outside the band. The crossover of the Bollinger band usually indi- 28.0
cates short-term exhaustion and it quickly pulls back again. However, this
isjust a process of pausing for breath un til the trend is then able to extend 27.6
again. By now, you will have noticed that the price often crosses the bands 27.0
several times before the trend reverses. The obvious question at this point
is: How do you know when the band has been crossed for the last time?
In other words, how do you know how to spot the bottom and top of a 26.0
move? The answer lies in the following rule.
• 'When the price traces out a reversal formation after it has crossed out-
side a band, expect a trend reversal. Some examples are shown in Chart
9-17, first in April and then inJune. In both situations, the price tried and
failed to cross the band successfully before completing a price pattern.

Sw mmu y
• One of th e basic assumptions of technical analysis is that stocks move in
trends. Since major tre nds comprise many m ino r fluctu ations in pri ces,
an MA is constructed to help smooth out th e data 50 that the un derlying
trend will be more clearly visible.
• Ideally, a simpl e MA should be plotted at the ha1fwily point of the time
period being monitored (a process known as antning). but since this
would involve a time lag during which pri ces co uld chan ge rapidly and
lose m uch of the potential profit of a move, th e MA is plotted a t th e end
of the period in question.
• Th is drawback has been larg ely overcome by th e we of MA crossovers,
which provide warnings of a reversal in trend and by the use of weighted
E.\iAs. which arc npecial ly se nsitive to ch anges in th e prevailing trend
since they ~ight data in favor of the most recent periods.
• There is no such thing as a perfect ave~. The choice of time span always
represents a trade-offberwecn timeliness (ca tching the trend at an early The methods of trend de termination co nsidere d so far have been con-
stage) and sensitivity (ca tch in g th e trend tum too early and cawing an cerned with an alysis of th e movem en t of th e price itself through tre ndlines ,
un d ue amount of whipsaws) . For short-term tre nds. 10-. 25-, 30-, an d ~ price patterns, and moving averages (MAs) . Th ese techn iques are extremely
d<ly spans are suggested. bu t for longer-term time spans, 4O-week simple useful. but they identify a change in trend aftd it has taken piau.
and 65-week EMA averages aft recommended. Helpful tim e spans using Th is cha pter will exami ne th e general principles of momentum inter-
monthly data an: 6, 9,12, 18, and 24 months. pretation that appl y in som e degree o r o ther to all mom entum ind icators .
The role 01 dumgt (ROC ) will be used as a case study. Th e next two chapters
will discuss other specific momentum indicators. .'

The co nce pt of upside mom entum is illustrated in the following example.
When a ball is th rown in to the air. it begin s its lr.ljectory at a very fast pace;
that is, it possoses strong mom entum. Th e speed at which th e ball rises grad-
ually diminishes un til it finally com es to a temporary standstill. Th e for ce
of gravity then causes it 10 reverse co urse. Th is slowing-down pr ocess,
knOWn as a lou 01 u.pward mommhnn. is a phenomenon that is 01150

180 181

experienced in financial markets. The flight of a ball can be equated to a in Olapter 24. Pritt m om entum can be co nstructed for ilfly pri ce series, but
market pri ce. The pric e's rate of advance begins to slow down noti ceably breadth momentum can only be calculated for a series that can be broken
before the ultimate peak in pri ces is reached. down into various components. .
On the other hand, if a ball is thrown inside a room and hits the ceiling Th is chapter outlines a number of th ese basic principl es using th e ~~
whil e its momentum is , till rising. th e ball and the momentum will ~ as an example. Remember that it is o nly one type of pri ce momentu~ andi-
at th e sam e time. Unfo rtunately. mom entum indicators in th e marketplace cater, Chapters 11. 12, 23. and 24 will discuss other oscillators for pnce an d
ar e not dissimilar. This is because there are occasions for which mom entum breadth momentum.
and pri ce peak sim ulta neously, eithe r because a ce iling of selling resistance It should be noted that the type of tre nd reversal si gnal~d by a momen-
is mel or because buying power is temporarily exha usted . Un de r such con- tum indicator depends upon th e time span over which it has been calcu-
ditions. th e ~l of mom entum is ofte n as helpful as its dim:titm in assessing lated . It is accepted practice to use daily data for ide ntifying sh ort:t~rm
m e quali ty of a price trend. trends. weekly d ata for intermediate trench. and monthly data for pnrnary
Th e idea of downward mo mentu m may be better understood by com- trends .
pari ng it to a car that is pushed over the top of a hill. Th e car begi ns to roll It is very important to note that the use of mom entum indicators assun:s es
downhill and. due to th e gradient of me hill, th e car starts to accelerate until that th e price of th e 5«urity is e xpe rie ncing a normal cyclic rhyth~, which
it reach es maxim um velocity at th e bottom. Although its speed th en begins is expressed in pri ce action by rallies and reactio~s. H~r. to so me
to decrease. the car continues to travel, but final ly it comes to a halt. Mark~t instances, countercyclical reactions are almost nonexistent, Price movement
pri ces act in a similar fashion: Th e rate of dechne (01'" loss of mom entum ) is then reflected as a lin ear uptrend o r downtrend. This is an unusual ph~
ofte n slows ah ead of th e final low. This is not alwa)'! th e case, however, since ncmen cn, and whm it develo ps momentum, oscillato rs fail to work.. ThIS
momentum and price some times (as at peaks) tum together, as pri ces meet is why the following principle is so very important,
a major level of suppo rt. Nevertheless, mom entum leach price ofte n enough
to WOlf11 of a po tential trend reversal in the indi cator 0 1'" market average that
is being monitored.
Mom entum is a gene ric term. Just as fruit describes apples, o ranges,
grapes, and so on , so momentum embrac es ma ny different indicators,
Exampl es include ROC, the rela tive strength indi cator (RSI). the moving
aver.tge co nverge nce divergence (MACD). the know sure thing (KSf) ,
breadth oscillators. and diffusion indexes.
There are two broad ways of looking at mom entum. The first uses price
data for an individual series, such as a currency, commodity, stock, or mar- The ROC Indicator
ket average. It is th en man ipulated in a statistical form th at is plotted as an The simplest way of measuring momentum is to calculate th e ra te at which
oscillator. We will call th is pria mommlum (alth o ugh volume can be manlp- a &tturity pri ce changes over a given period of time. which is kno~ as an
ulated in th e same way). Th e second is also plotted as an oscillato r, but is ROC indicator. lfit is desired, for e xam ple. to construct an ROC usmg a 10-
based o n statistical manipula tion of a number of market components, such w~d. time span, the current pri ce is divided by th e pri ce 10 weeks ago. If
as th e percentage of New York Stock Exchange (NYSE) stocks above a 54>- the la test price is 965 and that 10 Wttks ago was 985. th e ROC o r momen-
week MA This measure is referre d to as lwradUt mmnt'I'llum and is dUcussed turn indicator will read 98.0. that is, 965 divided by 985. The subsequent
~adinB in the indicator will be calculated by dividing next week's price by
the price 9 wee ks ago (see Table lo-} ). The res ult is a series that oscillates
aroun d a central reference point, This horiz ontal equilibrium line ~PTe­
lents the level at which th e price is un changed from its reading l~ weeks
ago (I«: Fig. }o-} ). IT an ROC calculation were made fOT a pn~e that
~n~d un ch anged , th e oscillato r would be re presented. by a honzontal
Itraight line .
182 Part I: Trend-Determining Techniques Momentum Principles 183

Table 10-1 Ten-Week ROC Calculation When an ROC is above the reference line, the market price that it is mea-
suring is higher than its level 10 weeks ago. If the ROC is also rising, the
DijA 10 weeks 100weekrate of change
Date DJIA (1) ago (2) (col. 1 divided col. 2) (3)
difference between the current reading of the price and its level 10 weeks
ago is growing. If it is above the central line but declining, the price is still
Jan. 1 985
8 980 above its level 10 weeks ago, but the difference between the two readings is
15 972 shrinking. When the ROC is below its central line and falling, the price is
22 975 below its level 10 weeks ago, and the difference between the two is growing.
29 965
Feb. 5 967 If the indicator is below its central line but rising, the price is still lower than
12 972 its level 10 weeks ago, but its rate of decline is slowing.
19 965 In short, a rising ROC implies expanding velocity, and a falling one a loss
26 974
Mar. 5 980 of momentum. Rising momentum should be interpreted as a bullish factor,
12 965 985 98.0 and declining momentum as a bearish one.
19 960 980 98.0 The ROC may also be calculated by the subtraction method; that is, the
26 950 972 97.7
Apr. 2 960 975 98.5
current price is subtracted rather than divided by a price Nperiods ago. This
9 965 965 100.0 indicator is called the momentum indicator on several charting packages,
16 970 967 100.3 but it is really a rate of change. My own preference is strongly for the divi-
23 974 972 100.2
30 980 965 101.6 sion calculation since it expresses proportionate moves in a similar manner
May 7 985 974 IOU to the ratio scale. This means that the indicator will not be distorted in
longer-term charts, as will an ROC calculated with the subtraction method.
There are two methods of scaling an ROC chart. Since the choice does
not affect the trend or level of the indicator, the method used is not impor-
tant, but a brief explanation is in order because the two alternatives can be
confusing. The first method is the one described previously and shown in
Fig. 10-1, where 100 becomes the central reference point. In the example,
100 (this week's observation) divided by 99 (the observation 10 weeks ago)
is plotted approximately as 101, 100 divided by 98 as 102, 100 divided by
102 as 98, and so on.
The alternative is to take the difference between the indicator and the
100 level and plot the result as a positive or negative number using a ref-
erence line of O. In this case, 101 is plotted as + 1, 102 as +2,98 as -2, and
so on (see Fig. 10-2).

Selection of Time Span
01 Choice of the correct time span is important. For longer-term trends, a
00 12-month or 52-week time span is generally the most reliable, although a
99 24- or IS-month period can also prove useful. For intermediate trends, a
98 9-month, 26-week (6-month), or 13-week (3-month) span works well. Price
97 mOvements of even shorter duration are often reflected by a 10-, 20-, 25-,
Or 3O-day span. Reliable short/intermediate movements are often reflected
Figure 10-1 ROC using percentage scaling. With a 45-day (9-week) and 65-day (l3-week) span.
184 Part l: Trend-Deterrninlng T~ 185

a.m to-I Long 's Drug$, 1996-1997, and thml ROCs. (From .)

25 long's DnIg$

Plice conllrms

, e.
-, • Those that d eal with oyerbougbt and oversold co n di tio ns, divergences.
-, and the like I will call these mommtum ~ lfyou study momen-
tum indicators or oscillators, you will find that they have certain charac-
FIgure 10-2 ROC using pbs aod mms 1CaIing. teri5tics that are associated with subsurface strengths or weaknesses in the
underlying price trend. It is rather like loo king under the hood of an
engine. Most of the time you can identify mechanical tro uble before it
becomes self-evident. Momentum and sentiment are close ly allied, and
In this way. trendlines, price patterns, and divergences, which may no t be the relationship between them is discussed in Chapter 26, deali ng with
obvious in one period. are more apparent in an other. The diKovery of signs mark~t sentime nt.
of a tre nd reversal in several indi cators cons truc ted from different time • The idrotification o f trend ~ in the momentum indicator itse lf
spans adds funh er Iuel to the weigbt of the eviden ce. An exa mple ar this is (momentum ~d-rnersal tKhniques) t n this case. we are ma king the
featured in Ch an 10-), assumption that whe n a trend in momentum is reversed , prices will
sooner or later follow.

Principles and Applications of Trend-determining techniques, such as trendline violation s, MA cross-

Momentum Indicators oven . and so on, when applied to momentum are just as valid as when uti-
lized with pri ce: The difference, and it is an important one, is that a tre nd
The following description of the principles and usc of momentum indica- reversal in momentum is j ust that, a reversal in momentum. Momentum
tors applies to all forms of oscillators. whether construc ted from an indi- typically reverses along with pri ce, ofte n with a smatl lag, but }wt btcOUSt
vidual price series or from an indicator that meas ures internal market oscillators changt dirntion dotm't always mean thal prica will 100. Normally, a
momentum, such as those described in Chapter 24. tn-erul in the momentum trend acts as confirming evidence ofa price trend
These principles ca n be roughly divided into two broad categories: reversal signal. In effect. this momentum signal performs the act of
186 Part 1: Trend-Determining Techniques

supplementary "witness" in our weight of the evidence approach. I will have

more to say on this one a little later, but for now take special note of the
I Momentum Principles

willmove 10 percent over a lO-day period; yet over the course of a primary
bull market extending over a 12-month period, a 25-percent increase would

fact that actual buy and sellsignals can only come from a reuersal in trend of the not be uncommon. Some indicators, such as RSI and stochastics, have been
actual price, not the momentum series. specially constructed to move within definite predetermined boundaries.
When a price reaches an overbought or oversold extreme, the probabil-
ities favor but by no means guarantee a reversal. An overbought reading is
Interpreting Momentum a time to be thinking about selling, and an oversold one warns that the cur-
Characteristics rent technical position may warrant a purchase. In many cases when a price
reaches an overbought extreme, the news is good, participants are opti-
Overbought and Oversold Levels mistic, and human nature tells us to buy. Unfortunately, the opposite is more
Perhap~ the most widely used method of momentum interpretation is the likelyto be the case. On the other hand, an oversold reading is usually asso-
evaluation of o~erboughtand oversold levels. This concept can be compared ciated with a negative news background. The last thing we want to do is raise
to a person takmg an unruly dog for a walk. The leash is continually being our shaking hand, pick up the phone, call our friendly broker, and tell him
pulled from one side of the person to the other as the dog struggles to get or her to buy, but that is often a reasonable time to do it, provided the over-
free. Despite all its activity, however, the dog can move no farther away than all technical position is favorable.
the length of the leash. In view of the variability of indicators such as the ROC, there is no hard
The same principle holds true for momentum indicators in the market- and fast rule about where the overbought and oversold lines should be drawn.
place, except that the market's "leash" should be thought of as made of rub- This can be determined only by studying the history and characteristics of
ber, so that it is possible for particularly strong or weak price trends to extend the security being monitored. The lines should be drawn such that they will
beyond the normal limits known as OVerbought and oversold levels. These areas act as pivotal points that, when touched or slightly exceeded, are followed by
are dra~ o?, a chart at some distance above and below the equilibrium a reversal in the oscillator. When a particularly sharp price movement takes
level, ~ m ~lg. 10-3'.The actual boundaries will depend on the volatility of place, these boundaries will become totally ineffective. Unfortunately, this is
the pnce being monitored and the time period over which the momentum a fact of life, but by and large it is usually possible to construct overbought
indicator has been constructed. and oversold benchmarks that are price-sensitive. Again, the market leash is
For example, an ROC indicator has a tendency to move to wider extremes made of rubber and can remain in overbought or oversold territory for long
over a longer period than over a shorter one. It is highly unlikely that a price periods. Consequently, it is essential to get confirmation from a reversal in
the trend of the price itself before taking any drastic action.

