Beruflich Dokumente
Kultur Dokumente
Research
Group
ASSET ALLOCATION
MODEL ATTRIBUTION: A CASE STUDY
JUNE 2015
TABLE OF CONTENTS
Executive Summary......................................................................................................................................................... 1
Introduction........................................................................................................................................................................ 2
Conclusion......................................................................................................................................................................... 19
This edition of NDR Solutions Quarterly discusses performance attribution for tactical asset allocation models. As
originally stated, the goal of this publication is to provide our client base with unique investment ideas each quarter that
stretch beyond Ned Davis Researchs traditional strategy content. Through this publication, the NDR Solutions team can
showcase the varied skill sets and diverse experience we regularly provide to clients with direct engagement, customized
output, and tailored solutions. As always, we welcome any feedback, as well as questions about our services, and hope
that you find this publication insightful, unique, and useful.
EXECUTIVE SUMMARY
Implementation of a performance attribution framework is essential for the proactive monitoring and maintenance of
quantitative models. The attribution process evaluates the performance of specific components in a models hierarchy at any
point in time. While much emphasis is traditionally placed on model development and historical returns, attribution determines how well a model meets expectations as we move farther away from its back-tested history. By identifying a models
successes as well as shortcomings, attribution maintains model transparency and targets areas for incremental improvement, potentially avoiding the expense of a complete reconstruction in the future.
Measuring individual asset contribution is the entry point for understanding the most important decisions
made by a model vs. its benchmark over time.
Back-tested statistics frame the baseline expectation for evaluating a models real-time performance.
We perform an attribution analysis on NDRs Global Balanced Account Model, a stock-bond-cash tactical asset
allocation model with 31 months of real-time history.
The Balanced Account Models real-time performance is, overall, in-line with the expectations set by the backtest. We identify and closely examine a four-month underperforming period in the summer of 2013.
We decompose the Global Balanced Account Model into its four sub-models and perform attribution on each.
Three of the four sub-models exhibited real-time performance in-line with back-test statistics.
Our objective analysis combined with subjective interpretation flags three indicators for potential revision. Of
three indicator alternatives tested, we select two for potential inclusion in the model at a later date.
Changes to the Global Balanced Account Model are not recommended at this time, but we will follow our
two indicator alternatives in tandem with their counterparts and perform a comparative analysis in six to 12
months.
While Eugene Fama is often thought of as the Godfather of factor returns decomposition in modern portfolio theory, it was a seemingly
innocuous article in the Financial Analysts Journal in the summer of 1986 by Brinson, Hood and Beebower that spurred popular interest
in attributing portfolio performance to specific elements. Nearly 30 years later, Determinants of Portfolio Performance continues to
inspire financial analysts to better understand the forces driving portfolio returns. In this vein, we apply our own attribution analysis to
an NDR flagship asset allocation model.
Quantitative timing models can be highly effective tools
for building a disciplined understanding of the asset allocation
landscape. Models make objective, data-driven decisions; they
are immune to the influence that dramatic market moves have
on investor behavior. A models performance history provides
insight into the risk and return characteristics of different
investment regimes. And when proactively monitored and
maintained, quantitative timing models can provide the asset
manager with years of valuable decision-making guidance.
750
500
Model
Model
Benchmark
Benchmark
Gain/Annum
750
8.9%
6.5%
250
250
Back-Test
1990
1992
1994
1996
Model Equity %
2008
2010
2012
2015-05-31 = 55.7%
RealTime
2014
100
75
75
50
50
25
25
Model Bond %
2015-05-31 = 34.8%
100
75
75
50
50
25
25
25
Model Cash %
2015-05-31 = 9.6%
25
20
20
15
15
10
10
NSQ2015Q1_01
500
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
The construction of this model started with the evaluation of individual indicators to gauge market risk consistently throughout
time. Indicators passing a rigorous initial screening process were then combined into equal-weighted diffusion indices and re-evaluated
in terms of their aggregate explanatory power, with special attention paid to avoid high correlation between individual indicators.2
In all, 22 indicators were selected for final inclusion in the model, organized in a hierarchical structure as follows:
Global Balanced
Account Model
Stock/Bond
Relative Model
Stock/Bond
Internal Model
(5 Indicators)
Bond/Cash
Relative Model
Stock/Bond
External Model
(6 Indicators)
Bond/Cash
Internal Model
(5 Indicators)
Bond/Cash
External Model
(6 Indicators)
The overall model consists of two individual relative models: a stock-bond component, evaluating the relative attractiveness
of stocks vs. bonds, and a bond-cash component, evaluating the relative attractiveness of bonds vs. cash. Each of these
components, in turn, is an equal-weighted composition of 2 diffusion-based sub-models: one internal (tape-based) model
composed of five indicators and one external (valuation, sentiment, macroeconomic) model composed of six indicators for
each of the stock/bond and bond/cash pairs. A Black-Litterman optimizer determines the periodic weight recommendations
utilizing the end-of-month model readings along with the asset covariance matrix and risk tolerance parameters.
