Beruflich Dokumente
Kultur Dokumente
Damien Challet
CentraleSupelec Paris
Encelade Capital SA
Damien Challet
Inefficient Markets
Summary
George Soros
Damien Challet
Inefficient Markets
Lecture structure
Market deconstruction
Market reconstruction
Profit!?
Damien Challet
Inefficient Markets
Why deconstruction?
Engineering failure
Damien Challet
Inefficient Markets
Engineering failures
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Philadelphia, 2015
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
www.tech.plym.ac.uk/sme/interactive resources/tutorials/FailureCases/sf2.html
Damien Challet
Inefficient Markets
Result:
(note: wrong way to plot)
source:
http://www.boeing.com/resources/boeingdotcom/company/about bca/pdf/statsum.pdf
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Part 1: Deconstruction
Myth 1: equilibrium
Wild price fluctuations:
dangerous news, unstable markets, or crazy traders?
1
structure+noise
news distribution
fixed-point?
non-linear dynamics, chaos
Damien Challet
Inefficient Markets
Part 1: Deconstruction
Myth 2: price efficiency
Event studies
Strategy backtest
Damien Challet
Inefficient Markets
And now ?
Damien Challet
Inefficient Markets
Part 2:
Reconstruction
Aim: to build a model of financial market dynamics
Building blocks:
1
Trader
actions (buy, sell, do nothing)
strategies
learning
Exchange rules
Walras
market orders
Damien Challet
Inefficient Markets
Part 2
Reconstruction: Agent-based models
ABMs allow one to study
1
feedback loops
BIG bonus:
1
Predictability dynamics
Trader ecology
EMH?
Calibration
Damien Challet
Inefficient Markets
Part 3:
Profit: strategies and trading
Backtest
HFT
Case studies
Algo trading
Damien Challet
Inefficient Markets
You
data analysis
graphs: reading, plotting
programming
mathematics
paper trading account
take risk
understand fluctuations
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
ABM
Trading strategies
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Reality:
MORE INTERESTING
Damien Challet
Inefficient Markets
Human made
laws
behavioural finance
Computer
Computers are useless. They can only give you answers. (Pablo
Picasso)
To err is human, but to really foul things up you need a
computer. (Paul Ehrlich)
Chaos
Non-linear systems
Unstability
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Crises
Remember
2012, 2011, 2010, 2009, 2008, 2007?
2001,2000 ?
1998, 1997?
1992?
1987?
1971?
1929?
Damien Challet
Inefficient Markets
Realistic view
internal/external fluctuations
cascading reactions
overreaction
instability
e.g. stick balancing, J. L. Cabrera and J. G. Milton, On-off intermittency in a
human balancing task, Phys. Rev. Lett. 89 158702 (2002).
Damien Challet
Inefficient Markets
Information removal:
Equibrium:
Well-behaved: :
fair?
???
only exceptionnaly
60-95%
Damien Challet
Inefficient Markets
Markets: systems
1
Theoretical approach:
1
2
passive
active: poke markets! see them react!
MECHANISMS
reverse-engineering
Engineering:
1
2
Damien Challet
Inefficient Markets
Optimistic hypothesis
At time t 0, price at fundamental price log p = 0 = x
At time t = 0, important news perturbation
Convergence towards x ?
force that pulls back price
force that brakes
Damien Challet
Inefficient Markets
dx
dt
Evolution equation
m
d 2x
= Fspring + Fdamp
dt 2
Damien Challet
Inefficient Markets
3 regimes, =
c
2 mk
critical damping = 1
under-damping < 1
Damien Challet
Inefficient Markets
1.0
1.0
0.6
0.4
cos(x) * exp(3 * x)
0.6
0.4
cos(x) * exp(x)
0.2
0.2
0.0
0.0
0.0
0.5
cos(x) * exp(0.1 * x)
0.5
0.8
0.8
1.0
200
400
600
800
1000
200
Index
400
600
800
1000
Index
Damien Challet
Inefficient Markets
10
20
30
Index
40
50
diff price
40
30
20
10
price
50
60
UBSN.VX
2002
2004
2006
2008
2010
Damien Challet
2002
2004
Inefficient Markets
2006
2008
2010
0.1
0.0
0.1
3.0
0.2
2.5
2.0
log price
3.5
0.2
4.0
UBSN.VX
2002
2004
2006
2008
2010
Damien Challet
2002
2004
Inefficient Markets
2006
2008
2010
0.0
cos(x) * exp(0.1 * x)
0.0
0.5
0.1
0.2
0.1
0.5
0.2
1.0
UBSN.VX
2002
2004
2006
2008
2010
200
400
600
Index
Damien Challet
Inefficient Markets
800
1000
0.5
0.0
0.5
cos(x) * x^0.2
0.0
0.1
0.2
0.1
1.0
0.2
1.5
UBSN.VX
2002
2004
2006
2008
2010
200
400
600
Index
Damien Challet
Inefficient Markets
800
1000
0.0
0.2
cos(x)/log(1 + x * 100)
0.0
0.1
0.2
0.2
0.1
0.4
0.2
UBSN.VX
2002
2004
2006
2008
2010
200
400
600
Index
Damien Challet
Inefficient Markets
800
1000
0.0
0.2
cos(x)/log(1 + x * 100)
0.0
0.1
0.2
0.2
0.1
0.4
0.2
UBSN.VX
2002
2004
2006
2008
2010
2000
4000
6000
Index
Damien Challet
Inefficient Markets
8000
10000
Omori Law
Earth quakes
Intensity of replica decreases as t
Large volatility decreases as t
More precisely
fix a threshold
n(t) = number of times |return| larger than
n(t) t 1
Damien Challet
Inefficient Markets
10
0.500
1cdf
0.005
0.001
Sample Quantiles
0
5
0.050
0.1
0.2
Theoretical Quantiles
0.5
1.0
2.0
Damien Challet
Inefficient Markets
5.0
10.0
2005
2007
Damien Challet
15/09/2008
2010
Inefficient Markets
2005
2007
Damien Challet
15/09/2008
2010
Inefficient Markets
Remember
2007: USA, Spain, UK, Ireland?