Oscillator Characteristics in Primary Bull and Bear Markets

I mentioned much earlier that the character of an oscillator alters accord-
ing to the price environment. In a bull market, oscillators tend to move into
an overbought condition very quickly and stay there a long time. In a bear
market, they can and do remain in an oversold condition for considerable
Overbought In effect, an oscillator is not unlike a migrating bird in the Northern
.-L\----n-----r'\-----;:/\::-:/- Hemisphere. I have divided the price action in Fig. 10-4 into a bear mar-
ket,followed by a bull, and finally another bear market. As we enter the bear
----------------~~- Equilibnum
phase, the true range of the oscillator shifts to the south, in a similar way to
a bird in the Northern Hemisphere migrating south to escape the cold
northerly winter. Then, when the bull market starts, the oscillator's trading
Figure 10-3 Overbought and oversold zones.
pattern migrates north again, just like the bird, finally shifting south again
188 Part I: Trend-De1e:nnln1ng T~ 189

digested, most people realize that thi ngs really have no t changed at all, and
the price declines again. Th us, the overbought reading more often than not
will correspond with the top of a bear market rally.
Looking at it from another perspecuve. during a bull market the price
will be far less sensitive to an overbought condition. Often it will be followed
by a small decline or even a trading range. as at poi nt A in Fig. I G-4. The
as a new bear market begins . Th is is useful inform atio n in itself because if rule then is do not count on a short-te rm overbough t cond ition to trigger
it is possib le to dT3Wpa rallel ho rizontal lin es like these against an oscilla- a big decline in a bull mar ket because the odds do no t favor it,
tor, it provides a valuable due as to whether the prevailing primary trend is Finally, people often point to an oversold condition and use that as their
bullish or bearish. ratio nale for a rally. Your favorite fina ncial columnist migh t say,· Analysts
Th e second point is thai if you have an idea of me direction of the pri- point out that the ma rket is d eep ly oversold an d a snap back rally is
mary trend. you can anticipate wh at might come from a specific overbought expected." Once again. it very mu ch depe nds on the enviro nment, In a
o r oversold readi ng . In a bull market, the price is extremely sensitive to an hull mar ket, yes. that is true, but the columnist is more lik.e1y to say that
oversold conditio n . That means th at when you arc lu cky enough to S~ one, "despit e a short-term oversold condi tio n , analysts are expecting lower
look aro und for some confinning signals that th e pri ce is about to rally, An prices beca use . • ." and then th e colum nist will go on to list a load of bear-
exam ple migh t be the violation of a down trendline and so on . The reason ish factors justifying his or he r pos ition. Remember, as a general ru le, the
for this sensitivity lies in the fact that the oversold reading very likely reflects media reflect the crowd, which is usually wro ng at turning points. and do
an extreme in short-term sentiment, Market participants are foc wing on die not make accurate forecasts, especially when qu oting "experts."
la test ba d ne ws and are usi ng that 35 an excuse to sell. Because th is is a bull In a bea r market, on the other hand, a mar ket or stock is far less semii-
market, they wou ld be better served by remem bering the positive lo ng-term tive to an oversold reading, often failing to signal a rally or possibly being
fundamentals that will soon emerge and by using this weakness as an oppor- followed by a trading noge. as at poin t B in Fig. IG-4.
tunity to buy. Th e matu rity of the trend, whether prim ary or inte rm edia te, often has
Th e same thi ng happens in reverse du rin g a bear market. Thade n are an effect on the limits that an oscillator might reac h. For example, when a
fOC\1Sed on bad news that sends the price down . Then some unexpectedly bull market has just begun. then: is a far greater tendency for an oscillator
good news hits the wires and the price rallies. However, when it is fully ~ move quickly into overbought territory and to remain at very high read-
~ngs for a con side rable pe riod of time. In such cases, the overbo ught read-
mgs tend to give premature warnings of declines. During the ea rly ph ases
of the bull cycle, when the market possesses strong mo ment um, reactions
to the oversold I~I are much more responsive to price reversals, a nd such

~adings therefore offer more rel iable signals. It is only when a bull trend
IS matu ring, or duri ng bear phases, that overbought levels can be relied
upon to signal that a is shcrtly to be aborted. The very fact that an indi-

~:~ ~o 2 years
calor is unable to remain at, or even to achieve . an overbought reading for
a long pe riod is itself a signal that the advance is losing momentum. Th e
opposite is true for a bear trend.

A "
C'\f"\/~ -
~. f\J\.L
Overbought and Oversold Crossovers

'"C V\M- B
In most cases, excellen t buy and sell alerts are generated when the momen-
tum indicator exceeds its extended overbought or oversold boundary and
ihen crosses back through the boundary on its way to zero. Figu re 10-5
"lemonstrat es this possibility. Thi s approach filte rs ou t many premature buy
190 Part I: Trend-Determining Techniques Momentum Principles 191

between buyers and sellers has unequivocally shifted in favor of buyers.

Think of a situation where a person uses all his or her strength to crash
through a locked door. It takes a tremendous aJ?lount of energy to achieve,
but once the door is finally shoved open, there is nothing to hold that per-
son back any longer. In the same way,a mega-overbought removes the price
from its bear market constraints, leaving it free to experience a new bull mar-
ket. An example is shown in Fig. }(4).
A mega-overbought is about the only instance when opening a long posi-
tion from an overbought condition can be justified. Even so, it can only be
rationalized by someone with a longer-term time horizon. This is due to the
fact that whenever an oscillator experiences a mega-overbought, higher
prices almost always follow after a short-term setback or consolidation has
taken place. A highly leveraged trader may not be able to withstand the
financial pressure or the countertrend move, whereas the long-term investor
Gill. In most instances, you will probably find that the immediate correc-
tion following the mega-overbought is sideways, rather than downward, but
there are just enough exceptions to trip up the overleveraged trader.
Figure 10·5 Overboughtand oversoldrecrossovers. Since a mega-overbought condition is associated with the first rally in a
bull market, it is a good idea to check and see ifvolume is also expanding.
If it takes the form of record volume for that particular security, the signal
and sell signals generated as the indicator just reaches its overextended is far louder, because record volume coming after a major decline is usu-
boundary, but one should still wait for a trend reversal in the price itself ally a reliable signal of a new bull market. Expanding volume is a more or
before taking action. less necessary condition since it is consistent with the idea that buyers now
have the upper hand and that the psychology has totally reversed.

Mega-Overboughts and Oversolds

As discussed in Chapter 26, there is a close connection between market sen-
timent indicators and the characteristics of oscillators. Since market senti- Primary bear market
lasts 1 to 2 years
ment differs widely during bull and bear markets, it follows that such mood
swings are occasionally reflected in changing characteristics of momentum
indicators. Observing these changing characteristics through short-term
oscillators is one of the few techniques that enables us to identify primary
trend reversals at an early stage..
As discussed in previous chapters, I have termed some of these phenom- ,
ena mega-overboughts and mega-oversolds. A mega-overbought is the initial

thrust in a bull market following the final bear market low. It is a reading
in the momentum indicator well beyond the normal overbought condition
witnessed in either the previous bull or bear market and should represent
a multiyear high. Such conditions are usually a sign of a very young and
vibrant bull market. The very fact that an oscillator is able to rally to such
a high level can be used, along with other trend-reversal evidence, to signal
that a new bull market has begun. It represents a sign that the balance

Figure 10-6 Mega-overbought.

192 19 3

Having said th at. th en: are occasions when a mega-overbought con d ition 0:Iut 10-2 8edIoo DIc:kaBon, 1981-1 985 , end e rnega-o.oe:rbou IIIId C/IIIlI!:I'IOld. (From
is followed n ot by a reversal, but by a chan ge in trend. In o ther words. the _
previ ous bear market emerges into a multiyear trading range rather than it
full-fledged bull market. The point here is that th e low preceding the mega_
Rising peaks and troogl1s
overbough t is n ot normally d ecisively violated for many yean.

8~ Sigoaleclllere
The same concept also appean in reverse fo r oversold extremes.
Consequently, wh e n it price d ecl in e following a bull market h igh pushes a
momentum indicator to a multiyear extreme low, ~U beyo nd an ythin g wit- 5
nessed eithe r d uring th e previous b ull or bear market, th e implicatio n is iliat
sellers now have th e up pe r hand. The fact th at it is po ssible for the momen-
tum indicator to fall so sh arply and so de eply is in itself a sign th at th e char- 20 Meg8-oWrbought --+
acter of the mark et has changed. when you sec mis type of action, you 10 ~=""';h~-.I...,...4.--+ ~-t--+-........t+1f-7--+'11
should , at the very least. q uestion th e bull market scenario. Look for telJ..
tale signs that a n ew bear market may be underway. What are th e volume o
co nfigu rations on the subseque nt nlly? Does volume now tre nd lower as the
p rice rises compand to previous rallies th at were associated with trends of -10 .JI--~+-'Il---1- --'--'----\1L.!-'--'---'--'---- --- -l
rising volu me?
Th e same possib ilities of a change, as o pposed to a reversal in trend, also 10 , '"_ 1981
• _
apply in th e sense that a mega-overso ld is typically th e fint d ecline in a bear
market, bu t occasionally it can also signal a change in trend from a p rimary
bull market to a multiyear trading range. An example of a mega-oversold ts
shown in Fig. 1{)'7. Both mega-conditions are wually best observed in short-
term oscillators rangi ng fro m 10 to as many as 30 days in a tim e span. Th ey Extreme Swings
n ever develop from ind icators whose co nstructio n co ns trai ns th eir Iluctua- Th e extreme swing is an other phenomenon that sign als a dramatic sh ift in
tiona between 0 and 100, such as the RSI and stoch astic indi cators. Examples psychology. It reflects the idea that some primary tre nd reversals are signaled
ofa mega-overbo ught and a mega-oversold are shown in Ch art 1{).2. by a swing from unbelievable exubera n ce, as th e bull market re ach es its
peak, to one ofco mplete despondency and depression, as the first bear mar-
kt setback gets under way. The opposite is true of a tnnsition fro m a pri-
mary bear to a primary bull market, In o rder for an extreme swing to
develop, it is necessary for th e market to experience a p ro longed up- o r
downtrend. The extreme swing th en appean in a momen tu m ind icato r by
an extre me ly strong move in the direction of th e then-prevailing trend, as
shown in Fig. 10-8. This is followed by an extre me reading in the o p po site
directio n.
In Fig. Io-B, we see a exhaustion blowo fI to the bull move as the oscilla-
tor reaches a very overboug h t reading. This is subsequently followed by a
Price decline th at p ush es it to the o th er ex treme . Such actio n indicates a
dramatic sh ift in sentiment as market particip ants change from a mood of
euphoria to on e o f desponden cy as th e security eventually rea cts in the direc-
tion opposite to that o rigin ally expected.
In order to qualify for an extreme swing, the fint swing must represent
Flgww 10-7 ~. Ihe strongest move in several yean, certai nly the strongest since the initial
194 Part I: Trend-Determining Techniques Momentum Principles 195

It's nice to see a

confirmation. too!! quickly followed by

Figure 10-8 Bearish extreme swing.

a mUltiyear oversold... ---+
I Figure 10-9 Bullish extreme swing.

Needless to say, extreme swings are quite unusual, but when you can spot
thrust from the previous bear market bottom. It is really a climax move for them, it really pays to follow their lead since a new trend invariably results.
the bull market. The second swing to the downside should really be a mega- An example of a bullish extreme swing is shown in Chart 10-3 featuring VF
oversold, though in some cases an extreme oversold will suffice. Corp.
This phenomenon is undoubtedly formed because the first swing encour-
ages the participants who have been right about the prevailing trend and
discourages those who have been wrong. In the case of a bull market, the Divergences
final rally also squeezes out all of the remaining shorts, so when the trend The ball example used at the beginning of the chapter showed that maxi-
reverses, there is virtually no buying activity from speculators covering short mum velocity was obtained fairly close to the point at which the ball leaves
positions. The preceding sharp advance also encourages buyers who could the hand. Similarly, prices in financial markets usually reach their maximum
see that there was only one way prices could go, and that was up. As a result, level of momentum ahead of the final peak in prices. In Fig. 10-10, this is
decisions on the buy side are made carelessly and without thought for the shown at point A. If the price makes a new high, which is confirmed by the
fact that prices may move the other way. When they do, such individuals are momentum index, no indication of technical weakness arises. On the other
flushed out of the market with no quarter given. Since there are few short hand, if momentum fails to confirm (point B), a negative divergence is set
sellers able to pick up the pieces, the price drops ferociously. up between the two series, and a warning of a weakening technical struc-
Extreme swings also develop between a bear and bull primary trend, as ture is given. Such discrepancies normally indicate that the price will
featured in Fig. 10-9. In this case though, the mood swing is from total undergo a corrective process. It can take the form of either a sideways or a
despondency and depression as the bear market squeezes out the last of the hOrizontal trading range, or (more likely) a downward one. However, the
bulls from the sharp and persistent downtrend. Even the strongest bulls are price will sometimes continue upward to a third top and be accompanied
forced to capitulate and eventually there is no one left to sell. Then during by even greater weakness in the momentum index (point C). Occasionally,
the rally phase, the shorts are forced to cover and new buying comes in the third peak in the momentum index may be higher than the second, but
because of the perceived improvement in the fundamentals. Since there is l?wer than the first. Either circumstance requires some degree of caution,
virtually no one left to sell, prices shoot up, and a mega- or extreme over- SInce this characteristic is a distinct warning of a sharp reversal in price or
bought is registered. a long corrective period.
196 Part l: T ~ T ~ 197

Owrt 10-3 \IF Corp ., 1989-1993, ard • posltiwIr extnJrne swing . (From _ .prlng.oom.1 Figu re l Q..10 aha shows a posilivt ~ In th is instance. th e pri ce
mak es its low a t poi nt E, but this is prec ed ed by th e oscillator that bo ttom ed
al a .
50 \IF COrporstion whenever any diverg en ce between mom en tum and price OCC U I1i , it is
essential to wait for a confirm atio n from me pri ce itself that its trend has
also been reversed. This co nfirm ation em be ach ieved by (1) the violatio n
or a simple trendli ne, as shown in Fig. 10-1 0. (2) th e CJ CISSClW'T of an MA, or
(5) th e comple tion or a price pattern o r (4) a reversal in peak/trough pro-
~ equals an extreme $Wi'lg gression. This fonn of insurance is well worth raking, since it is not unknown
for a securi ty to continually lose and regain mom entum with ou t sufferi ng
30 1~8y ROC
a break in tre nd d uri ng a long cyclical advan ce. Exam ples of th is phe nom-
enon occurred d uring me 1990's bull market in U.S. stocks a nd inJap an ese
10 }---;,-,.I,1rJ"....-+--f-.!>I.J!~~W-:Y>/""'"F'V-~ stocks between 1982 and 1990.
·'0 A good exam ple can be seen in Chan 104, which shows th e Nikke i Dow
.JO violating an irnportanr !P/ t-year seco ndary u endline after th e 13-we ek ROC
Multiyear 0'tIef50Id
indicator had neg;ttivd y d iverged several times with th e index. As a resu lt.
990 991 992 the final rally was accompanied by very little upsi de mom entum. It would
have been a mistake to sell o n an y of th e prior d ivergences, but a very timely
sd l signal w;u ge nera ted by waiting for a confirmatio n in th e fo rm of a tre nd
break in th e ind ex itself.
At point C in Fig. IO-Il , the price moves to a sign ificant new high, but
c the mom entum indicator is barely able to remai n ab ove the equilibrium
line. Such a situa tion demands th e utmost cautio n whm aaompanUd by a
tmul bR:al, because it is usual ly a sign of extreme technical weakness and is
often, th ough certai nly not always, followed by an above average declin e.
The opposite type of situation (see Fig. 10-12) in a bear market sho uld be
e ewed a.s a very positive charactera uc, especi ally if th e upward trend break.
in price is acco mpanied by high volume. Th e mor e expl osive th e volume ,
the more re liable th e sign al.
In a sense, it is possible to equa te momentum divergen ces and price trend
break! with dark clouds and ra in. If you look up a t the sky and observe dark
clouds, commo n sense tells you that it will probably rai n. bu t you do no t
Flgur_ 10-10 Momentum and divefgmees. know for sure until you can hold ou t your han d and actually feel rain falling .
In o the r words. the clo uds (lik.e th e divergen ces) warn of th e deteriora ting

198 Part I: Trwd-Deterrnnng T~ 199

Oiart 10-4 NiU<ei, 19~1990 . and neg.atl\oe dNergenoes. AJ. the 1990 peak, the NlkkeI.
expeiielad nume:roosdM!rgerw;:e$ with Il$ 13-weekROC. They __ c:onftrmecI \lA}en IIw price
eventually \Ilcll.!Iled tts 1986-90 up tren:IIlne .and lain on aossedbelow Its 65-week exponen-
tial movlog everage (EMA). These negallw diYergences took pece over a 3-yeu period so It
was not surprising hi the Nikkel experienceda mapr bear market lolIowIng thb event . (From


HOI<. -
.. ,
.. , .-.-....
.., ... ,

, 2
f'IsF- 10-12 Emenw buI1ish dtverg>ence.
': .j:,d.,.Po.r.~-:..I.~:--1I\1l--+:-'::'",,;:'--=:::\c,6,~\---J.\-1
~~ .. ... .. ... ...

weather (tech nical condition), but the change in weather is signaled only
by the tim raindrop (reversal in the price). It is possible to take the anal-
ogy a step further by concluding that the- darker the clouds (the greater the
numbe r of diverge nces ), me heavie r me rainstorm (the sharper the price

Price Disaepancy 01""""""",

A further indication of subtle strength or weakness is given when th e
momentu m series moves strongly in one direction. but the accompanying
move in the price index is a muc h smalle r one . Such a develop ment sug-
gests that the security is tired of moving in the direction of the prevailing
trend, because despite a strong push of energy from the oscillator. prices
an: Unable to respond. This unusual but powerful phenomenon is illustrated
fOr both tops and bo ttoms in Figs.. to-I ! and 10-14.