The Global Balanced Account Model is an ideal candidate for this case study for the following reasons:
Simplicity: The model consists of just three assets (stocks, bonds, and cash) and recommends tactical positions relative
to static benchmark weights. In addition, the model is long-only; the recommended weights must always sum to 100%; it
rebalances at the end of every month; and no out-of-benchmark assets are allowed.
Available History: No changes have been made to the model or its components since it launched on October 31, 2012.
Thus, it offers 31 months of real-time performance to compare with more than 22 years of back-tested history.
Organizational Structure: The hierarchy of the model, as outlined above, naturally lends itself to component isolation and
re-evaluation at any level.
Real-Time Excess Return: Thus far, the model has logged a positive track record in its out-of-sample real-time period (discussed in the next section). Therefore, we are able to invest the time to thoughtfully evaluate the components and test ideas
for future enhancements rather than introducing hasty fixes to address a situation of lagging returns.
2. Since this paper focuses on model performance analysis, the details of indicator development on both a stand-alone basis and in combination will not be covered. A
complete explanation of NDRs model building process is covered in Chapter 2 of Ned Davis book, Being Right or Making Money (Third Edition).
Please see important disclosures at the end of this report.
Equity Line
Model
Model
Benchmark
Benchmark
Gain/Annum
130
11.6%
10.9%
125
120
120
115
115
110
110
105
105
100
100
Benchmark Based On 55% Stocks, 35% Bonds, & 10% Cash
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
100
Model Equity %
2015-05-31 = 55.7%
100
75
75
50
50
25
25
100
Model Bond %
2015-05-31 = 34.8%
100
75
75
50
50
25
25
25
Model Cash %
2015-05-31 = 9.6%
25
20
20
15
15
10
10
NSQ2015Q1_02
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
www.ndr.com | Periodical | Issue #CRS201506251
4
!
!!!
! ! !
MODEL EXCESS RETURN AND INDIVIDUAL ASSET CONTRIBUTION FOR FEBRUARY 2015
Asset
Benchmark
Weight (%)
Overweight vs.
Benchmark (%)
Asset
Contribution (%)
Stocks
60.75
55
5.75
5.62
0.323
Bonds
33.9
35
-1.1
-0.591
0.007
Cash
5.35
10
-4.65
0.33%
T_NSQ201506251.1
3. This is the Market Allocation Contribution component of the attribution formula introduced in Study Session 17 of the CFA Level III curriculum.
Please see important disclosures at the end of this report.
Monthly Excess Return of GBA Model and Assets During Back-Test History
Average: 0.19%
Std. deviation: 0.68%
Risk-adjusted ratio: 0.29
5
4
3
2
1
0
-1
-2
+2 SD
+1 SD
Mean
-1 SD
-2 SD
5
4
3
2
1
0
-1
-2
Average: 0.17%
Std. deviation: 0.66%
Risk-adjusted ratio: 0.26
Average: 0.03%
Std. deviation: 0.15%
Risk-adjusted ratio: 0.2
Average: -0.01%
Std. deviation: 0.01%
Risk-adjusted ratio: -0.47
5
4
3
2
1
0
-1
-2
+2 SD
+1 SD
Mean
-1 SD
-2 SD
1.25
1.00
0.75
0.50
0.25
0.00
-0.25
-0.50
+2 SD
+1 SD
Mean
-1 SD
-2 SD
0.15
0.10
1.25
1.00
0.75
0.50
0.25
0.00
-0.25
-0.50
0.15
0.10
0.05
0.00
-0.05
0.05
+2 SD
+1 SD
Mean
-1 SD
-2 SD
0.00
-0.05
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
NSQ2015Q1_03
Please see important disclosures at the end of this report.
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers
refer to www.ndr.com/vendorinfo/
|
|
www.ndr.com
Periodical
Issue #CRS201506251
Overall, the
performance is
+1 SD
0.5
0.0
-0.5
0.5
Mean
0.0
-0.5
-1 SD
1.5
1.0
+1 SD
0.5
0.0
-0.5
0.5
Mean
0.4
0.2
0.0
-0.5
-1 SD
0.3
real-time model
in-line with our
+1 SD
0.1
0.0
-0.1
0.1
Mean
0.0
-0.1
-1 SD
-0.2
0.05
0.04
0.03
-0.2
Average: 0.0% 0.05
Std. deviation: 0.0% 0.04
Risk-adjusted ratio: -0.96
0.03
0.02
0.01
0.00
-0.01
-0.02
0.02
0.01
+1 SD
0.00
-0.01
Mean
-0.02
-1 SD
Dec
Feb Apr
2013
Jun
NSQ2015Q1_04
Aug
Oct
Dec
Feb Apr
2014
Jun
Aug
Oct
Dec
Feb Apr
2015
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimerwww.ndr.com
at www.ndr.com/copyright.html
For data vendor disclaimers
refer to#CRS201506251
www.ndr.com/vendorinfo/
| Periodical
| Issue
Model
Benchmark
130
125
120
115
115
110
110
105
105
100
Title
Gain/Annum
Model
11.9%
Benchmark 10.5%
95
Std Dev
4.7%
4.0%
Downside Dev
2.5%
2.5%
Bat Avg
71.0%
Sharpe
2.54
2.62
Info Ratio
1.05
Tracking Err
1.3%
100
Max Drawdown
-2.0% (2013-05-31..2013-06-29)
-2.1% (2013-04-30..2013-06-29)
95
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
100
Model Equity %
2015-05-29 = 51.7%
100
75
75
50
50
25
25
1.25
1.00
0.75
0.50
0.25
0.00
-0.25
-0.50
-0.75
-1.00
Stock Contribution
Bond Contribution
1.25
1.00
0.75
0.50
0.25
0.00
-0.25
-0.50
-0.75
-1.00
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR
Disclaimer
www.ndr.com/copyright.html
For data
vendor
disclaimersbe
refer
to www.ndr.com/vendorinfo/
5. A 50/50 benchmark enables us to directly map the model scores to asset
weights
andatremove
any performance that
would
otherwise
attributed
to Black/Litterman
optimization. In this scenario, a neutral score of 50 perfectly maps to a 50/50 model allocation with excess return of zero. The higher the score above 50, the higher (lower)
the model's allocation to stocks (bonds); the lower the score below 50, the lower (higher) the model's allocation to stocks (bonds).
Model
Benchmark
135
130
125
120
120
115
115
110
110
105
105
100
Title
Gain/Annum
Model
13.7%
Benchmark 10.5%
95
Std Dev
5.5%
4.0%
Downside Dev
2.4%
2.5%
Bat Avg
71.0%
Sharpe
2.50
2.62
Info Ratio
1.49
Tracking Err
2.1%
100
Max Drawdown
-2.1% (2013-05-31..2013-06-29)
-2.1% (2013-04-30..2013-06-29)
95
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
100
Model Equity %
2015-05-29 = 70.0%
100
75
75
50
50
25
25
3.0
Stock Contribution
Bond Contribution
2.5
3.0
2.5
2.0
2.0
1.5
1.5
1.0
1.0
0.5
0.5
0.0
0.0
-0.5
-0.5
-1.0
-1.0
-1.5
-1.5
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers
refer to www.ndr.com/vendorinfo/
|
|
www.ndr.com
Periodical
Issue #CRS201506251
Model
Benchmark
125
120
115
115
110
110
105
105
100
Title
Gain/Annum
Std Dev
Downside Dev
Bat Avg
Sharpe
Info Ratio
Tracking Err
10.1%
4.1%
2.6%
51.6%
2.44
-0.29
1.5%
Benchmark 10.5%
4.0%
2.5%
Model
95
2.62
100
Max Drawdown
-2.0% (2013-04-30..2013-06-29)
-2.1% (2013-04-30..2013-06-29)
95
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
100
Model Equity %
2015-05-29 = 33.3%
100
75
75
50
50
25
25
Stock Contribution
Bond Contribution
1.0
0.5
0.5
0.0
0.0
-0.5
-0.5
-1.0
-1.0
-1.5
-1.5
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer
data vendor disclaimers
refer to www.ndr.com/vendorinfo/
| Periodical
| Issue #CRS201506251
www.ndr.com
10 at www.ndr.com/copyright.html For
Sharpe
Ratio
% Months
Positive
Excess
Return
Number of
negative
outlier
events
140 bps
2.54
71
320 bps
2.50
71
-40 bps
2.44
51.6
13
Excess Gain/
Annum vs.