200 Part I: Trend-Determining Techniques Momentum Principles 201

One approach is to plot two momentum indicators of differing time spans

on the same chart, as shown in Fig. 10-15. Since this method tries to mon-
itor two separate cycles, it isjust as well to choose two widely different time
spans. For example, not much could be gained from the comparison of 12-
and 13-week ROGs since they would move very closely together. On the
other hand, comparing 13- and 26-week spans would dearly reflect differ-
ent cycles.
Most of the time, the two indicators are moving in gear, so this study does
not give us much information. On the other hand, when the longer-term
indicator reaches a new peak and the shorter one is at or close to the equi-
librium line, they are clearly in disagreement or out of gear (see point A 2
in Fig. 10-15). This normally, but not necessarily, indicates that a reversal in
trend will take place, and it is usually a significant one. Even so, it is very
important to make sure that any such divergence is confirmed by a rever-
sal in the price trend itself. In Fig. 10-15, a trend break does occur, but in
Figure 10-13 Bullish price discrepancy divergence. Fig. 10-16, no reversal takes place and the price continues upward.
Complex divergences also occur in a positive combination, as indicated
at point B} in Fig. 10-15, but again, it is mandatory to wait for that trend-
reversal signal in the price itself.
An example in Chart 10-5 features Lowe's using 20-and 45-<lay ROes. The
45-day series makes its bottom in January of 1995, but by that time the 20-
day series is already around zero. Soon after, the price confirms this diver-
gence by rallying above a small resistance trendline.

----- Long-term MA
(e·9·,40 weeks)

Figure 10-14 Bearish price discrepancy divergence. Negotive divergence

AI 0.
, I

Complex Divergences ", '.-
It is widely recognized that price movements are simultaneously influenced
by several cyclic phenomena. Because a single momentum indicator can
----- 26- week ROC 8 2

Positive divergence
monitor only one of these cycles, it is always a good idea to compare several
different momentum indicators based on differing time spans. Figure 10-15 Complex divergence.
202 Part I: Trend-Oetenninin Technkp.e 203

Momentum Trend-Reversal
Trend1ine Violations

- ....
- - - - - 1..on9- ..",MA
Occasiona lly. it is possible to co ns truct a trendline o n th e mom entum ind i-
cato r by co nnecting a series of peaks o r troughs. An example of a n uptrend
reversal is shown in Fig. 10.17. When the line is violated. a trend-reversal
1'"-9_.40 _ d
!ignal for the oscillator is generated .
The construction a nd signi ficance of the b reak should be based on th e
princip les outlined in Chapter 8. This type of momentum weakness m ust
be regarded as an ale rt, and action should be taken only when confirmed
by a break in the price trend itself (in dicated at point Ail in Fig. 16-17) . In
effect. the mo mentum trend break is reinforcing the price trend break and
offen an ad d itional piece of evidence th at th e trend has reversed.
An example sign aling a new u ptre nd is featured in Fig. I Q..18.1t is possi-
ble fOT the mome n tu m trend break to precede that of the price by some
time, }'1:t it docs not generally lose its potency because of this .
- l ' - _ ~ ROC
--- - - 26-_ ROC

Agure 1D-16 Complex dio.oerge:nao w1lh nonconftTmllltion.

0uu1 10-5 !..owe's, 1994-1995, and a complex d!Ylrrgerv;c. (From



- - 4 S-day ROC • • ••• 2O-<lay ROC


~ 111: _
2 04 205

flgure 10-19 Momentum price pettem completion .

Ch- rt 10-6 AIc:oe, 1999-2000, and momentum price pattnns. lhis charlfu.tura AkoB
lIIith II l Ckioy ROC. Note tKPN II r1!YI!ne H&S In the l ROC WIIS completed).S before
ltIl' price broke II nla! trendline to the upside. Adding 10 the signlfleance or the sigNll was the
flld thol the price crossed /I00ve lis MA more or Ies$ simullllneously WIth the trendllne viol,,·
lYJn. SB1ce both the line ard the MArepresent resistance, the brellkoul offered two pIeee$ of
"'o'ld\<f"lCe lhIII the trend had fe\II!f5Ied. 1.$ soon as the ROC had broken out, II begiln to fonn
Figure 10-18 BuIish rT1OmeI'1llnI trend break.. an H&S top. The break from this momentum distribution pattern _ then coofinnedby the
prieto violating II srnaI up tn!ndIine. (From WlIIUI.p

It sho uld also be noted that mom entum trendline breaks can be can-
finned by any legi timate trend-reversal technique in th e price. be it an MA
CTOSSO\'ff, a price pattern or peak-trough progressi on reve rsal, and so on.
." ~~~
..-_.' . ---- -----
or all th e m eth ods of mom en tum interpretation, th e j oint tre ndline viola-
tion techniq ue is one of the simplest and without doubt one of th e most " H

Momentum Price Patterns
Mome ntum indicators an: also capable of tracing out price patterns.
Because of th e shorter lead times normally associated with reversals of falling


mo mentum. a breako ut from an accumulat ion pattern, whe n accompanied
by a reversal in the downward trend of the pri ce itself, is usually a highly mo..
re liable indication that a wonhwhi le move has just begun. An exam ple is
shown in Fig. 10-19 an d Chart 10.6.
It is important to use a little com mon sense in in terpreting mom entuot
pri ce patterns. fi gure 10-20. for example, shows a breakout from a reverse
hmd-and-~ (H&S) patt ern th at takes place as a result of an overbcugb! Equilibrium Crossovers
condition. Th is is not to say that such signals will never be valid. but it stallds Some technicians have devised indicators tha t offer buyand sell signals when
to reason thai a breakout from an extreme level is w=ry unlikely 10 result in ~e momentum indicator crosses above an d bel ow iu equilibrium o r zero
a sustainable price move. Remember; t«hniall analpis bals willa probabibJits, line. Man y markets do not lend themselves to this approach. so iu imple-
and the odds of a favorable o utcome in this case are low. m ~ m.a tion depends very much on a trial and erro r basis through

206 Part I: Trend-Determining Techniques 207
Momentum Principles

Chart 10-7 The Economist Commodity Index, 1968-2001, and zero crossovers. (From

Mornenlum Index

100 -r-4;;;/--"''--+--+-""t---t-------~-----~"-T-

Oversold fine

Figure 10-20 Overbought momentum pattern completion.

experimentation. In any event, it is always a good idea to use this method in

conjunction with a reversal in the price itself. Chart 10-7 shows how zero 12-
month ROC crossovers, used in conjunction with 12-month MA crossovers,
have consistently given reliable buy signals for the Economist All Items
Commodity Index. The two sets of ellipses point out few whipsaw signals.

Momentum and MAs

By now, it is apparent that all the trend-determining techniques used for
price are also applicable to momentum. The interpretation of oscillators,
as described previously, depends to a considerable extent onjudgment. One S.II
Momen1um lOdex
method of reducing this subjectivity is to smooth the ROC index by using
an MA. Warnings of a probable trend reversal in the price being monitored
are offered by momentum MA crossovers, as indicated in Fig. 10-2l.
One of the problems associated with this approach is that the momen-
tum indicator is often much more jagged than the price index that it is try-
ing to measure, causing an unacceptable number of whipsaw signals. It is
possible to filter out some of these whipsaws by using a combination of two
MAs, as shown in Fig. 10-22. Buy and sell alerts are given when the shorter-
term :tvlA crosses above or below its longer-term counterpart. Fil"'~e 10·21 MA crossovers.
208 Part I: Trend-Determining Techniques Momentum Principles 209

+ Price


.,. t


pr i ce
t index

t t
- Smoothed
t---+---,f---\:---f-- momentum


Figure 10-22 MA crossovers smoothed. Figure 10-23 (a) Directional changes of smoothed momentum MAs. (b) Overbought/over-
sold recrossovers of smoothed momentum MAs.

This concept of momentum is explained in greater detail in the next

chapter, since smoothed momentum forms the basis of the trend deviation smoothing the MA itself with an additional calculation. Another possibility
and MACD indicators. is to construct an oscillator by combining the MAs of three or four ROes
and weighting them according to their time span. This possibility is discussed
at length in Chapter 12.
Smoothed Momentum Indicators Chart 10-8 shows the effectiveness of combining two smoothed ROC indi-
Another way of incorporating MAs into momentum studies is to smooth the cators. In this case, the smoothing is a IO-month weighted average of
momentum indicator by a long-term MA. The meaning oflong-term in this 11-and 14-month ROCs of the S&P Composite monthly closing prices. Since
case will depend on the type of trend being monitored. For example, a 20- this indicator has been found useful for market bottoms rather than tops,
to 3D-daytime span would be suitable for a short-term price movement, but the momentum curve is significant only when it falls below the zero refer-
a 6-, 9-, or 12-month smoothing, or one even longer, is more appropriate ence line and then rises. The arrows show that bull market signals between
for a primary trend. Warnings of a probable trend reversal in the price would 1960 and 2000 were particularly timely. The ellipse represents a signal where
be offered by a reversal in the smoothed momentum index itself, as shown the indicator was timely, although in this case it did not drop below zero. I
in Fig. 10-23 (a), or by a penetration of the MA through a designated over- have traced this indicator back to 1900 and found that very few whipsaws
bought or oversold level, as in Fig. 10-23 (b). The level of the dashed over- Were triggered. These could have been eliminated by waiting for a 12-month
bought and oversold barrier would be determined on a trial and error basis, },fA crossover as confirmation. Clearly, this is an excellent track record
with reference to a historical study of the relationship between the price and (derived from an approach discovered by E.S.C. Coppock).
the momentum curve. A further variation on the construction ofa smoothed momentum index
If the momentum curve is found to be unduly volatile, it is always possi- is to take the ROC of an MA of a price index itself. This method reverses
ble to smooth out fluctuations by calculating an even longer-term MA or by the process described previously, because instead of constructing an ROC
210 Part I: Trend-Determining Techniques

Chart 10-8 S&P Composite and the Coppock Indicator, 1960-2001. (From

1000 S&P Composite

Indicators I
In this and the next chapter, several specific momentum indicators will be
examined. It is recommended that you study them all and then choose three
or four in which you have confidence and with which you feel in~uitively
and then smoothing the resulting momentum index, the price index itself comfortable. Following too many indicators usually leads to confusion.
is first smoothed with an MA, and an ROC is taken of that smoothing.

Relative Strength Indicator

The Formula
• Momentum is a generic term embracing many different types of oscillators.
The relative strength indicator (RSI) was developed by Welles ~il~er.1 It is
• Momentum measures the rate at which prices rise or fall and gives use-
a momentum indicator, or oscillator, that measures the relative mternal
ful indications of latent strength or weakness in a price trend. This is
strength of a security against itself. This should not be confused with co~­
because prices usually rise at their fastest pace well ahead of their peak
parative relative strength, which compares the performance of one secunty
and normally decline at their greatest speed before their ultimate low.
to another. The formula for the RSI is as follows:
• Since markets generally spend more time in a rising than a falling phase,
the lead characteristic of momentum indicators is normally greater dur-
ing rallies than during reactions. RSI = 100 - ...1QQ....
1 + RS
• Oscillators reflect market sentiment and have different charactersistics in
primary bull and bear markets.
average of x days' up closes
• There are two basic methods of interpreting momentum: momentuIll RS=
characteristics and momentum trend reversal. average of x days' down closes
• Momentum signals should always be used in conjunction with a trend-
reversal signal by the actual price.
'Wells Wilder, New Concepts in Technical Trading Systems, Trend Research. Greensboro. NC,



. ,.
212 Part I: Trend-Determining Techniques Individual Momentum Indicators I 213

where RS = the average of x days' up closes divided by the average of x days' Chart 11-1 RSI versus the ROC. (From
down closes. The fonnula aims to overcome two problems involved in the
construction ofa momentum indicator: (1) erratic movements and (2) the 40 I Philadelphia Gold and
need for a constant trading band for comparison purposes. Erratic move-
45-day ROC
~ iii!"" Silver Share Index

ments are caused by sharp alterations in the values, which are dropped off 20 ; "~,~\ ~ ~ J~.~~
II,., ' I r .,1 ,I " ~
in the calculation. For example, in a 20-day rate of change (ROC) indica- ( ... r,; \1. '
tor, a sharp decline or advance 20 days in the past can cause sudden shifts
in the momentum line even if the current price is little changed. The RSI
attempts to smooth out such distortions.
The RSI formula not only provides this smoothing characteristic, but also
results in an indicator that fluctuates in a constant range between 0 and 100.
The default time span recommended by Wilder is 14 days, which he justi-
fied on the basis that it was half of the 28-day lunar cycle.

The RSI Enables Comparisons Between Securities

The nature of the RSI calculation enables the accurate comparison of dif-
ferent securities on the same chart. In Chart 11-1, there are two series, the
Dow jones Utilities Index and the Philadelphia Gold and Silver Share
Index. The upper panel plots a 45-day ROC and the lower one a 45-day RSI. in an opposite way for the RSI. For the RSI, equilibrium is the halfway point,
With the ROC, it is not easily possible to compare the two because the which in this case is the 50 level. It is therefore traditional to place the over-
Utilities Index is far less volatile. On the oilier hand, you can see that the bought and oversold lines equidistant from this point.
divergence in volatility is far less. Because of this, it is much easier to estab- We should remember that longer time spans in the RSI calculation result
lish universal standards for the overbought and oversold benchmarks. Using in shallower swings and vice versa. Consequently, the 70/30 combination is
the 14-day default, these levels are traditionally set at 30 for oversold and inappropriate when the time span differs appreciably in either direction
70 for overbought. from the standard 14-day period. Chart 11-2, for example, features a 9-day
RSI for the Eurodollar for which an 80/20 combination gives a much bet-
ter feel for the overbought/oversold extreme than "the 70/30 default value.
Constructing Overbought/Oversold Unes This is due to the fact that shorter time spans result in wider RSI oscillations.
The lower panel features a 65-day RSI in which the narrower swings result
The default time span for the calculation of an RSI is 14 periods. The over- in a more appropriate 65/35 combination. In this instance, neither of the
bought and oversold lines are typically drawn at 70 and 30, respectively. In default 70/30 values is reached at any time.
an article entitled "How The RSI Behaves," Peter W. Aan argued that the . The terms long and short time spans refer to the type of data under con-
average value of an RSI top and bottom occurred close to the 72 and 32 lev- SIderation in a relative sense. For example, a 60-day RSI would represent a
els, respectively. This research would indicate that the 70 and 30 levels reC- long span for daily data, but for monthly numbers, a 60-day (2-mont1l) span
ommended by Wilder should be moved farther apart to better reflect the
average overbought and oversold values.
It is important to note that the magnitude of the oscillations of the RSI
~s i~verse .to th~t of most other momentum series. For example, the ROC
indicator IS subject to wider fluctuations the longer the time span. It works

2FutU1~s,Janual)' 1985.
2 14 215"

Oiart 1 ) ·2 EurodoIar. 1994-1997, RSI ouerbought.!ownoIdllnes. (From iUUlUI.p1fng.mmJ Oaart 11·3 RSI OOlllplIrtng time fnunes. (From

-:n t-- rr+-+" o:-::.--I--\-:-Itt-/r--+-+- -"----'I+lfhl-- ]
~yRSl Owrboughtl owrwkl are at 80120
55 6O-day RS!

1995 996 1997 993

would be very sha n. Some considera tion sho uld therefore be given to this are many oilier cycles apart from the lunar cycle. Working on this a5$ump-
factor when the choice of a specific RSJ time span is being mad e. Chan tion, for example , would mean that a 14-hour RSI would be inappropriate
11-3 plots two RSls for id entical tim e periods (60 days or 3 m onths). if the dominant cycle was som ething other than 28 hours. The same would
How~r. the OYerbo ug h t and oversold lin es are d ra wn at different levels be true for weekly and monthly data.
because one calculation is based on dail y and the other on monthly data. In practice. a 14-day time span works quite well, but only for shorte r peri-
. Because RSIsbased o n shorte r-term time spans experien ce greater volatiJ.. ods. I also we 9-. 25-, 3o., and 45-day spans. For weekly data, the calendar
I~', they are more suitab le for poin ting o ut overbought and oversold con- quarters ope rate effectively, so 13-, 26-, 39-, and 52.week spa ns are adopted.
dl~ons . ?n the other hand . longer-tenn spans are more stable in their M for monthly charts , the same recommended spans for the RO C are also
traJec~ones and, therefore lend themse lves better to the purpose of con- suitable for the RSI, that is, 9, 12, 18, and 24 months.
struc tmg trendh nes and pri ce patterns. For longer-term charts, covering perhaps 2 yean of weekly data, a time
span of about 8 weeks o ffers enough information to iden tify interm edi ate-
tenn turning points. A 26-week RSJ results in a momentum series that oscil-
lates in a narro wer ra nge , but nevertheless usually lends itself to tre ndline
<orutrucrion. Very long term charts, goi ng back 10 to 20 years, see m to
respond well to a 12-month time spa n. Crossovers of the 30 pe rcent over-
sold and 70 percen t overbought barriers provide a very good indi cation of
~or long-term buying and selling points. 'When the RSI pushes th rough
lhese extremes and then crosses back toward the 50 level, it often warn s of
;I, reversal in the primary tre nd .