Benchmark
Stock/Bond Top-Level
(50% Internal 50% External)
Model
T_NSQ201506251.2
Sharpe
Ratio
% Months
Positive
Excess
Return
Number of
negative
outlier
events
120 bps
1.4
64.5
130 bps
1.24
64.5
120 bps
1.58
67.7
Excess Gain/
Annum vs.
Benchmark
Bond/Cash Top-Level
(50% Internal 50% External)
Model
T_NSQ201506251.3
11
108
105
105
102
102
Model Bond %
2013-12-31 = 33.3%
75
75
50
50
25
25
2013-12-31 = 66.7%
Model Equity %
Bonds
Stocks
Stocks
Bonds
Stocks
Stocks
Stocks
Stocks
Bonds
Bonds
Neutral
Bonds
Bonds
Bonds
Bonds
Bonds
Stocks
Bonds
Stocks
Bonds
Bonds
Bonds
Bonds
Bonds
Stocks
Stocks
Stocks
Bonds
Bonds
Bonds
Bonds
Bonds
Bonds
Bonds
Bonds
Bonds
Stocks
Stocks
Stocks
Stocks
Stocks
Stocks
Stocks
Stocks
Stocks
Stocks
Stocks
Bonds
Bonds
Bonds
Bonds
Bonds
Stocks
Stocks
External Model
50.0
50.0
41.7
16.7
33.3
33.3
50.0
66.7
66.7
Internal Model
60.0
60.0
50.0
50.0
60.0
60.0
70.0
70.0
80.0
Total Composite
55.0
55.0
45.8
33.3
46.7
46.7
60.0
68.3
73.3
Mar
Apr
Jun
Jul
Aug
Sep
Oct
Nov
Jan
2013
Feb
NSQ2015Q1_08
May
Dec
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
12
Global Stock Market Sentiment favored bonds throughout the year, with the exception of a neutral reading posted at the end of May.
Relative Strength of the global Staples, Health Care,
Utilities, and Telecom (SHUT) sectors vs. the broad market changed to favor bonds over stocks at the end of April
and held this position throughout the summer until the
end of September, when it returned to favor stocks over
bonds.
The chart below shows the U.S. Market Sentiment standalone model during the real-time period from October 31,
2012 through May 31, 2015. The middle clip displays the U.S.
market sentiment time series along with standard deviation
brackets that generate the signals6 and is shaded blue when
the indicator favors stocks, black for bonds, and grey for
neutral.
The Equity Risk Premium indicator favored bonds the entire time.
Global Central Bank Policy favored stocks the entire time.
Model
Benchmark
125
120
115
115
110
110
105
105
100
Title
Gain/Annum
Model
10.1%
Benchmark 10.5%
95
Std Dev
6.1%
4.0%
Downside Dev
3.8%
2.5%
Bat Avg
45.2%
Sharpe
1.65
2.62
Info Ratio
-0.10
Tracking Err
3.9%
Max Drawdown
-4.2% (2013-05-31..2013-08-31)
-2.1% (2013-04-30..2013-06-29)
100
95
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
75
75
70
65
70
65
60
60
55
55
50
45
3
50
Model is 100% stock when 6-month smoothing crosses below lower bracket and reverses.
Stock Contribution
Bond Contribution
Model is 100% bond when 6-month smoothing crosses above upper bracket and reverses.
45
3
2
1
-1
-1
-2
-2
-3
-3
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
6. Discussion of signal generation techniques for specific indicators will only be covered on an as-needed basis so as not to drift too far off topic.
Please see important disclosures at the end of this report.
13
Max number
of consecutive
months
missing median
excess return
Number
of signal
changes
Excess Gain/
Annum vs.
Benchmark
Sharpe Ratio
% Months
Positive Excess
Return
-40 bps
1.66
46.7
14
-670 bps
1.28
20
160 bps
2.01
63.3
-710 bps
1.36
30
730 bps
2.38
70
Global PMI
260 bps
2.11
60
Model
T_NSQ201506251.4
14
Model
Benchmark
125
120
115
115
110
110
105
105
100
Title
Model
95
Gain/Annum
8.5%
Benchmark 10.5%
Std Dev
Downside Dev
5.2%
2.7%
4.0%
2.5%
Bat Avg
35.5%
Sharpe
1.62
Info Ratio
-0.58
2.62
Tracking Err
3.5%
Max Drawdown
-4.2% (2013-05-31..2013-08-31)
-2.1% (2013-04-30..2013-06-29)
100
95
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
U.S. Stock Market Sentiment Composite 6-Month Smoothing
+1.25 Rolling 5-Week Standard Deviation
-2.0Rolling 5-Week Standard Deviation
70
65
70
65
60
60
55
55
50
45
3
50
Model is 100% stock when 6-month smoothing crosses below lower bracket and reverses.