To i50late major buy ca ndidates, it is importa nt to rem ember that the be st

oPPOrtun ities lie where lon g-term momentum , such as a 12-month RSI, is
OYtnold. If you can also identify an interm ediate- and a short-term oversold

216 Part I: Trend-Determining Techniques Individual Momentum Indicators I 217

condition, all three trends, primary, intermediate-term, and short-term, are Chart 11-5 Spot Nickel, 1981-2001, RSI, trendlines, failure swings, and extreme crossovers.
then in a classic conjunction to give a high-probability buy signal. (From

1600 Spot ($) Nickel LME
RSI Interpretation
Extreme Readings and Failure Swings Any time an RSI moves above
its overbought zone or below its oversold zone, it indicates that the security
in question is ripe for a turn. The significance depends upon the time frame
under consideration. For example, an overbought reading in a 14-day RSI,
as shown in Chart 11-4, is nowhere near as significant as an RSI constructed
with a 12-month time span, as in Chart 11-5. An overbought or oversold read-
ing merely indicates that, in terms of probabilities, a counterreaction is over-
done or overdue. It presents an opportunity to consider liquidation or
acquisition, but not an actual buy or sell signal. This can come only when
the price series itself gives a trend-reversal signal.
For instance, in Chart 11-6 featuring Centura Banks, the overbought read-
ing in June 2000 was confirmed by a nice trend break in the price, as was
the oversold reading that had previously been registered in February.
Compare this to the December 1999 oversold reading that was not con-
firmed by a price break and did not generate any meaningful rally.
More often than not, the RSI traces out a divergence, as indicated in Fig.
11-1. In this case, the second crossover of the extreme level at points A and

Chart 11-6 Centura Banks, 1999-2000 RSI, trendlines, and price patterns. (From
Chart 11-4 German Bunds, 1995-1996, RSI and trendlines. (From

German Bund Futures 66
218 Part I: Trend-Determining Techniques Individual Momentum Indicators I 219

B usually offers good buy and sell alerts. These divergences are often called Chart 11-7 J.P. Morgan, 2000-2001, RSI, trendlines, and failure swings. This chart shows
failure swings. Chart 11-7 shows an example of a bullish failure swing for J.P. a 14-day RSI for J.P. Morgan. Note the two trendJine violations that signaled declines in
September 2000 and February 2001. Also, there was a bullish failure swing at the October
Morgan in October 2000. Chart 11-5 shows a longer-term failure swing for low. It's true that the price confirmed with a trendline violation, but the trendline was not that
Spot Nickel. impressive since it was quite steep and was only touched by one rally. Uttle wonder that the
stock needed to test the low again before taking off to the upside. (From

Trendline Violations and Pattern Completions The RSI can also be Failure swing confirmation
used in conjunction with trendline violations. Generally speaking, the
longer the time span for any particular period (daily, weekly, or monthly),
the better the opportunity for trendline construction. Important buy and
sell signals are generated when trendlines for both price and the RSI are
violated within a relatively short period (refer to Chart 11-4). An example
of the RSI's capability to form price patterns is shown in Chart 11-6, where
we see three situations in which it traced out a head-and-shoulders (H&S) for-
mation. Each one was confirmed by a trend break in the price. Note how
the inverse pattern, formed in the summer of 1999, involved several false
oversold crossover buy signals. However, it was not until the pattern was com-
pleted that the price confirmed any of this by rallying above its 3-month
down trendline.

Smoothing the RSI It is a perfectly legitimate technique to smooth the

RSI. One of my favorite approaches is to smooth a 9-day RSI with an 8-day
MA. Because the fluctuations are not as great as the raw data, the overbought
and oversold lines are drawn at 70 and 30, not at my usual default of 80/20
for a 9-day span. Chart 11-8 features such an indicator plotted inversely with the 3D-year
yield to correspond with movements in bond prices. This smoothing tech-
nique is very useful from the point of view of flagging reversals when the
indicator moves beyond its normal overbought and oversold extremes. The
dashed arrows point to changes in trends signaled by the RSI moving to an
extreme, but where no reasonable price confirmation was given. The solid
Price arrows show the three instances where the overbought and oversold
extremes were confirmed with price trend breaks. This form of interpreta-
tion works pretty well except in periods experiencing strong linear up- or

RSI Application to Peak-and-Trough Progression

The RSI often traces out a series of rising or falling peaks and troughs,
Which, when reversed, offer important buy or sell alerts. Chart 11-9 shows
that the 14-day RSI for SunTrust Banks experienced two peak-and-trough
reversals, each of which was confirmed by a price trend break. These are
Figure 11-1 RSJ failure swings. flagged by the arrows.
220 Part I: Trend-Determining Techniques Individual Momentum Indicators I 221

Chart 11-8 30-year Government Yield, 1998-200 I, RSI and overbought and oversold Two Variations on the RSI
crossovers. (From
Chande Momentum OscUIator
The Chande momentum oscillator (CMO), named after its inventor, Dr.
Tushar Chande, is a variation on the RSI, yet is uniquely different. It has
three characteristics:

• The calculations are based on data that have not been smoothed. This
means that extreme short-term movements are not hidden, so the indi-
cator reaches overbought and oversold extremes more often, but not
enough to result in too many signals.
• The scale is confined within the -100 to + 100 range. This means that
the zero level becomes the equilibrium point. With the RSI, the 50 level
is the equilibrium point and is not always readily identifiable. With zero
as the pivotal point, it is easier to see those periods when momentum is
positive, and those when it is negative. The zero equilibrium therefore
makes comparisons between different securities that much easier.
• The formula uses both up and down days in the calculation.

Interpretation ~
Chart 11-10 compares a 14-day RS1 with a 14-day CMO. The first thing to
notice is that the CMO reaches an overbought and oversold extreme more
Chart 11-9 SunTrust Banks' 1997 RSI and peak-and-trough analysis. (From times than the RSI, such as in February andJune 2000 and in January 2001. Sometimes it is possible to construct more timely and better trendlines for
the CMO, though occasionally it works the other way. The March 2000
break, for instance, came off as a better trendline for the CMO than the RSI;
so too did the breaks from trendlines AB and cD. The two trendlines in the
50 \f\lhl1\J'\ IJl 1h j I summer of 2000 were slightly better for the RSI. Note also that both series
experienced positive divergences at the October 2000 low, but the one for
the CMO was a much stronger signal since the September bottom was well
above that of mid-October. It does not always work in favor of the CMO, but
I prefer this indicator because of the more numerous overbought and over-
sold readings, and the plus and minus scaling, which makes it easier to spot
Positive and negative readings.
One approach that I have found helpful is to plot a 20-day CMO and
smooth it with a 10-day MA, such as that plotted in Chart 11-11. I then take
~ smoothing of this indicator, in this case a 10-day simple MA, the dashed
hue that hugs the CMO. MA crossovers are then used to generate buy and
sell alerts. However, since there are numerous crossovers, it is important to
May Illakean attempt at filtering out those that are not likely to work out by only
222 Part I: Trend-Determining Techniques Individual Momentum Indicators I 223 .

Chart 11-10 FTSE, 2000-2001, comparing the RSlwith the CMO. (From Chart 11-11 Hang Seng's 1998-2000 smoothed CMO.

Hang Seng


using those that develop at an extreme level in view of the fact that they tend
to be more accurate. This should then be confirmed by a trend break in span with a l(klay momentum factor. Since it is an RSI-based indicator,
the price itself. Some examples are shown in Chart 11-11. longer-term spans involve less volatility. Note that the fluctuations in the 45-
dayseries are much less pronounced than the 14-day RMI.
Generally speaking, the longer-term span offers slower and more delib-
Relative Momentum Index erate movements that lend themselves more easily to trendline construction.
The relative momentum index, or RMI as it is known, is another variation Severalexamples are shown in Chart 11-12. I particularly like the late 1998
on the RSI. In the calculation of the RMI, the standard RSI formula is mod- signal since it is confirmed by a simultaneous ,breakout above the trendline
ified to allow for a momentum factor. This indicator first came to my atten- and the 200-day MA. Whenever a price crosses above a trendline and a reli-
tion as an article in the February 1993 Stocks and Commodities magazine by able MA at the same time, it emphasizes the strength of the signal since they
Roger Altman. reinforce each other as dynamic resistance areas.
This modification has two effects. First, it smooths the indicator and, sec- accentuates the degree of the fluctuation. The result is a lessjagged
oscillator that experiences more overbought and oversold readings. The RSI Conclusion
RMI requires two parameters: the time frame and the momentum factor. Most of the time, the RSI and its two variations, like all oscillators, do
If the RMI has a momentum factor of 1, the indicator is identical to the not tell us very much. The RSI can be really useful when it triggers diver-
RSI. It is only when the momentum factor is greater than 1 that the twO gences, completes price patterns, or violates trendlines. When such an
series diverge. ?Ccurrence is also confirmed by a trend-reversal signal in the price itself, it
Chart 11-12 shows two variations on the RMI. The middle panel features IS Usually a wise policy to pay attention, because the RSI has a good record
a 14-day span with an 8-day momentum factor, and the lower one a 45-day of reliability.
224 Part I: Trend-Determlnlng T~

O&art 11 ·12 AT&T, 1996-200 1, two RMI varla tloos. tFtQm .-o.pnng .rom .)
50 AT&T ... .... M" · IO"1o
- Mni.. ~
--. -. .. ....- 0....-.:1 0.
: .. .."'''-10''"
100 RMI (1418)
JOtI. fe ll. Mor. Apr .


Agure 11 -2 Envdopes and I1'1OItlelltum .

a-t 11 -13 S&PCo.nposlla, 1996-1 997, calotlatlon ol .~IrdcaIof. tFrom
. ." ... ... _ .prl

800 S&P Composle

Trend Deviation (Price Oscillator) 750
A tren d-deviation indicator is obtai ned by dividing or subtracting a securi tr's
pri ce by a measure of trend, which is usually a Conn of MA. It is also possible
to base a trend d eviation using linear regression tec hniques. However, we wiD
concentrate on th e MA meth od here . This app roa ch is also called a priaOStil-
lasor in some c harting packages. Th ere are two methods of calculation: sub-
traction and d ivision. Division is preferred. since it is more reflective of
pro portion ate moves. For a discussion of this topic, you are referred to
Chapter 5 and Chapter 8. which co mpare logarith mic a nd arithmetic scales.
Since the ave rage re presents the trend being mon itored, this oscillalO r
ind icat es how fast the price is advancing or de clining in relation to that
tre nd . An oscillator based on a tre nd-deviatio n calculation is, in fact, a hor-
izo ntal represe ntation of the envelope anal ysis discussed in Chapter 9, bur
in th is graphic form at, it also shows sub tle changes of underlying technical
stre ngth and weakn ess.
Figures 11-2 (a) and (bl show these two approaches for th e same indica-
to r. The up~r and lower e nvelopes are bo th drawn at a level that is 10 per-
ce nt fro m the ac tual MA, which mea ns th at when the price touches the l ~ The int erp reta tion of a trend-deviation indicator is based on th e sam e
line, it is really at the same level as th e MA. When the momentum index lS P?nciples described in Cha pter 10. This method can be used to identify
at 110, th e price index is 10 pe rce nt above its MA. and so o n. Chart II -I! ~noerge:nces as well as overbo ught and oversold zones, bu t it appears to come
for the S&P Co mposite Index also indicates ho w a trend-deviation indica- Into its own when used in conjunctio n with trendline co nstruc tio n and MA
to r using the closing price and a 25-day MA is calculated. I crossovers.
226 Part I: Trend-Determining Techniques Individual Momentum Indicators I 227

Trendline Construction ing price divided by a 52-week MA. The second series is simply a 1Q-week
MA of the first. Buy and sell alerts are then triggered as the smoothed trend-
Chart 11-14 shows the Morgan Stanley Capital International (MSCI)
deviation indicator crosses above or below its 1Q-weekMA. Then look for a
Hungary Index together with a trend deviation calculated from a close
confirmation from the price itself. Two examples are shown in Chart 11-15,
divided by a 45-day MA. This is a fairly jagged indicator and lends itself to
one for a top and the other for a bottom.
overbought/oversold, trendline, and price pattern analysis. In March 2000,
This is very much a guerilla approach because the buy alert indicated by
we see the completion of an H&S top in the indicator, which was later con-
. the dashed arrow was signaled almost at the top of the rally. This example
finned by a trendline violation in the price. Later, the oscillator traces out
demonstrates the importance of picking and choosing between signals, only
a double bottom formation and the price violates a 3-month down trend-
selecting those that develop close to a turning point. If this filtering
line and crosses above its 25-day MA. This should have warranted a good
approach is not taken, then there is considerable risk that action will be
rally, but instead the price initially advanced and then traded in a sideways
taken close to the end of the trend.
range, later to be followed by a new low.
A useful method that greatly reduces such whipsaw activity, but still
In theory, this breakout should have worked. In retrospect, it was because
offers timely signals, is to advance the 52-week MA by 10 weeks when the
the price had begun a bear market, but this example is a good reminder
trend-deviation calculation is being made. This means that each weekly close
anyway that even the best laid technical principles can and do fail from time
is divided by the 52-week MA as it appeared 10 weeks before. This new
to time: all the more reason toplan an exit strategy priorto making a tradeorinvest-
calculation has been plotted in the center panel of Chart 11-16.
ment in case things do not go according to plan.
In this example, the whipsaw in late 2000 was filtered out since the trend-
deviation indicator fails to cross decisively below its MA. I am not suggest-
ing this as the only legitimate combination for weekly charts, but it is one
Trend Deviation and MAs
that appears to operate quite well. There is always a trade-off when you try
AIr alternative approach with trend-deviation indicators is to smooth out to make signals less sensitive and in this case we find that there is occasionally
unwanted volatility with the aid of two MAs, as shown in Chart 11-15. The a small delay compared to the nonadvanced 52-week MA. The most obvious
actual trend-deviation series is calculated by taking a 2frweek MA ofthe clos-

Chart 11-14 MSCI Hungary Index and a trend-deviation indicator, 1999-2000. (From Chart 11-15 S&P Airlines, 1995--2001, and a smoothed trend-deviation indicator. (From

800 MSCI Hungary Index
700 600 S&P Airlines

Trend deviation (1/45) H 0.1 Trend deviation (26/52)
228 Part I: Trend-Determining Techniques Individual Momentum Indicators I 229

Chart 11·16 S&P Airlines, 1995-2001, and two smoothed trend-deviation indicators. (From MACDs can be used with many different time periods. Gerald Appel of Signalert,~ who has done a considerable amount of research on the subject,
recommends that buy signals on a daily chart be constructed from a com-
S&P Airlines
bination of 8, 17, and 9 exponential MAs, but he feels that sell signals are
500 more reliable when triggered on the basis of a 12, 25, and 9 combination.
On the other hand, the popular MetaStock program plots the default val-
ues as 12 and 26 with the signal line at 9.
Chart 11-17 shows an example from General Electric with an MACD indi-
02 cator. As mentioned before, one of the principal ways in which the MACD
0.1 is interpreted is to use the signal-line crossovers as buy and sell alerts. The
0.0 ...p..;'<::-=.,.-::;:".4-''----------=o,~-_r_"---~---_r"-'-----__1 problem I have with this approach is that it results in far too many whip-
.0,1 saws.In Chart 11-17, we see numerous examples. Amore reliable technique
Trend Deviation (26/52) is to construct overbought and oversold lines, trendlines, and price patterns
and to look for divergences. In the chart, both series complete H&S pat-
terns at the end of the year 2000. The MACD also experiences a negative
divergence. Note how the divergence, flagged by the right shoulder, is barely
able to rally above zero. The result is an above-average decline. Note also
that the indicator was unable to rally above zero and touched its oversold
level several times during the balance of the period covered by the chart.
This type of action reflects bear market activity. The MACD is often plotted
one on this chart developed at the beginning of 1997, where the lagged in a histogram format along with the signal line as in Chart 11-18.
series in the center panel crossed its MA at a slightly higher price. In most
instances though, this is a small price to pay if a costly whipsaw can be
Chart 11-17 General Electric, 1999-2000, and MACD. (From
Moving Average Convergence 50
Divergence 46
The moving average convergence divergence (MACD) trading method is a 40
form of a trend-deviation indicator using two MAs, the shorter being sub-
tracted from the longer. The two MAs are usually calculated on an expo- ffiJ
nential basis, in which more recent periods are more heavily weighted than 2.0
in the case of a simple MA. It is normal for the MACD to then be smoothed
by a third exponential moving average (EMA) , which is plotted separately 0.5
on the chart. This average is known as the signal line; the crossovers ofwhich O.O+---~=---~-----"":""'~'-----A--------,~r----;
generate buy and sell signals. This indicator obtains its name from the fact .(l.6
that the two EMAs are continually converging and then diverging from each ·1.0
other. The MACD has gained great popularity over the years, but in effect,
it is really just another variation on a trend-deviation indicator that employs
two EMAs as its method of deviation. A visual of its construction is there-
fore very similar to Chart 11-13. 'Signalen, 150 Great Neck Road, Great Neck, NY11021.