Stock Contribution
Bond Contribution
Model is 100% bond when 6-month smoothing crosses above upper bracket and reverses.
45
3
2
1
-1
-1
-2
-2
-3
-3
Monthly Excess Return of Stock/Bond Model for U.S. Market Sentiment (New)
NSQ2015Q1_10
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
15
Although the revised indicator generated improved backtested results over its history from 1981 (not shown), the
performance from October 31, 2012 through May 31, 2015
actually degraded to -200 bps/annum of excess return with 17
allocation changes, contrary to what we anticipated. Given the
indicators adequate performance in its original form, we dont
see an immediate need to test more radical adjustments at this
time. We will continue to watch the indicator going forward and
revisit it on an as-needed basis.
Model
Benchmark
125
120
115
115
110
110
105
105
100
Title
Gain/Annum
Model
3.8%
Benchmark 10.5%
95
Std Dev
3.1%
4.0%
Downside Dev
2.1%
2.5%
Bat Avg
19.4%
Sharpe
1.20
2.62
Info Ratio
-1.95
Tracking Err
3.4%
Max Drawdown
-3.5% (2013-04-30..2013-06-29)
-2.1% (2013-04-30..2013-06-29)
100
95
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
Global Stock Market Sentiment Composite Deviation from Trend Line(5-Week/32-Week Smoothing)
1.4
1.4
1.2
1.2
1.0
1.0
0.8
0.8
0.6
0.4
3
0.6
Model is 100% stocks when deviation from trend line crosses below 0.5 and reverses.
Stock Contribution
Bond Contribution
Model is 100% bonds when deviation from trend line crosses above 1.25 and reverses.
0.4
3
2
-1
-1
-2
-2
-3
-3
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
16
Model
Benchmark
125
120
115
115
110
110
105
105
100
100
Title
95
Gain/Annum
Model
9.6%
Benchmark 10.5%
Std Dev
Downside Dev
Bat Avg
Sharpe
Info Ratio
Tracking Err
6.6%
4.0%
3.2%
2.5%
45.2%
1.44
2.62
-0.21
4.3%
Max Drawdown
-4.2% (2013-04-30..2013-08-31)
-2.1% (2013-04-30..2013-06-29)
95
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
2.5
Global Stock Market Sentiment Composite Deviation from Trend Line(5-Week/32-Week Smoothing)
+0.75 Rolling 1-Year Standard Deviation 2015-06-19 = 1.166
-1.75 Rolling 1-Year Standard Deviation 2015-06-19 = 0.557
2.0
2.5
2.0
1.5
1.5
1.0
1.0
0.5
0.5
Model is 100% stocks when deviation from trend line crosses below -1.75 and reverses.
Stock Contribution
Bond Contribution
Model is 100% bonds when deviation from trend line crosses above 0.75 and reverses.
4
3
-1
-1
-2
-2
-3
-3
Monthly Excess Return of Stock/Bond Model for Global Market Sentiment (New)
NSQ2015Q1_12
Please see important disclosures at the end of this report.
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers
refer to www.ndr.com/vendorinfo/
www.ndr.com | Periodical | Issue #CRS201506251
17
Model
Benchmark
130
125
120
120
115
115
110
110
105
105
The SHUT Index is composed of the Consumer Staples, Health Care, Utilities, and Telecom Sectors.
This largely defensive index helps distinguish market leadership as being either cyclical or defensive.
100
Title
Gain/Annum
Std Dev
Downside Dev
Bat Avg
Sharpe
Info Ratio
Tracking Err
12.1%
5.9%
3.6%
64.5%
2.04
0.37
4.4%
Benchmark 10.5%
4.0%
2.5%
Model
95
2.62
Max Drawdown
-3.4% (2013-12-31..2014-02-01)
-2.1% (2013-04-30..2013-06-29)
100
95
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
140
140
135
135
130
130
125
125
Model is 100% stock when 15-day smoothing crosses below 120-day smoothing by 1.5%
4
3
Stock Contribution
Bond Contribution
Model is 100% bond when 15-day smoothing crosses above 120-day smoothing by 1.5%.
4
3
2
-1
-1
-2
-2
-3
-3
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers
refer to www.ndr.com/vendorinfo/
www.ndr.com | Periodical | Issue #CRS201506251
18
135
Model
Benchmark
130
125
120
120
115
115
110
110
105
100
Title
The SHUT Index is composed of the Consumer Staples, Health Care, Utilities, and Telecom Sectors.
This largely defensive index helps distinguish market leadership as being either cyclical or defensive.
Gain/Annum
Std Dev
Downside Dev
Bat Avg
Sharpe
Info Ratio
Tracking Err
Max Drawdown
Model
12.7%
Benchmark 10.5%
95
5.3%
4.0%
1.6%
2.5%
58.1%
2.37
2.62
0.59
3.8%
-3.5% (2013-04-30..2013-08-31)
-2.1% (2013-04-30..2013-06-29)
105
100
95
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
140
140
135
135
130
130
125
125
Model is 100% stock when 15-day smoothing crosses below 120-day smoothing by -0.25%
4
3
Stock Contribution
Bond Contribution
Model is 100% bond when 15-day smoothing crosses above 120-day smoothing by 0.25%
4
3
2
-1
-1
-2
-2
-3
-3
Monthly Excess Return of Stock/Bond Model for SHUT Market Index (New)
NSQ2015Q1_14
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
CONCLUSION
In the preceding pages, we have applied an objective, realtime attribution framework to an actual Ned Davis Research
asset allocation model, identifying the successful components
as well as those in need of increased attention. We performed
a full decomposition of the model and evaluated three ideas
for improvement at the individual indicator level.
19
Stat
125
125
0.195
median
0.053
115
stDev
0.683
115
110
min
-2.561
4.757
110
105
max
# obs
273
105
skewness
2.884
kurtosis
19.027
100
120
100
95
Frequency
Value
mean
120
95
90
90
85
85
80
80
75
75
70
70
65
65
60
55
50
60
55
50
45
45
40
40
35
35
30
30
25
25
20
20
15
15
10
10
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
Excess Return(%)
NSQ2015Q1_15
Please see important disclosures at the end of this report.
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers
refer to www.ndr.com/vendorinfo/
| Issue #CRS201506251
www.ndr.com | Periodical
20
Number of
Consecutive Months
2
3
4
5
6
7
8
9
10
11
12
12
9
6
1
1
1
2
0
0
0
0
16
8
1
0
0
0
0
0
0
0
0
T_NSQ201506251.5
Monthly Data 2009-10-30 to 2010-05-28 (Log Scale)
107
Model
Benchmark
106
106
105
105
104
104
103
103
102
102
101
101
100
100
Title
99
Return %
Std Dev
Downside Dev
Bat Avg
Tracking Err
2.4%
2.5%
1.9%
14.3%
0.96
-3.8%
Benchmark 3.2%
2.5%
1.5%
Model
Oct
0.3
0.2
0.1
0.0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
-0.7
Nov
Stock Contribution
Dec
Cash Contribution
Max Drawdown
99
-4.4% (2010-04-30..2010-05-29)
1.27
Jan
2010
Bond Contribution
0.21
-3.8% (2010-04-30..2010-05-29)
Feb
Mar
Apr
Negative Model Excess Returns
May
0.3
0.2
0.1
0.0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
-0.7
21
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
The model also logged one sevenmonth and one six-month instance
during the back-test history when it
failed to achieve the median excess
return. These time periods, October 2001
April 2002 and March 2004 August
2004, are presented in more detail in the
charts at right and below.
Model
Benchmark
All performance numbers do not include transaction costs.
104
103
103
102
102
101
101
100
100
Title
Return %
Model
1.5%
Benchmark 2.7%
99
Oct
Std Dev
Downside Dev
Bat Avg
Sharpe
Info Ratio
Tracking Err
1.1%
1.5%
0.9%
0.9%
33.3%
0.55
1.26
0.36
-3.5%
Nov
Stock Contribution
0.50
Dec
Bond Contribution
0.25
104
Jan
2002
Cash Contribution
Max Drawdown
-2.1% (2002-03-28..2002-05-01)
-2.0% (2002-03-28..2002-05-01)
Feb
99
Mar
Apr
0.50
0.25
0.00
0.00
-0.25
-0.25
-0.50
-0.50
-0.75
-0.75
-1.00
-1.00
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer atMonthly
www.ndr.com/copyright.html
For data vendorto
disclaimers
refer to www.ndr.com/vendorinfo/
Data 2004-03-31
2004-08-31
(Log Scale)
100
99
99
Return %
Std Dev
Downside Dev
Model
-0.6%
0.6%
0.7%
Benchmark
-0.3%
0.5%
0.6%
Apr
0.3
0.2
Stock Contribution
Bat Avg
20.0%
Sharpe
-1.80
May
Bond Contribution
Tracking Err
-2.7%
Cash Contribution
Max Drawdown
-1.5% (2004-06-30..2004-07-31)
-1.3% (2004-06-30..2004-07-31)
Jun
Positive Model Excess Returns
0.1
Info Ratio
0.10
-1.47
Jul
Negative Model Excess Returns
Aug
0.3
0.2
0.1
0.0
0.0
-0.1
-0.1
-0.2
-0.2
-0.3
-0.3
-0.4
-0.4
-0.5
-0.5
Negative outlier excess return
-0.6
-0.6
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
22
While the distribution of back-tested excess returns is positively skewed, the distribution of real-time excess returns is
negatively skewed (0.06% mean return vs. 0.074% median
return), due to the magnitude of the negative outlier events.