230 231

Alternatively, the MACD can be plotted as a smoother series wing two rel- The stochastic th erefore attempts to measu re the poin ts in a risin g trend
atively long-term MAs. This more deliberate series then lends itself better at which th e closing prices tend to cluster around the lows for th e period
to signal-line crossovers. For example, on hourly charts, a 65/90 combina- in question, a nd vice versa, since these are the co nd itio ns that sign al tre nd
tion with a 12-pe riod signal lin e appean to work q uite well . The same prin- uversals. It is plotted as two lines. the %K line and the %D line. The %D
ciple can be applied to daily, weekly, o r monthly charts. Iro nically th e default provides th e major sign als an d is th erefore m ore Important,
time span used for daily charts (26/ 12 with a 9-<by signal line) appean 00 The fonnula for calculati on of %K is
work better on monthly ones because it manages to retain the primary trend
swings ~ l th e sign al line whipsaw crossovers are kept to a mi nimum. %K= loo[(C - 41 / (H, -41)

where C is th e mos t rec en t close, 4 is th e lowest low for th e last five trad-
S tochastic Indicators ing periods, and H 5 is the high est high for th e same five trading periods.
Introduction and Formula Remember th a t the calc ulation of stoc hastic indicators d iffers from that of
The stoc has tic ind icator o riginally gained a great deal of popularity among most o ther momentum in dicato rs in that it requires h igh . low, and closi ng
futures traders, with the result that the standard Cannula uses very sho n term data for the period in question .
time spans. The th eory be hind the indicator. which was invented by George Th e stochastic formula is similar to the RSI in tha t the plots can neve r
Lane: is that prices tend to close near th e u p per e nd of a trading range dur- exceed 0 or 100, but in thi s case , it measures the closing pri ce in relation
in g OlD uptrend. k th e trend matures , th e te ndency for prices to close 3WOIY to the total price ran~ for a selec ted number of pe riods. A very high read-
from th e higher e nd of the trading range becom es pronounced. In a down- ing. in excess of 80. would put th e closing price fo r th e period nea r the top
ward-moving mark et. the reverse co nd itions hold uue. of the range , wh ile a low reading. under 20. would place it ncar the bottom
of the ran ge.
The seco nd line , %D. is a smoothed version of th e %K line. The no rm al 1 1-18 Hornest4ke ~ and an MACD in histogum for mat . ThIs c:hort shows . ruue is three periods. The %D formula is as foll ows:
dame H&S pattern. Nol:e that the MACD histogram g~ became weaker as the Plttem
progressed . Thb was cri111 short-term sell signllI, but the price eo.oentuaIIy fen be10w the slgoaI
ieYeI. \From Telesam.I %D = 100 X (H,I4)
":----- -- -- - ----.c:--------, where H, is the three-period sum of ( C- 4 ) and 4 is th e three-period sum
of (Hs - 4) .
The mome ntum indicat or that results from th ese calcu latio ns is two lines
that fluctuate betwee n 0 and 100. The %K line is usually plotted as a solid
line, while the slower %D line is usuall y plotted as a dash ed line . A good
ro _ way to differentiate between th em is to think o f the fast %Kas "kwick" and
i-II '-' the slow %D as "dawdle."
",- 1" t.
The popularity of th e stochastic can no d oubt be explai ned by th e smooth
manner in which it moves fro m an overbo ugh t to an oversold co ndition ,
lUlling a trader into a Feeli ng tha t price trends ar e much mor e orderly th an
would appear from an obsen..a tion of an RSl o r an ROC ind icator.
longer-term time frames . use d o n monthly and weekly charts. ap~ar to
Work much better than the shorter-term stochastic ind icators used on daily
futures charts. Colby and Meyen in '1'1u Encydoptdia of T« hnical Marlen
11IdUatorr' no ted th at the stochastic tested " ery poorly relative to MA
crOssoven and o the r mom entum indicaton.

' Robe:n W. Colb)' and Thomu A. M ~-en. J1u CnCJd"!'ftlin r/ rrch'llullf Ma1l<tl!,,,Ii.caton. Dow
'I n''l:Mm~n l EdUoC...tnn Inc:urponotro. On Pb.inn .IL 61:1018. Jonr.- lnrin, Homewood, 14 1988.
232 Part I: Trend-Determining Techniques Individual Momentum Indicators I 233

Overbought and oversold bands are usually plotted in the 75 to 85 per- Divergence Failure An important indication of a possible change in
cent area on the upside and in the 15 to 25 percent area on the downside, trend arises when the %K line crosses the %D line, moves back to test its
depending on the time span in question. An overbought indication is given extreme level, and fails to cross the %D line, as in Fig. 11-4.
when the %D line crosses the extreme band, but an actual sell alert is not
indicated until the %Kline crosses below it, VVhen the two lines cross, they
behave very similarly to a dual MA system. If you wait for the penetration, Reverse Divergence Occasionally, during an uptrend, the %D line will
you can avoid getting trapped into shorting a strongly bullish move or buy- make a lower low, which is associated with a higher low in the price, as in
ing into an extremely negative one. Fig. 11-5. This is a bearish omen, and conventional wisdom suggests look-
ing for a selling opportunity on the next rally. This condition is sometimes
referred to as a bearsetup. A bull setup develops at the end ofa downtrend.
Crossovers Normally, the faster %Kline changes direction sooner than
the %D line. This means that the crossover will occur before the %D line
has reversed direction, as in Fig. 11-3(a). When the %D line reverses direc-
tion first, a slow, stable change of direction is indicated, and %D is regarded
as a more reliable signal [see Fig. 11-3(b)]. I

---" ./

Figure 11-4 Stochastic failures.

Left crossing

Right crossing

Figure 11-3 Stochastic crossovers. F'

Igure 11-5 Stochastic reverse divergences.
234 Part I: Trend-Determining Techniques Individual Momentum Indicators I 235

Extremes Occasionally, the %Kvalue reaches the extreme of 100 or O.

This indicates that a very powerful move is underway, since the price is con-
sistently closing near its high or low. If a successful test of this extreme occurs
following a pullback, it is usually regarded as an excellent entry point.

Hinges When either the %Kline or the %D line experiences a slowdown

in velocity, indicated by a flattening line, the indication is usually that a rever-
sal will take place in the next period (see Fig. 11-6).

Divergences The stochastic indicator often sets up positive and negative

divergences in a similar manner to other oscillators. Some possibilities are
indicated in Fig. 11-7. Buy and sell alerts are triggered when the %Kline
crosses %D after a divergence has taken place.

Slowed Stochastic Indicator It is also possible to extend the calcula-

tion in order to invoke a slowed version of stochastic indicators. In this
instance, the %K line is replaced with the %D line, and another MA is cal-
culated for the %D. Many technicians argue that this modified stochastic
version gives more accurate signals. Chart 11-19 shows the Amex Brokers
Figure 11-7 Stochastic divergences.
Index together with a lO-day %K slowed by a factor of 5. The %D line has
a slowing factor of 5 as well, hence the 10/5/5 label.

Chart 11-19 Amex Brokers Index, 2000-2001, stochastic and several interpretive tech-
niques. (From

Figure 11-6 Stochastic hinges.

236 Part I: Trend-Determining Techniques

• The RSI is bounded by 0 and 100. Overbought and oversold lines should
be wider the shorter the time span.
• The RSI lends itself more easily to comparing the momentum of differ-
ent.securities than does the ROC.
• The RSI can be used in conjunction with overbought and oversold lines,
divergences, price patterns, trendlines, and smoothings.
• Trend-deviation indicators are calculated by dividing the close or a short-
term MA by a longer-term one.
• Trend-deviation indicators can be used with trendlines, price patterns, and
MAs. They also lend themselves to overbought-oversold and divergence
analysis. \
• The MACD is a form of trend-deviation indicator.
• The stochastic indicator assumes that prices close near the low at the end
of a rally and near their highs at the end of a downtrend.
1 Indicators II
• The stochastic indicator is confined between 0 and 100 and consist of two
lines, the %K and the %D.
• The stochastic indicators lend themselves to crossovers, divergences,
hinges, extremes, and reverse divergences and are usually plotted in their
'slowed version. The Know Sure Thing (KST)*
The Long-Term KST
Chapter 10 explained that the rate of change (ROC) measures the speed
of an advance or decline over a specific time span and is calculated by divid-
ing the price in the current period by the price N periods ago. The longer
the time span under consideration, the greater the significance of the trend
being measured. Movements in a 10-day ROC are far less meaningful than
those calculated over a 12- or 24-month time span.
The use of an ROC indicator helps to explain some of the cyclical move-
ments in markets, often giving advanced warning of a reversal in the prevail-
ing trend, but a specific time frame used in an ROC calculation reflects only
one cycle. If that particular cycle is not operating, is dominated by another
one, or is dominated by a combination of cycles, it will be of little value.

~~Maj()tT~c~icalPrinciple .Price atanY?hetirneisd~termined by

';~\Welnt~racti(;mofrnanY different time cycl~:Ahindicator that takes
~S~s irltos()nsideration is likely to be more timely without losing too
';;: much inth~ way of sensitivity.

. • The design of this indicator was adapted from a technique used by Ian S. Notley, in my
\'lew the world's leading analyst of financial cyclic phenomena.

238 Part I: Trend-Determining Techniques Individual Momentum Indicators II 239

This point is illustrated in Chart 12-1, which shows three ROC indicators Chart 12-2 shows the Standard & Poor's Composite during the 1974-90
of different time spans: 6 months, 12 months, and 24 months. The 6-month period. The oscillator is a 24--month ROC smoothed with. a 9-mo~th ~.
ROC tends to reflect all of the intermediate moves, and the 24--month series This series certainly reflects all of the primary trend SWIngs dunng this
sets the scene for the major swings. The arrows flag the major turning points. period. However, if we use the indicator's changes in direction as signals,
They show that for the most part all three ROCs are moving in the same close examination shows that there is a lot to be desired. For example, the
direction once the new trend gets under way. A major exception occurred 1984 low is signaled with a peak in the oscillator. Similarly, the 1989 signal
at the 1984 bottom. Here we see the price rise, but immediately after, the develops almost at the rally peak. VI/hat was required was an indicator that
24--month ROC declines while the others continue on up. During the reflected the major trend yet was sensitive enough to reverse fairly close to
period covered by arrow A, the speed of the advance is curtailed because the turning points in the price. A good way of achieving this is to construct
of the conflict between the three cycles. Later, though, all three ROCs get an indicator that includes several ROCs of differing time spans. The func-
back in gear on the upside and the rally approximated by arrow B is much tion of longer time frames is to reflect the primary swings, while the inclu-
steeper. In effect, major turning points tend to occur when several cycles sion of the shorter ones helps to speed up the turning points. The formula
are in agreement and speedy advances and declines develop when more for the KST is shown in Table 12-1.
cycles are operating in the same direction. This is a fairly limited view
because there are far more than three cycles operating at anyone point
in time.
Table 12-1 Formulas for Time Frames
Clearly, one ROC time span taken on its own does not give us a complete
picture. This was one of the factors that I considered when designing the Time Frame Smoothing Weight
KST. Another was that I wanted an indicator that fairly closely reflected the 9-monlh 6--MA X 1
major price swings over the time period under consideration, primary 12-month 6--MA X 2
trends for monthly charts, short-term trends for daily charts, and so forth. I8-month 6--MA X 3
24-month 9-MA X 4

Chart 12-1 S&P Composite, 1978-1988, and three ROCs. (From
Chart 12-2 S&P Composite, 1973-1991. and a smoothed ROC. (From

200 S&P Composite 300

100 ._~- 200 S&P Composite



240 Part I: Trend-Determlning Techniques Individual Momentum Indicators II 241

Since the most important thing is for the indicator to reflect the primary Chart 12-4 S&P Composite, 1963-1979. and a long-term KST. (From
swings, the formula is weighted so that the longer, more dominant time
spans have a larger influence.
Chart 12-3 compares the performance of the smoothed 24-month ROC S&P Composite ~
to the long-term KST. It is fairly self-evident that the KST reflects all of the
major swings being experienced by the smoothed 24-month ROC. However,
the KST turning points develop sooner than those of the ROC. The verti-
cal arrows slice through the ROC as it bottoms out. In every instance, the


~ \J.
KST turns ahead of the arrow, the lead time varying with each particular
cycle. Note how in 1988 the KST turns well after the 1987 bottom, but just
at the time when the market begins to take off on the upside. The ROC 50
reverses direction much later. There is one period when the KST under-
performed, and that is contained within the 1986-1987 ellipse where the
KST gave a false signal of weakness, unlike the ROC that continued to rise.
The dominant time frame in the KST's construction is a 24-month period, -60
which is half of the 4-year business cycle. This means that the KST will work
best when the security in question is experiencing a primary up- and down-
trend based on the business cycle. For example, Chart 12-4 shows the KST
during the 1960s and 1970s where the S&P was in a clearly defined business
cycle-typetrading range. Periods of accumulation and distribution occur third when the MA also reverses direction. In most cases, the MA crossover
between the times when the KST and its MA change direction. offers the best combination of timely signals with a minimum of whipsaws.
There are really three levels of signaling. The first occurs when the indi- Changes in the direction of the 9-month MA offer the most reliable signals,
cator itself changes direction, the second when it crosses its MA, and the but these usually develop well after the turning point. The most timely and
reliable signals thereby develop in those situations where the MA reverses
close to a turning point.
Chart 12-3 S&P Composite, 1973-1991, and a smoothed ROC versus the KST. (From For the most part, the indicator has been very reliable, but like any other
technical approach, it is by no means a perfect technique. For instance, the
same calculation is shown in Chart 12-5, but this time for the Nikkei.
During periods of a secular or linear uptrend (as occurred for Japanese equi-
S&P Composite ties in the 1970s and 1980s), this type of approach is counterproductive since
100 many false bear signals are triggered. However, the vast majority of markets

Smoothed 24-month ROC
are sensitive to the business cvcle and so the summed ROC concept works
extremely well. It is for this reason that I call this indicator the KST, which
50 stands for knoio sure thing. Most of the time the indicator is reliable, but you
know that it's not a sure thing.

100 Short- and Intermediate-Term KSTs

The KST concept was originally derived for long-term trends, but the idea of
four smoothed summed ROCs can just as easily be applied to short, interme-
?iate, and even intraday trends. Formulas for various time frames are shown
In Table 12-2. The formulas presented here are by no means the last word and
242 Part I: Trend-Determining Techniques Individual Momentum Indicators II 243

Chart 12-5 The Nikkei, 1975-1992, and a long-term KST. (From Chart 12-6 Cash copper, 1993-2001, and an intermediate KST. (From

Nikkei 2600

. c_ -. i'J?_!~~~~_~~~lI~_~il:!~?J _
, ' , ~

Table 12-2 Suggested KST Formulas"

ROC MA Weight ROC MA Weight ROC
MA Weight ROC MA Weight an overbought crossover, a 65-week exponential moving average (EMA)
Short-termt 10 10 1 15 10 crossover, and a head-and-shoulders (H&S) top in the price-classic stuff.
2 20 10 3 30 15 4
Short-term! 3 3' 1 4 4' 2 6 6' 3 10 8' 4 There were a couple of false buy signals on the way down, but the rally peaks
Intennediate-tenn l 10 10 1 13 13 2 15 15 3 20 20 4
Intennediate-tennl 10 10' 1 13 13' 2 15 15'
in the KST lent themselves to a nice trendline construction. The violation
Long-terms 3 20 20' 4
9 6 1 12 6 2 18 of the line, the oversold crossover, and the completion of the base in the price
Long-terms 6 3 24 9 4
39 26' 1 52 26' 2 78 26' 3 104 39' 4 combined to offer a nice buy signal in late 1996. The next time the KST
*It is possible to program all KST fonnulas into MetaStock and the Computrac Snap crossed its oversold level in 1998 there was no good place to observe a trend-
Module (see Resources, at the end of the book). reversal signal in the price. That was not true in early 1999, where a positive
tBased on daily data.
IBased on weekly data. divergence, an EMA crossover by the KST, and a trendline break in the price
sBased on monthly data. offered a good timely entry point. Note that after the price broke above the
'EMA down trendline, it subsequently found support at the extended line.
Charts 12-7 and 12-8 feature a short-term KST suitable for use with daily
are su?"gested .mer~ly as good starting points for further analysis. Readers may data. First, the KST for the Bombay SE Index experienced nice fluctuations
expenmen~ WIth d~erent fonnulas for any of the time frames and may well between 2000 and the first quarter of 2001. The arrows on the chart indi-
come up WI~ supenor results. When experimenting, strive for consistency, cate the approximate periods when the KST reverses and crosses above or
never perfection, for ~ere is no such thing in technical analysis. below its 1O-dayMA. These MA crossovers are a classic way in which the KST
Chart .12-6 sh.owsan Intem:edi.ate KST for the London copper price. One is used. Unfortunately, not all situations are as usable as this.
of th.e thIll~ I ~Ike ~b.out the indicator, especially in its short-term and inter- Chart 12-8, for instance, shows the KST with the Arnex Brokers Index. After
~edlate ;anetIes, .IS its fl.exibility of interpretation. Pretty much all of the two false KST MA crossovers, the index finally succumbed to a trend break
mterprenve techmques discussed in Chapter 10 can be applied. In Chart 12- and a negative KST divergence in the spring of 1999. Occasionally, the KST
6, we see an overbought crossover at the tail end of 1994. It did not amount will move in a very quiet and subdued manner, as indicated by the two con-
t? anyt?ing bec.ause it was not possible to come up with any trend-reversal verging trendlines in the fall of 1999. Once the indicator breaks out and this
signals III the pnce. Later, though, in early 1995 we see a negative divergence, is confinned by the price, a strong and reliable signal is usually given.
244 Part I: Trend-Determining T~ Individual Momentum Indicators 11 245

Chart 12 -7 Bombay SE Index and a short-term KST. (From ww .) OI art 12-9 Apple Co mpute r and two intraday KSTs. (From WWW. pri n9.oom.)