Stat
Number of Observations: 31
Number of Observations greater than 1 S.D.: 4 (12.90%)
Number of Observations less than 1 S.D.: 4 (12.90%)
Number of Observations within 1 S.D.: 23 (74.19%)
8.00
7.75
7.50
7.25
7.00
6.75
Frequency
6.50
Value
8.25
mean
0.06
median
0.074
stDev
0.19
7.75
min
-0.409
7.50
max
0.452
7.25
# obs
31
skewness
-0.673
7.00
kurtosis
3.826
8.00
6.75
6.50
6.25
6.25
6.00
6.00
5.75
5.75
5.50
5.50
5.25
5.25
5.00
5.00
4.75
4.75
4.50
4.50
4.25
4.25
4.00
4.00
3.75
3.75
3.50
3.50
3.25
3.25
3.00
3.00
2.75
2.75
2.50
2.50
2.25
2.25
2.00
2.00
1.75
1.75
1.50
1.50
1.25
1.25
1.00
1.00
-0.40 -0.35 -0.30 -0.25 -0.20 -0.15 -0.10 -0.05
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
Excess Return(%)
NSQ2015Q1_19
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
www.ndr.com | Periodical | Issue #CRS201506251
23
Number of
Consecutive Months
T_NSQ201506251.6
Monthly Data 2013-05-31 to 2013-09-30 (Log Scale)
101
100
100
99
99
98
98
Title
Return %
Model
1.3%
Benchmark 1.7%
0.2
Downside Dev
0.6%
0.6%
Bat Avg
50.0%
Jun
May
2013
0.3
Std Dev
1.2%
1.4%
Stock Contribution
Bond Contribution
Sharpe
1.06
1.24
Info Ratio
0.21
Tracking Err
-1.8%
Max Drawdown
-1.9% (2013-05-31..2013-06-29)
-2.0% (2013-05-31..2013-06-29)
Jul
Cash Contribution
Aug
0.1
Sep
0.3
0.2
0.1
0.0
0.0
-0.1
-0.1
-0.2
-0.2
-0.3
-0.3
-0.4
-0.4
-0.5
-0.5
-0.6
-0.6
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
24
Stock/Bond External Model With New Global Market Sentiment Indicator Real-Time Performance
130
125
130
Model
Benchmark
125
120
120
115
115
110
110
105
105
100
Title
Gain/Annum
Model
11.1%
Benchmark 10.5%
95
Std Dev
Downside Dev
Bat Avg
Sharpe
Info Ratio
Tracking Err
4.7%
4.0%
3.2%
2.5%
54.8%
2.33
2.62
0.32
1.7%
100
Max Drawdown
-2.6% (2013-12-31..2014-02-01)
-2.1% (2013-04-30..2013-06-29)
95
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
100
Model Equity %
2015-05-29 = 33.3%
100
75
75
50
50
25
25
2.0
1.5
Stock Contribution
Bond Contribution
1.0
2.0
1.5
1.0
0.5
0.5
0.0
0.0
-0.5
-0.5
-1.0
-1.0
-1.5
-1.5
Monthly Excess Return of Stock/Bond External Model With New Global Market Sentiment Indicator
NSQ2015Q1_21
Please see important disclosures at the end of this report.
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
www.ndr.com | Periodical | Issue #CRS201506251
25
125
125
120
120
115
115
110
110
105
105
100
Title
The SHUT Index is composed of the Consumer Staples, Health Care, Utilities, and Telecom Sectors.
This largely defensive index helps distinguish market leadership as being either cyclical or defensive.