Bombay SE


, .. 1 Short-term KST (15 b3.
o r~=~.,.-f.'-+-+-f~-;-..,¢-::'-':~C+----J

.., . ',

Long-le T (15 ba r)

( ", -( " . b
" " • No. .b , ,
C ha rt 12·8 Amex Brokers Index, 1998-2000, and II short-term KST. (From
www .) of th ese tim e frames will de pend on the characte r of the market bein g mo n-
.., itored and th e nature o f the trend . By its vcry const ru ction . th e Cann ula
assum es th at m arkets are in flue nced by the usual -t-yea r busin ess cycle .
roo Wheneve r that Lime frame is unduly sho rte ne d or lengthe ned , th e m onth ly
." KSTsu ffe rs in its per fo rm ance .

Using the KST with t h e Market C ycle Model

Three Main Trends Ch apter I expla in ed th at the re are seve ral tre nds
ope rati ng in the m arke t at a ny particul ar time. They range from intraday,
hourly trends righ t th ro ug h to \'er)' long term o r secu la r tren ds that evo lve
over a 20- o r 3O-year pe rio d . Fo r investme n t p u rpo ses, the most widely
'DO recogni zed tr en ds are sho rt te rm, inte rm ediate term, an d lo ng te rm . Sho rt-
..,': ~=~~-_---':~--/,;""",.."..,,;j, i-..:i;,0-7.~i:--\l terrn tr ends are usu ally monitored ....-i th daily pri ce s, In te rmediate-term
trends with wee ki)' p rices. and lo ng-te rm tre nds with monthly p rices. A

'" ." ••
hyPOIhe tical bell-shaped curve in corporat in g all three tren d s is shown in
Fig. I-I in Chapte r 1.
From an investme nt p oin t of view, it is importa nt to u nd ers tand th e d ire c-
tion of the m ain . or primary, tre nd. T h is makes it possible to g-ain so me pe r-
Cha n 12-9 shows a I5-minutc bar of Apple Compu ter with a short and spective o n the current pos itio n of the overa ll cycle. The co ns truc tion o f a
in termediate KST su itab le for in tra day tradin g. Several ove rso ld crossover long-tenn KST is a usefu l sta rt in g point from wh ich to id en tify m aj or ma r-
signa ls are featured. Generally speaking, the monthly }{Sf is far more reli- ket cyclejunctu res. The int roduction of short and in te rm ediate se ries as well
able than its da ily and weekly cou nterparts. The degree of success for any now enables us to repl icate the Market Cycle Mod el.
246 Part I, Trmd ·Determining Techniques Individual Momentum Indicators [[ 247

TIle ~Sl in vesa:nents are made wh en th e p rimary trend is in a risin g mode The best b uying opportu niti es see m to occur eit her when th e long-te rm
an d ,th e m te~ ediate- and short-te rm market movem en ts are bottom ing out, index is in th e term inal phase o f a decline o r when it is in an uptre nd , but
Du n ng a p n mary bea r market, me
best se lling o pportunities occur whe n has no t yet reac hed a n ove rextended po sition. These indicators d iffer from
in term ediate- and short-te rm trends are peaking. the previous charts in t.hat th ey are smo othed by EMAs rather than by sim-
In a se nse, any inv~tm en ts made du rin g th e early and middl e stages of ple MAs. Trial and error suggests this su bstitu tion as more suitab le fo r this
a bull m.arket are balled out by the fact that th e primary trend is rising, type o f arran gemen t. TIle lon g-term indicator is construct ed fr om the same
whereas investors have to be m uch more agile d uri ng a bear m arket in order time spans as th e m on th ly se ries; th us, 12 mo n ths becom es 52 weeks . and
to capitalize o n th e rising intermediate-term swings. so on .
Only a71 IUtUal EMA cms,wver sh ou ld he interpreted as a bu y or se ll alert
for a momen tum series based on expone n tial sm oo th ing , no t o n a reversal
Combin ing the Three Trends Ideally. it would be very help ful 10 track in direction. Quite ofte n , th e to ng-te rm series sta bilizes bu t does n ot re verse
th e KS-:: for month ly. weekly. an d daily data o n the same chart, bu t p lolting direction , th ereby leaving the ob se rver in doubt as to its true in te ntion. Vital
constraints do n o t easi ly permit this. It is possible. however, to sim ulate these clues ca n o ften be glea ned from the ac tio n of th e short-term and in rerme-
th ree tre n ds by usi ng d ifferent tim e spa ns based o n weekl y data. shown for diate series in conjuncti on with the pri ce action itself
th e Ch icago lu m be r price in Ch an 12-10. This arrangement facilitates ide n- For ex am ple. in early 1997 the long-te rm KST stabilized, but. precedi ng
tification of bo th the d ir ectio n and the maturity of the primary tren d this even t the interm ed ia te series had cro ssed bel ow its EMA and violated
(shown at the bottom) as well as th e in terrd ationshi p between the short and a trendline. The sh o rt-te rm se ri es also co m pleted a top and crossed below
the inte~ediat~ tre nds. The line' above the price re flec t primary bu llish zero, while the price itsel f viola te d a nice up tre ndline . All these signals sug-
and bearish environments as re flec ted in the lon g-term KST crossing above gested th at th e price was a t best like ly to move sideways. Even u nd er th at
and below its 26-wee k EMA. Note h ow th e sho rt-te rm KST rarely m oves to scenario , th e long-term KST could be projected to d eclin e since th ere wou ld
th e oversold zo n e duri ng the b ullish e nvironments and rarely to overbought be a tre me ndous dissipation o f upside m om entu m resulting fro m these
when the m ain trend i, d own . signals.
Th e KST can also be adap ted to rel ative strength lin es. Th is is especially
useful fo r the long term wh e n ap plied to in dustry groups or in di vidu al
stocks. This is because the gro up rotation around the bus iness cycle m eans
Chart 12-10 Lumber, 1994--2001 , and three KSTs. (From www .)
that linear up- an d downtrends a re far less likely t.o develop th an with
absolute price data. For a fulle r discussion on these m atte rs, re aders a re
l umber referred to Chapter 16, o n rela tive streng th, Chapter 19, on gro up rotation,
roo '~ " " " ' ~ • ••• _ _ _ _ and Chapter 3L, on individual stock sele ctio n . For those who do not have
access to a ch arting package ca p able of plotting t.h e KST, a use ful subst itute
. . .... " ' . .. . ' . . .. .. ' " '' is the MACD, suitably smoo thed with longer-t.erm MAs.

Short-term KST
0 The Directiona l
...... _ . Movement System
', The o bj ective of the Dire ctional Movement System designed by We lles
10 rmedlate KST Wilder is to determine wheth er a market is likely to ex perience a tre nd ing
or trading ra n ge environment. The d istinction is impo rtant because a
sec ~endin g m arket will be bette r sign aled by the ad op tion of trend-follo win g
0 ". IIldicators suc h as m oving averages (MA5), whereas a tra d ing ra nge envi-

"" '" '''' '" ... ,... ronment is more suitable for oscillators. In practice , 1am no t im pressed with
the capability of t.h e Direction a l Movem ent Syste m to accom plish this objcc-
248 Part 1: Trencf. Detennirmg T~ 249

uve , o th er than 10 ide ntify a chang e in trend . O n the othe r h and. th ere are davs of negauve di rectional movemen t fro m the positive ones. However,
several o ther ways in which this ind icato r ca n be usefu lly applied. when the - Dl is greater than th e + DI, the negative sign is ign o re d . This
mean s m at th e ADX o n ly te lls us wh ethe r th e se cu ri ty in questio n is ex pe-
riencing d irectio n al m ovem en t o r not. Af,,rain, th e n ormal defau lt ti me spa n
The Concept is 14 davs.
The AoX is calculated in such a way that the plot is always containe d
Th e calcu la tio n o f th e Directional Move me nt System is quite invo lved and
within th e Ka le of 0 to 100. H igh readi ngs indicate th at m e security is in a
time does not permit a fu ll d iscussi on h ere. For that, reade rs ace re ferred
to Wilder's Nno C07lUpts in Ttdl1lical Tradi ng. trending mode (it h as a lo t of d irec tional movement), and low read ings indi-
To simplify matters , the direction al movement in d icator is plo n ed by cal- cate a lack of d irectional movement and an- m ore indicative of trading range
mar kets. Un like o the r oscillators, th e ADX tells us no thing about th e direc-
culating th e m axim u m range th at th e p rice h as m oved. ei ther during the
period und er consideratio n (a day, week. l a-m in ute bar, and 50 o n) o r from
tion in wh ich a pri ce is m oving , only il\ tren d ing o r no n trending chara c-
teristics. Use o the r oscillators fo r thi s task.
th e previous period's d ose to the extr eme poi nt reach ed duri ng the pe riod.
In effect, th e system tri es to measure di rectional m ovem ent. Sin ce there art:
two di rection s in which prices can move, ther e are two direction al move- I
men I indicato rs, called + DI an d - DI. Since th e raw da ta de rived from the The Two Dis
ca lculation are unduly volatile , th ey are each calculated as an av~rage ove r
a specific time period a nd th e result is ploued o n a chan. No nnally, th e Db
are ove rlaid in the sa me cha rt panel. The sta ndard, o r default, time span is
DIm 12-11 features th e p rice o f ISM to gether with e Dis using a l4-day time
span. Buy alerts a re signaled when th e + DI crosses above the - DI a nd vice
versa. In th is example. there are several occasio n s whe n it is possible to co n-
14 periods. Chan 12· 11 shows th e two Db wing a l 4-day span. Crossove rs
of th e Dis are the n u sed as buy and sell signals.
firm such crossove rs with a tr endline violation in the price . MA crossovers
or price pattern s could j ust as eas ily be su bstituted. In this example . th e I
The re is o ne othe r importa nt in d icator in corporated in th is syste m and crossovers are p ret ty good and not su bject to any whlpaaws. Unfortuna tely,
things d o not always work out as well as th is. That's o n e reason why it is
that is th e Aven ge Directional In dex (ADX ). The ADX is simply an aver-
age of th e + Db and - Dis ove r a specific period. In effec t, it su b trac ts the importan t to make su re th at su ch crossovers are co nfinned by the p rice. I
Culrt 12 ·1 1 IBM, 2000-2001, ard two Db. (from w\Qu'-pring.rom.) OI&rt 12-12 Akoa.. 1999-2000, two smooth0:iOb. andan ADX- (from _ .) I

«I .. . . . . -1 ~yOl

315 '-' :
30 :~ • ;' ". ,f:

"" J<,/" •,.,.,;\"" ~:1'."

'' V ' "'J' ..... . . •. . • - l "·day Dl (10 MAl
" .
10 " -. ~-'

, . .. • , ... • • F. . .
250 Part I: Tmd-Determlning T«Iw1~ b;Ividual Momentum IndJur,lon U 251

Another ~y aro~nd thia is to smooth the two DIs. as I have done in Olan the ADX. having indicated a strong d irectional movemen t. the n falls back
1 2.1 ~. TIllS certainly reduces th e whipsaws, but there is a U'ad~fT in that below its 40 level. The Tt"VeT5a1 warns us that th e tr e nd or s trong directi onal
me sign als an: occasio nal ly delayed because this approach is less sens itive. movement has now reversed, a nd we should be on the lookout fo r a ch ang e
of lre nd. In th e first instan ce , it s fro m u p to sideways, in the seco nd from
up to d own , a nd o n th e far righ t of the chart from down to u p .
n., AD X
There is ra rely an optimum tim e spa n for any indicator. H owever, fo r the
A high ADX reading does not tell us that th e mark et is overbough t and ADX. the 14-period span and th ose very close to it appear to offe r th e best
a~ut to do d?WII. P1~ note that a clumgrin trend is different from iI mArT- results. Chart 12-14 sho ws two extre mes: a &day and a 3O-day sp an . They
JOl In trend sinc e a change in trend can be from u p to down up to side- demons tra te that shorter-te rm time fram es return an ADX mal is too
ways. or down to u p . Simibrly... downtrend could change to a Bnge 'Olatile, while longer-term ones result in a step-like pattern.
o r to an uptrend. The arrow agai nst the 6-day series abo goes to show that a reversal in d irec-
In Chart 12·13, a l +day ADX has been plotted against Alcoa. I ha ve- abo lion does no t always resu lt in a change in tre nd . In this case , a reduction in
p laced a~ Dve.nold (li ttle directional movement) at 15 and an overbought directio nal m ovemen t was ac curately p red icted by th e reversal .
(I,o ts o f di rectional m ovem en t) at 50. Since th e vola tility of each security it Unfortu na tely, the re duc tion was not sufficient to resul t in a change of trend,
. hi
n t, suc h lines sho uld be d rawn o n a trial-and-error• 1.. _- '

wang lhe merely a slowing down in the existin g (do wnward ) o ne. Fortuna tel y. th is
price !Story (as much as poMib le) as a guide. Th«e ~ th~ periods when kind of action is the exce ption rather than th e rul e.
Chart ]2-12 zeros in o n o ne of these periods. but it also includes a l o-day
MA of the l4-day u p and down Dis . These a re displayed in o rder to o ffe r
some idea as to the im plication of th e reversal in th e ADX . In th is case, the
crossing of the 50 level (n ote that th is overbought readin g is n o t the sa me
as that in Chart 12-13) is foll owed by a negative CT0550ver o f th e sm oothed
DI above ill! + 01 cou n terp art. This indicates th at the reversal in the ADX
is likely to be fo llowed by a d ecline .

OY.rt 12·13 Alcoa. 1999-2001. and an AOX.lFrom UIlIPUl.pr1ng.CDf7IJ Ot.ut 12 ·14 AT&T, 199 8-2001 , and two ADXs. (FtOm _ .)