Gain/Annum Std Dev Downside Dev Bat Avg Sharpe Info Ratio Tracking Err
Max Drawdown
Model
10.2%
Benchmark 10.5%
95
4.0%
4.0%
2.4%
2.5%
45.2%
2.51
2.62
-0.25
1.4%
100
-2.0% (2013-04-30..2013-06-29)
-2.1% (2013-04-30..2013-06-29)
95
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
100
Model Equity %
2015-05-29 = 41.7%
100
75
75
50
50
25
25
Stock Contribution
Bond Contribution
1.0
1.0
Above median excess return
0.5
0.5
0.0
0.0
-0.5
-0.5
-1.0
-1.0
-1.5
-1.5
Monthly Excess Return of Stock/Bond External Model With New SHUT Indicator
NSQ2015Q1_22
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
Stock/Bond External Model With New Global Sentiment and SHUT Indicators Real-Time Performance
130
125
120
130
Model
Benchmark
125
120
115
115
110
110
105
105
100
Title
The SHUT Index is composed of the Consumer Staples, Health Care, Utilities, and Telecom Sectors.
This largely defensive index helps distinguish market leadership as being either cyclical or defensive.
Gain/Annum
Std Dev
Downside Dev
Bat Avg
Sharpe
Info Ratio
Tracking Err
Max Drawdown
Model
11.2%
Benchmark 10.5%
95
4.6%
4.0%
2.9%
2.5%
51.6%
2.41
2.62
0.43
1.6%
100
-2.2% (2013-12-31..2014-02-01)
-2.1% (2013-04-30..2013-06-29)
95
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
2013
2014
2015
100
Model Equity %
2015-05-29 = 41.7%
100
75
75
50
50
25
25
1.5
Stock Contribution
Bond Contribution
1.0
1.5
1.0
0.5
0.5
0.0
0.0
-0.5
-0.5
-1.0
-1.0
-1.5
-1.5
Monthly Excess Return of Stock/Bond External Model With New Global Sentiment and SHUT Indicators
NSQ2015Q1_23
Copyright 2015 Ned Davis Research, Inc. Further distribution prohibited without prior permission. All Rights Reserved.
See NDR Disclaimer at www.ndr.com/copyright.html For data vendor disclaimers refer to www.ndr.com/vendorinfo/
26
Before joining NDRG, Lisa was a Vice President in Credit Risk Technology
for Goldman Sachs Group, Inc., where she was responsible for quantitative
risk measures of derivative portfolios.
including registered representative, option trader, and most recently as the option
specialist for Schwab Private Client Investment Advisory, where he developed
strategies to generate income and hedge equity risk in client portfolios.
Matt is also a CFA charterholder and a member of the CFA Institute, CFA Tampa
Bay, and the Market Technicians Association.
27
VENICE
Ed Clissold, CFA
U.S. Market Strategist
Joseph Kalish
Chief Global Macro Strategist
BOSTON
50 Federal Street
6th Floor
Boston, MA 02110
(617) 279-4860
ATLANTA
DISCLAIMER
The data and analysis contained herein are provided as is and without warranty of any kind, either
expressed or implied. Ned Davis Research, Inc. (NDR), d.b.a. Ned Davis Research Group (NDRG), any NDRG
affiliates or employees, or any third-party data provider, shall not have any liability for any loss sustained
by anyone who has relied on the information contained in any NDRG publication. NDRG disclaims any
and all express or implied warranties, including, but not limited to, any warranties of merchantability,
SAN FRANCISCO
not necessarily be updated as views or information change. All opinions expressed herein are
8 Bouverie Street
London
EC4Y 8AX
+44 (0)20 7779 8579
communication reects our analysts opinions as of the date of this communication and will
subject to change without notice. NDRG or its affiliated companies or their respective shareholders,
directors, officers and/or employees, may have long or short positions in the securities discussed
herein and may purchase or sell such securities without notice.
Using any graph, chart, formula or other device to assist in deciding which securities to trade or when to trade
them presents many difficulties and their effectiveness has significant limitations, including that prior patterns
may not repeat themselves continuously or on any particular occasion. In addition, market participants using
such devices can impact the market in a way that changes the effectiveness of such device.
Further distribution prohibited without prior permission. For data vendor disclaimers, refer to
www.ndr.com/vendorinfo.
Copyright 2015 (c) Ned Davis Research, Inc. All rights reserved.
With a
we
deliver
Macro
damental
Fun
award-winning
Idea
Te c h nic a l
S e nt
i m e nt
AC H
LONDON
NDRGs past recommendations and model results are not a guarantee of future results. This
RO
50 California Street
Suite 1500
San Francisco, CA 94111
(415) 277-5477
A
360
PP