... ."'"
" •

... 1
~ t'e'-""".:L!AO";;Xi-- f-t-----,I-Jl;----- __1r
"10 f'---''--'---¥---\rt~W--1;AAA-,.J.~--''i~~ "
20 3Cklay

s o
252 Part I: Trend-Determining Techniques Individual Momentum Indicators II

Low readings in the ADX indicate a lack of directional movement, and How Does It Work?
these can be helpful as well when it is fairly clear that a rising trend of direc- The parabolic shows up on the chart as a parabolic-shaped .curve tha~ is plot-
tional movement is underway. Consider the situation in November 1999. ted above and below the price, as in Chart 12-15. ThIS curve IS o~te~
The ADX had been in a narrow trading range at a very low level. Then it referred to as the SAR, which stands for stop and reversal system. This I.S
broke to the upside. This indicated that the price was now likely to trend, because the parabolic, when triggered, is often. used not .only to stop a .pO.SI-
but in which direction? The answer lay in monitoring the two smoothed DIs tion, but to actually reverse it. Thus, a parabolic wo~ld SImultaneously tng-
and the price. In this instance, the +DI crossed above the minus series and ger the liquidation of a long position and the entenng of a short .one.
the price was confirmed with a nice breakout. Remember that the rising Personally, I prefer to select an entry point and use. the ParabolI: Syste~
ADX did not signal rising prices, only rising directional movement. If the for the purposes of setting the stop, rather than continually pOSI-
- DI had crossed below the + DI and the price had broken to the downside, tions. Remember, most losing trades deoelop when you go agamst the dtrectwn oj
the rising ADX would then have been associated with a declining trend in the main trend. Strict use of the parabolic with the stop and reversal approac.h
the price. totally ignores this discipline. That's why I prefer to use the Parabolic
System as a stop system only, and no~ a stop an~ reverse system ..
Strict use of the parabolic would involve buymg when the pnce crosses
above it, placing the stop well underneath the purc~ase price. This is the
The Parabolic Indicator
downside because there is considerable risk at this point. Gradu~lly, the par-
The Concept abolic picks up steam and the risk is dramatically reduced as upside momen-
The Parabolic System, devised by Welles Wilder, is not a momentum indi- tum increases. In Chart 12-15, the sell signal comes pretty close to the top
cator, and so strictly speaking, it does not fall within the scope of this chap- ofthe rally. The short signal in November also results in a nice. profit..
ter. However, it is being mentioned briefly since it has become a very Like most other indicators, the parabolic does not do as well m a tra~mg
popular trading mechanism and can be used to generate timely stop-loss range. This is because it is unable to gather sufficient momentum to quickly
One of the most valid criticisms of trend-following systems is that the
implied lags between the turning points and the trend-reversal signals oblit-
Chart 12-15 GM and a parabolic, 2000-2001. (From
erate a significant amount of the potential profitability of a trade. The
Parabolic System is designed to address this problem by increasing the speed
of the trend, so far as stops are concerned, whenever prices reach new prof- Ii ..~---Sell
itable levels. The concept draws on the idea that time is an enemy, and unless
a trade or investment can continue to generate more profits over time, it
should be liquidated. Since it is a stop-loss system, it can be used with any
momentum series, once that indicator has been used to filter out a good
entry point for a trade. It is also a trailing stop-loss system, which means that
the stop is continually being moved in the direction of the position, that is,
60 ~ .......,.--- Buy
up for a long position and down for a short one.
When the position is first initiated, it is given a relatively long leash, so to ";jf-----Slop
speak. Then, as time passes, and the price increases, the stop is gradually 55
tightened. The expression parabolic arises from the shape of the curve of the
stops as it appears on the chart. In a rising market, the stop is 50
being raised, never lowered. In a declining market, the opposite wIll
hold true.
254 Part I: Trend-Determnng T~ 255

Qu,rt 12·16 HewietI~ am iii parabolic.. (From lWIU'.po1ng.corrr.1 less timely th e signal, and the h igher th e ac cele ration factor, th e more se n-
sitive and timely are th e sign als. Ho weve r, tl ley res u lt in more wh ipsaws.
,. In Ch an 12-17. th e u pper panel featu res an acce leration factor of .0 1 with
a m ax im u m of .2. whereas th e lower ch an sh ows a m ore aggressive .2 accel-

" eration with a .2 m ax imum. Th e d iffe rence is very dea r, The sma lle r acce l-
eratio n facto r featu red in the to p panel results in fa r fewer wh ipsaws tha n
that in th e lowe r p anel, b ut at the same tim e is less tim ely, Setting leve ls fo r
th e acce lerating factor is a trial-and-error p rocess just like the p rocess of
det ennining th e o p timum span fo r an MA In m y experience, th e defau lt
" V
va lue provided with many software programs and Wilde r's rec o mmendation
.,f .02 and .2 are q u ite satisfactory in m OSI situatio ns.

A Practical Use

Th ere arc man y waY' in whi ch the parabo lic can be applie d . One o f my
" , , , favorites is to en te r a trade on an MA crossover and exi t the trade o n a
<t- • • • • reverse crossover. Howeve r, if the parabolic crosses the MA. m e n il shou ld
be wed to exi t th e tra de. An exa mp le is sh own in Chart 12-18 with General
Mot on . For a mo re in-de p th exp lana tio n , please refer to m y own
Mom entum Exp lain ed book a nd CD-RO M tuto rial.
reduce the risk. In Chart 12·16, we see five tra des, th ree long (the solid
arrows ) and two sh ort (the dashed alTOW!I) . No ne ma ke muc h mo ney. bu t
th e winners slightly outnumber the losers th ree to two.
O\m 12 -11 HevAett-Packard and two parabolia_(From WWUI_prlng.eom .1
Setting the Parameters
Most charti ng packa ges enable th e user to se t two J>M'lffieten . These ace
the acceleration factor and th e maximum acceleration factor. The acceler-

a rio n fac to r is the am o um th e para bolic is in creased every time th e pric~

m ak es a n ew hi gh in an u p tre nd (a new low in a downtre nd). The maxI-
mum factor is a limit above whi ch the acceleratio n factor canno t go. For "
e xa m ple. if the acc elera tio n factor is .2 and th e m ax im um is .8. after th e
p rice h as m ad e four n ew high s since the parabolic began (that is, .2 tim~
• Pafil bok 2J.2
4), lh e acce leration factor remains a t .8 un til the pri ce crosses the para bohc .
an d a n ew (SAR) trad e is begun. The level of the acceleratio n factor is the
most im po rtan L
with lo.1A!, the re is always a trade-off between tim eliness and sensitivity. A.
sho rt-te rm MA gives very timel y signals, but is so se nsitive th at it also gen-
erates n umerous whip saws. O n the o th er hand, a long-term average has
fewe r wh ipsaws, b u t crossovers are los timely. The Parabo lic System wor U
in exactly same way, but in this case th e lowe r the ac cel e ration facto r, the ., ....
256 Part I: Trend-Determining Techniques

Chart 12-18 GM, a parabolic versus an MA crossover. (From


+-- 25-day MA

Candle Charts


~ Candle Construction
., 19 26 3 10
Candle charts originated in Japan several centuries ago, but hav: recen~y
gained a following in other countries. This system of plotting. pnc: da~ IS
an alternative to the familiar bar chart. Candles can be used to Identify pnce
Summary patterns, and they also lend themselves to trendline constructio~. They ~re
constructed from opening, high, low, and closing data for a specific penod
• The KST can be constructed for any time frame, from intraday to primary.
such as an hour, day, or week.
• The KST is calculated from the smoothed ROC of four time spans, each
Bar charts are expressed as vertical lines, with the left and right handles
of which is weighted according to the length of time.
representing the opening and closing levels. Candle charts, on th: other
• Long-term, short-term, and intermediate KSTs can be combined on one
hand, are plotted as vertical rectangular boxes that connect the opemng and
chart to reflect the Market Cycle Model.
closing prices, and with vertical lines that extend from the rectangles to
• The KST lends itself to numerous momentum-interpretive techniques and
encompass the extreme high and low. Bar charts treat all.da~ m~re or less
can successfully be applied to relative strength analysis. .'
equally. It is also important to differentiate between which IS higher, the
• The + DIs and -DIs measure positive and negative short-term direcuon-
open or the close. . .
• When the raw or smoothed DIs cross, they trigger buy and sell momen-
Candle charts can be plotted only for markets in which opernng pnces
tum signals.
as well as closes, highs, and lows are known. Proponents of this system
• The ADX measures the directional movement of a trend.
believe that it provides all the information contained in bar charts as well
• A rising ADX indicates an increase in directional movement and vice
as additional ideas contained only in candle charts. Candle charts are ~ot
the Holy Grail, but they are certainly a useful adjunct to the techmcal
• When the ADX reverses direction from a high reading, the prevailing
trend is likely to change.
• The parabolic is a stop-loss system. d
• The parabolic was originally designed as a stop and reversal me tho .
Major Technical Principle .·In candle.sticK charting, .the greatest
However, it is better used as an exit mechanism due to the large risk ofte.n
emphasis isgiven to the opening and Closing prices and the trading
associated with the initiation of a position based on an initial paraboh c
range between them.

258 Part I: Trend-Determining Techniques Candle Charts 259

arsenal. In this description, we will refer to days since they are the most com- a bullish tendency. Figure 13-1 (b) is the long black line, a wide trading range
mon timeframe, but bear in mind that candIe charts can be constructed for where the opening is close to the high and the close near the day's low. This
any time period. candle has a bearish tendency. Figures 13-1 (c) to (e) show doji lines, where
A typical candle consists of two parts: the real body, that is, the rectan- opening and closing prices are identical. Interpretation depends upon the
gular part, and the shadow or wick, that is, the two vertical extensions. The context in which they appear, but they essentially indicate a balance between
top and bottom of the rectangle are determined by the opening and clos- buyers and sellers. Figures 13-1 (j) and (g) illustrate umbrella lines, where the
ing prices for the day. If the closing price ends up above the opening (the real body is narrow and develops at the high end of the day's trading range.
real body), it is plotted in white. When it closes below the opening, it is plot- Umbrella lines are bullish at bottoms and bearish at tops. Figures 13-1 (h)
ted in black [see Figs. 13-1 (a) and (b)]. The top of the real body represents and (i) show spinning tops, representing days when the trading range is very
the opening price, the bottom the close. This is reversed in the case of a small. They do not have any significance in trading ranges, but are impor-
white rectangle where the close is plotted at the top and the open at the tant in some of the price formations discussed later.
bottom. Candle charts offer indications of both reversal and continuation phe-
The thin, vertical shadow lines that protrude from the real body reflect nomena.just as bar charts do. Some examples in this chapter demonstrate
the high and low for the day. Since the closing and opening prices can be
identical, or identical with the high or low, there are a number of possible
combinations that need to be represented. Some of them are shown in
I candlestick formations (see Charts 13-1 to 13-4). Candlesticks really come
into their own in the identification of shorter-term reversals and continua-
tion situations.
Fig. 13-I.
Candlesticks provide essentially the same information as bar charts, but
their more pronounced visual representation of the material enables tech-
nicians to identify characteristics that are less obvious on bar charts. Certain
phenomena illustrated in bar charts have been given their own names, such Chart 13-1 Financial Times Stock Exchange 100 cash, 1990. (From
as key reversal days or island reversal days, likewise, with candle charts. Because
of the large number of potential variations for both individual days and price

·· .
formations encompassing several days, it has been common practice to give
e ase .0

exotic names to the various possibilities. The characteristics of the more

common candles are also shown in Fig. 13-I.

Figure 13-1 (a) shows the long white line, a wide trading range where the
opening is close to the low and the high is near the close. This candle has
· •,"=-.,,·L
tUO.O 2::U:O.O


.~ ~;h961
~;~Q~~ t. tr1rt:'~"'"
" 2100.0

ffi t~a~~QtI 20$0 _ 0

toClCl.O . . . . ~t
(a) (b) (e) (d) (e) (f) (g) (h) Q) ltSO. Q \/L

Figure 13-1 Selected one-bar candles.

260 Part I: Trend-Determining Techniques Candle Charts 261

Chart 13-3 Microsoft, 2000-2001, candle volume. (From

Reversal Phenomena
Hammers and Hanging Men (Takuri and Kubitsuri) 80
These formations [see Figs. 13-1 (f) and (g)] have probably gained more
notoriety than all the others because of their imposing titles. A hanging man
is an umbrella line that develops after a rally. It looks rather like the body
Dark Cloud
of a man with dangling legs and, as its name implies, is a bearish pattern.
If a hanging man appears after a prolonged upmove, it should be treated
65 Harami
with respect, especially if it occurs after a gap. A hanging man can be iden-
tified by the fact that the shadow, or wick, is at least twice the height of the
real body. The color of the body is not important, 60
A hammer is identical to a hanging man, but occurs after a market decline
when it is a bullish sign. It gets its name from the idea that the price is "ham- 66
mering out" a bottom. In effect, it represents the kind of trading day when
the price temporarily slips quite sharply, because there is a run on the sell-
ing stops. Nevertheless, the technical position is sufficiently constructive to
cause buyers to come into the market and push the price back up toward
or above the opening level.

Chart 13-2 New York light crude three-month perpetual, 1989-1990. (From Chart 13-4 Treasury bonds, 1990. (From



.• • • • • • •.• . . . . . . • • • .• • • • • • • .•.~< . .•. . ~fl'

" .00

tIE .00
n.tlo H.OO

"······&V~r&··· .
L .. n,oa ~!l' .00

1'1.00 ,OJ .00

:t1. 0 0

n,pa ~:!I _00



+\l][ . "!to




19.00 . - . J! ,00

, •. 50

... "0 tlOU


Dark Cloud Cover (Kabuse)

Part I: Trend-Determining Techniques

In real life, dark clouds (see Fig. 13-2) hint at the possibility of rain, so a
I, Candle Charts

Engulfing Pattern (Tsutsumi)


This formation (see Fig. 13-4) develops significance after a prolonged price
dark cloud candlestick formation implies lower prices. Its bearish connota- move. It is characterized by two consecutive more or less shadowless real
tions are most pronounced during an uptrend or in the upper part of a con- bodies, in which the second day "engulfs" the first. It is bullish in a down-
gestion zone. It is a form of key reversal, since the price closes down on the trend, when the second day is a white body, and bearish in an uptrend, when
day after a gap's higher opening. It consists of 2 days. The first is a strong, it appears as a black one.
white real body, and the second is a black body, in which the close occurs
in the lower half of the previous white real body.
Stars (Hoshi)
Stars (see Fig. 13-5) are common phenomena in candle charts and come
Piercing Line (Kirikorni) in four different reversal varieties. Stars are combinations of wide real bod-
This pattern (see Fig. 13-3) would be more aptly named "sunny sky" because ies and spinning tops. The morning star heralds a new day (upmove) and is
it is the exact opposite of the dark cloud and is therefore bullish. It is impor- bullish. It consists of two long real bodies separated by a spinning top. The
tant to note whether the second day's white body closes more than halfway star is represented by the spinning top, which is made on a gap. The third
above the previous body. If it does not, conventional wisdom indicates that body should be white and should result in a closing price more than
additional weakness is likely. halfway up the body of the first the dotted horizontal line in Fig. 13-5. The
evening star is a precursor of night. It has the opposite characteristics and
implications of a morning star.

Bullish Bearish

Figure 13-2 Dark cloud cover.

Figure 13-4 Engulfing patterns.

---'-----0------ Q,
--0-- - -
0+ O·
EveningStar Morning Star CoilStsr Shooting Star

Figure 13-3 Piercing line. Stars.

264 PlIrt I: Tnmi-Determlning T~ 265

.... dnji JW is a bearish sign and occurs after a lengthy It consis ts of a ill o r dose to its session low (see Fig. 13-K) . You can see that none o r th e
gap and .a d oji lin~. t\ Jlwoli ng sta r is like a short-term top where the daily candles have a lower ....ick (see Fig. 13-9 ) a nd each o r the three rea l bodies
pnce acuo n e xpe rie nces a small gap and the black real body ap~rs at the ope ns Voithin the ra nge of th e p revious session's real body.
end o f a long wick o r upper shadow.

Tweezer Tops and Bottoms (Kenukl)

Upside Gap Two Crows (Narabl Kuro) If you h old a tweezer upsid e down, YOII will see tha t the two poi nts are at
Thls bearish fonnatio n (see Fig. 13-6) co nsis ts ofa long while line fo llowed ide ntical levels (see Fig. 13-10). The sam e is true o r a t""tur lop, which con-
by lWO black lines . TIle first black line gaps to th e upside. The- third day often si5LS of two candles that th e h igh o f the day is ide n tical to . Actu ally, it's po'i-
doses th e gap. but because it is a black line where the close is bel ow the sible for a tweezer to com ist o f m o re th an 2 days with an iden tical lOp . Make
open, its im pl icatio n is bearish . no mistake about it, we are talki ng aboar the hig h . ~,..h ich can be a shadow
or real body close o r open . This patlern is sho rt-term bearish be<alL~e the
fint cby 's h igh acts as re!oi.'ltance ; whe n the second day is unab le to punch
Three Bla ck Crows (Sanba Garasu)
The th ~ block rnnvs pattern in Fig. 13-7 co nsists o f Ih R'C d ec lini ng bla ck am-
dlesticks that form after an advance . TIley ind icate lower prices. Each black
can dl e sho uld o pen wilh in th e real body of its black p redecesso r and close

Agure 13 -6
Upside gap two crows.
Agure 13-8 Three bl.Kk crows ,

o· *-;;----
Agu", 13·7 Three bIadc. crows.
26 7
2 66

thro ug h the ho rizo n ul line th at ma rks the area of th e top. it in d ica tes a 10Il!J
o f upside m omentum. In this example. th e secon d d ay is also an engulfing
panem, which is an important point since a tweezer often contains a price
pattern as pan of its fo rma tion .
A twrtUT oottum occurs whe n, after a declin e . two o r m ore candles make
a n ide n tical low. This agai n indicates a loss o f downsid e mome n tum since
the p rice finds support in the area o f th e low. In Fig. 13-10, we SC':c a tweezer
literally hammered out because th e second lo w to touch the horizontal line HIIfIl)"'II Man
is a part of a ham me r cand le.
O n e factor th a t will Incre ase the sign ificance of a tweezer is th e natu re of
the pattern be ing formed. Fo r e xa m p le . if th e seco nd day of a tweeze r top
is a h angi ng man, as in Fig. 13-11, we have two pi ec es o f evidence that the
tren d m ay be about to reve rse : the tweez e r a nd the hanging man . flgvnr 13-11 Tweuer top and a hanging man.
Alternatively, the second d ay m igh t be a d oji, so th e tweezer is reall y a harami
cross (see Fig. 13-20 ). It co uld also be a shooting star. Fin ally, a tweezer bot-
to m may also be ;I moming star. haram j, hamme r. an d .50 forth.

Belt-Hold Unes (Yoriklri)

A bullish Mr hold (see Fig . 13-12) is a one-candle pattern co ns isting ofa long
ca nd le in wh ich the p rice opens o n th e sess io n low and th en works its "'':1y
higher throughout the sessio n . The p rice does not have to close at th e high ,
b ut the lo nger th e rea l body. th e m ore positive the ca n d le. Also , if a belt
hol d has n ot appeared on the chart for quite a wh ile. it is the refore an

F\gur. 13-12 Bulhsh belt hold,

unusual phenomenon. As a resul t. it ga ins in Importance. This is because

traden are m akin g a very stron g statem e nt abo ut thei r fe elings towards Ih~
market with a belt h o ld co m p ared to the sma ller can d les th a t were 111"1: \1 -
ously th e norm. th in k o f a crowd mu rm u rin g. a nd then ;I loud m ice
comes from ...i thin the crowd. It is obviously a penon whowanl! 10 be h eard.
.Well, the bel t h old fo llowi ng a long period of sma lle r can d les am o unts to
the same thing. It 's a l-day p altem th at says lou dly. "Liste n 10 m e because
. ~7 1. am Id lin g yo u the sh o rt-term trend ~las chan ge~. - .
if '.~~l l·'T' A bearish belt hold (see Fig. 13-13) 15 the o p po" l!e. 1t \ S ;I lon g black ca n-
die' in which the price o pens at the hi l(h and the n works its ",-a~ lower as ~ he
flgur_ 13 - I O T_ lops and bortoms. . ~on progresses. Bell ho lds are often im porta nt pivotal days since the lu gh
268 Part I: Trend-Determining Techniques Candle Charts 269

Bearish belt hold

. - on the high

Bullish counterattack line ---+

Figure 13-13 Bearish belt hold.

Figure 13-14 Bullish counterattack line.

and low occasionally act as support and resistance areas in subsequent price
action. The halfway point of the real body of a belt hold should also be mon-
itored for a possible price reversal during later price swings.

Counterattack or Meeting Unes (Deai Sen/Gyakushu Sen)

A bullish counterattack line (see Fig. 13-14) develops when, after a decline, a
black candle is followed by a white candle and both close or "meet" at the
same level. This is why this 2-day pattern is sometimes referred to as a meet-
ing line. The first day is usually a long black candle. The second day opens
sharply lower, leading most traders to believe prices will continue to give way.
However, by the end of the day, the price has regained everything lost (a Figure 13-15 Bearish counterattack line.
counterattack by the buyers) and closes unchanged. The meeting line
therefore indicates that the downside momentum has probably dissipated
and a reversal in trend is likely.
A bearish counterattack or meeting line (see Fig. 13-15) is formed when, More specific rules for identifying these patterns are as follows:
after an advance, a white candle is followed by a black candle and both close
at the same level. The psychology behind this one is fairly evident. The • The first day is colored in the direction of the prevailing trend and the
sharply higher opening on the second day has the bulls in a euphoric mood second day forms in the opposite color (white/black for tops and
since these new gains come on top of an already sharp rally. However, eupho- black/white for bottoms).
ria turns to disappointment as the price unexpectedly returns to the • Both real bodies extend the prevailing trend and are long.
unchanged level. • The closes are identical.
270 Part I: Trend-Determining Techniques Candle Charts 271

Continuation Formations
Upside Gap (Tasuki)
A tasuki gap (see Fig. 13-16) occurs after an advance. The requirement is
an upside, white line gap followed by a black line that does not close the
gap. This type of pattern is usually followed by higher prices. However, if
the gap is filled, the formation deteriorates into an upside gap with one crow Bullish Rising Bearish Falling
and therefore loses its bullish portent.
Figure 13-17 Rising and falIing three methods.

Rising and Falling Three Methods (Uwa and

Shita Banare Sanpoo Ohdatekomi)
These formations (see Figs. 13-17(a) and (b)] are very similar in conceptto
a flag in bar charting, except that they take only a few days, not weeks, to
develop. The rising method is a bullish pattern and consists of a powerful
white line followed by a series of three or four declining small black lines.
These lines should be accompanied by a noticeable contraction in volume

o o'u
that indicates that a very fine balance is developing between buyers and sell-
ers. The final part of the pattern is a very strong white line that takes the Window.--
price to a new closing high. If volume data are available, this final day should
record a significant increase in activity. The bearish falling three method is
exactly the opposite except that volume characteristics are of no significance ' " Closed
on the last day. An example is featured in Fig. 13-17(b).
Figure 13-18 Windows.

Windows (Ku)
Japanese chartists refer to gaps as windows (see Fig. 13-18). Whereas gaps
are said to be "filled" in traditional bar charts, windows are "closed" in can-
dle charts. Windows therefore have the same technical implications as gaps.
(Refer to Chapter 7.) Harami tines (Yose)
In Chapter 8, it was mentioned that trendline violations are followed either
by a reversal or a temporary consolidation. The harami formation, shown in
Fig. 13-19, is similar to the consolidation trendline break in that it indicates
a loss of momentum. The main difference is that harami lines are of much
shorter duration and consist of 2 days' price action. The second one, the
forms a real body that is sufficiently small to be engulfed by the prior
long real body. If the harami is also a doji, as in Fig. 13-20, it is called

Figure 13-16 Upside gap.

o UpsideGap
a harami cross. After a sharp rally or reaction, these patterns indicate a bal-
between buyers and sellers after one or the other has predominated.
means that haramis often warn of an impending trend change. In some
.. ',!IStan{'I>.~ this will be from up or down to sideways and in others an actual
272 Part I: Trend-Determining Techniques Candle Charts 273

Chart 13-5 Microsoft, 2000-2001, candles and an RSI. (From


Figure 13-19 Haramt,

Harami Cross

Figure 13-20 Harami cross.

Candle Charts and

Western Techniques
There is a tendency among many technicians to look at candlesticks in iso- Candle Volume Charts
l~tion. My pre.fer~nce, reme~bering the weight of the evidence approach
dlscusse~ earhe~, I.S to combine selected Western charting techniques with Candle volume charts are the same as regular candle charts with one impor-
candlesticks..This mvolve~, amongst other things, the inclusion of price pat- tant difference. The width of the real bodies varies with the level of volume
terns, trendlines, and OSCIllators into the analysis. during that particular session. The greater the volume, the wider the real
.Chart 13-5, for instance, shows a head-and-shoulders (H&S) top for body and vice versa. This is a very useful way of presenting the data because
Mlcr~soft that was completed in November 2000. Note the sell off during the signals from the regular candlesticks are preserved, yet the width of the
the nght shoulder was an identical three crow pattern. Later, we see a dou- real bodies offers a quick and simple overview of the volume pattern. We
ble bottom. The rally from the second low consisted of a bullish belt hold, will have a lot to say on volume in Chapters 22 and 23, including the con-
which in ~tself in?icated that prices were headed higher. This bottom was cept of equivolume on which the candle volume technique is based.
also associated With a reverse H&S in the RSI. Finally, the H&S top in the Chart 13-6 shows a candle volume chart for Wal-Mart. Notice how the early
RSI was confirmed with a harami. November window following the bullish engulfing pattern was closed soon
One important question is where to draw the trendlines. Should they after. However, the very thin candles that were involved in the retracement
touch the wicks, real bodies, or a combination of both? The answer lies in move indicated a lack of volume, which is precisely the type of thing that is
applying ~ommon sense. Since the opening and closing prices are gener- ~equired in a pullback of this nature. Heavy volume in such a situation would
ally ~or~ Important than the high and low, lines that exclusively touch real mdicate selling pressure as opposed to the situation here where prices were
bodies Will generally be more significant than those only touching the wicks. clearly falling because of a lack of buying interest.
However, a longer line that has been touched on more occasions and only The previous rally experienced a series of very thin candles, which indi-
touches the wicks will probably be more significant than one that is relatively Cated that prices were rising on lower volume. This is opposite to the norm,
short and only touches two real bodies. Where rising prices and volume are healthy. The very thin candles therefore
Warned that the days of the rally were numbered.
"-- '-- - 27 5
2 74 13-6 Wlll-Mart. 2000-2001, and cardIe (From .)

QIatt 13-1 M .""""""...
-~- - .. 2000-2001. and andko wbne. (from WU/UI.prlng,com .)

. VlJaI-Mart (Caodle Volume) +-- High volume , long wMe line


67 Low vokl me aod .........
SIT1lI ~. res t bodies

".."" 1
~I 'f~l'Il\I
... 16

."... I s

... w_ I
.- • .... 1 • • ......,

:~ • 2

The late November/early December rally was associated with wider can-
dies, which W3.5 a good sign . However, as we reach th e wee k o r th e fourth,
th e candles move sideways, but are very thin. This indicates th at th e balance
0w113-8 BoMg. 2000-2001. and undle vcIune. 'From lDU'UI' .)
between buyers and sellers W<U more n'enly m atched. The doji on the day
or the high also re flects th e even balance. This characteristic is often rol-
lowed by a trend reversal , especially ifvolume p icks up o n th e downside.
That's preciselywhat happen ed her e, as the up tre ndt tne is violated and the
ca ndles th icken up.
The long white line that developed at th e very e nd or December looked

good at th e time , since this candle W3.5 a pretty wide one, indicating ht'3 1
volume. However. there W4S no Iollcwthrough o n the upside, which indi-
cated tha t the lon g white line W3.5 a buying climax. This W:iI.5 co nfirmed fj"t
wi th the establishment or a harami on the subsequent day and by a long
black ca ndle that retraced all of the ground gained by the long white line,
Cha n n -7 shows th e same pe riod fo r Microso ft, as sbo....m in Chart l ~.
but this time with can dle volume. Not e that from Novembe r 30 the bulk of
th e ca ndles ar e thin, ind icating low volume. Th e o nly thick candles are black
ones , which in dicate potential selling pressure. Finally, the righ t-shoulder
rally consists of thin candle s, indicati ng low (bearish) volume, The carly
December bear rally is also associated with th in can dles. Rising prices on '"~
• 20 1
..n ....
wcak volume is bearish because it indi ca tes that pri ce! are ral lying on a lact.
276 Part I: Trend-Determining Techniques

of sellin~ rather. than enthusiastic buying. Then, on the day of the rally high,
the relatIvely thick black candle indicates that the days of the bulls are num-
bered because selling pressure has started to pick up.

You can see that the February down trendline break is associated with a
thick belt hold and a trendline break in the RSL Also, it's important to note
that the nature of the RSI changes with candle volume, because the wider
candles involve a slower, but more deliberate price action by the oscillator,
Thin lines work in the opposite way, of course.
Chart 13-8, featuring Boeing, shows a piercing line in October. Note that
the white candle is fairly thin. Normally, we would like to see a thick one
because that would mean expanding volume. The engulfing day in january
Point and
is a thick candle and the high volume that this reflects adds to the bearish-
ness of the pattern. Finally, note how the candles break out from the sym-
Figure Charting
metrical triangle at the same time the RSI completes a base.

• Candle charts can be constructed only from data that include opening Point and Figure Charts
prices, and therefore the technique is not one that can be applied to all versus Bar Charts
Point and figure charts differ from bar charts in two important ways. First,
• Candle charts provide a unique visual effect that emphasizes certain mar-
bar charts are plotted at specific time intervals regardless of whether there
ket characteristics not easily identifiable from bar or closing charts.
is any change in price. A new plot on a point and figure chart, on the other
• Candle patterns can be reversal or continuation in nature.
hand, is made only when the price changes by a given amount. Point and
• Western charting techniques can be used in conjunction with candles for
figure charts are only concerned with measuring price, whereas bar charts
superior results.
measure both price (on the vertical axis) and time (on the horizontal axis).
• Candle volume charts add an additional dimension to the analysis.
The second major difference is that bar charts record every change in
price for the period they are measuring, but point and figure charts ignore
all price movements that are smaller than a specified amount. For exam-
ple, if a box is set at price movements of 5 points for the DowJones Industrial
Average (DjIA), only price changes in excess of 5 points will be recorded,
and smaller fluctuations will not appear.

Construction of Point
and Figure Charts
and figure charts are constructed using combinations of X's and O's,
as boxes. The X shows that prices are moving up, the 0 that they
are Inoving down. Once the amount of historical data to be plotted has been
established, there are two important decisions to be made before a chart
~n be constructed.

point and Figure Charting
278 Part I: Trend-Determining Techniques

First, the size of each box must be determined. For individual stocks, it ~ox
is common practice to use a l-point unit or box for issues trading above $20 xo xo x o
XO oxo
and a Y2-point unit for lower-priced stocks. However, for very long term 71 x oxo
x 0 ox
x x x oxo
charts or for averages consisting of much higher numbers, it is more con- X X xox 0 x oxo
venient to use 5-, 10-, or even 20-point boxes. As the box size is decreased, 70 ~g~g~8~g~-- g~8x
xo 0 0 0 oxoxox x
the detail of price movement graphically displayed is increased, and vice x oxoxoxoxo
x oxo oxoxo
x 0_ oxox
versa. In following the price action of a stock or market over many years, it 69 x gxg~g~ox XI
is more convenient to use a relatively large box since small boxes will make xo x x oxg~g~g~ 1
X XOX ~ xo x oxoxox l
the chart unduly large and unmanageable. Often, it is a good idea to main- 8~xoxox8xOXO X oxoxo
XOXOX xoxo 0 0 I
tain two or three different versions, just as daily, weekly, and monthly bar xo oxOxo
X 0 I
charts may be plotted. 71--~U~ I
The second decision is whether to use a regular point and figure formula xox I
xox I
or to use a reversal chart (which should not be confused with a reversal pat- XO
6 6>-~ I
tern). The straight point and figure chart is plotted just as the data are X I
X Ix
recorded. If the price moves from 64 to 65, five X's will be plotted on a 20¢- X
6 5~X
point chart, as in Fig. 14-1(a). If the price reverses from 67 to 66, five O's xo~ oxo
xox 0
will be posted. Reversal charts, on the other hand, follow a predetermined OX
4 OX
rule: A new series of X's or O's cannot begin until prices have moved by a OX
specified amount in the opposite direction to the prevailing trend. The use
of the reversal technique therefore helps to reduce misleading or whipsaw 63

signals and to greatly compress the size of the chart so that more data can
be plotted. Figure 14-1(b) shows the same data plotted as a line chart. 62

The construction of 1/2-point, 5-point, or 10-point charts, or charts by any

other measure, is identical to the previous method except that a new box
can be posted only when the price has moved by the degree specified, that (a)
is, by 1/2 point, 5 points, or 10 points, respectively. Since only price is
recorded, it could take several days or even weeks before a new box is plot- Figure 14-1 20¢ point chart.
ted. Hence, a common practice is to record dates either at the foot of the
chart or in the boxes at the appropriate points. A combination of both date
locations is used for longer-term charts. For example, the year is recorded
at the bottom of the chart against the column that the first posting of that basis corresponds to keeping several point and figure charts using various
year was made for, and the beginning of each month is recorded in a box unit sizes.
using the number of the month, 1 for January, 2 for February, and so on. Point and figure charts are plotted on an arithmetic scale. If drawn on
The decision about unit size (and thus the degree of price change paper, they would have traditionally been constructed with 8, 10, or 12
required to trigger a new column of O's or X's) is essentially based on per- squares to the inch. Occasionally, point and figure charts are plotted on a
sonaljudgment. It is determined by the price range and degree of volatil- semilogarithmic or ratio scale, though this is not the norm, because price
ity of the indicator, stock, or market under consideration. Reducing the size Objectives are calculated in a different way than those on regular charts with
of the units (figures) increases the detail of the price movement portrayed. a time scale.
Making the unit larger expands the base of data, that can be included, but Data published in the financial press covering the high, low, and close for
this limits the number of fluctuations that can be illustrated (see Chart 14-1). specific stocks are not suitable for accurate point and figure charts. For
Following a market with bar or line charts on a daily, weekly, or monthly example, if a $15 stock has an intraday price range of $1 1/2, it is impossible
280 Pa rt I: Trend-[)e termining Techn~