1995: UK?
1980s: Switzerland?
1995-2000: dot-com?
1720: South Sea Company?
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Role of Markets
Exchange goods
Fair price
Easily
E.g.: Fish markets, Ebay, NYSE
The more people, the more efficient
Damien Challet
Inefficient Markets
Exchange assets
Fair price
efficiency
Easily
liquidity
small spreads
Damien Challet
Inefficient Markets
Exchange assets
Fair price
efficiency
rational agents
Easily
liquidity
small spreads
many agents
Damien Challet
Inefficient Markets
Price information
Ideal case:
rational agents
no arbitrage
Historicity:
neo-classical Economics
simple mathematics
no computers
little data
Arrogant traders
Damien Challet
Inefficient Markets
Rationale
Markets: complex
Better assume they are perfect
Damien Challet
Inefficient Markets
Rationale
axiomatic
Markets: complex
Better assume they are perfect
axiomatic
axiomatic
axiomatic
Damien Challet
Inefficient Markets
This course
Financial markets:
complex, noisy,
non-linear systems
with feedback
Damien Challet
Inefficient Markets
This course
Market (in-)efficiency
High-frequency
Fluctuations rather than expectations
Endogeneous vs exogeneous
Predictability
ROAD to efficiency
Agent-based models
Mathematics of interaction and heterogeneity
Dynamics of efficiency
Damien Challet
Inefficient Markets
This course
Market (in-)efficiency
High-frequency
Fluctuations rather than expectations
Endogeneous vs exogeneous
mechanistic
Predictability
ROAD to efficiency
Agent-based models
Mathematics of interaction and heterogeneity
Dynamics of efficiency
Damien Challet
Inefficient Markets
mechanistic
Market efficiency
Hypotheses: at time t
Pt = price of asset
Pt = fundamental price
t = all available information
Efficient Market Hypothesis (EMH)
Pt = E (Pt |t )
Current prices incorporate all available information
without bias
Damien Challet
Inefficient Markets
Fair price
Damien Challet
Inefficient Markets
Price evolution
|
Pt+1 = E (Pt+1
t+1 )
Best forecast:
| )
E (Pt+1
t
t+1 ?
= (1 + r )Pt
t+1 N (0, )
Damien Challet
Inefficient Markets
EMH: consequences
Damien Challet
Inefficient Markets
EMH: consequences
useless
Information dynamics:
Information price:
To mathematical finance
Damien Challet
Inefficient Markets
useless
useless
priceless
Market efficiency?
Hypotheses: at time t
Pt = price of asset
Pt = fundamental(?) price
t = all available information
Efficient Market Hypothesis (EMH)
Pt = E (Pt |t )
Current prices incorporate all available information
without bias
Damien Challet
Inefficient Markets
Market efficiency??
Hypotheses: at time t
Pt = price of asset
Pt = fundamental(?) price
t = all(?) available information
Efficient Market Hypothesis (EMH)
Pt = E (Pt |t )
Current prices incorporate all available information
without bias
Damien Challet
Inefficient Markets
Market efficiency???
Hypotheses: at time t
Pt = price of asset
Pt = fundamental(?) price
t = all(?) available(?) information
Efficient Market Hypothesis (EMH)
Pt = E (Pt |t )
Current prices incorporate all available information
without bias
Damien Challet
Inefficient Markets
Market efficiency????
Hypotheses: at time t
Pt = price of asset
Pt = fundamental(?) price
t = all(?) available(?) information
Efficient Market Hypothesis (EMH)
Pt = E (Pt |t )
Current prices incorporate(?) all available information
without bias
Damien Challet
Inefficient Markets
Market efficiency?????
Hypotheses: at time t
Pt = price of asset
Pt = fundamental(?) price
t = all(?) available(?) information
Efficient Market Hypothesis (EMH)
Pt = E (Pt |t )
Current prices incorporate(?) all available information
without bias(?)
Damien Challet
Inefficient Markets
Market efficiency????????
Hypotheses: at time t
Pt = price of asset
Pt = fundamental(?) price
t = all(?) available(?) information
Efficient Market Hypothesis (EMH)
Pt = E (Pt |t )
Current prices incorporate(?) all(?) available(?)
information(?) without bias(?)
Damien Challet
Inefficient Markets
Market efficiency?????????
Hypotheses: at time t
Pt = price of asset
Pt = fundamental(?) price
t = all(?) available(?) information
Efficient Market Hypothesis (EMH)
Pt = E (Pt |t )
Current prices incorporate(?) all(?) available(?)
information(?) without bias(?)
FLUCTUATIONS ?
Damien Challet
Inefficient Markets
Arbitrages
Risk-free
Sell short over-priced asset
Buy under-priced asset with same characteristics
Similar asset?
Risk-free?
Damien Challet
Inefficient Markets
Risk-free? LTCM
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
Shleifer chap 2:
Noise traders
Risk-adverse arbitrageurs
Price does not converge to fundamental price
Damien Challet
Inefficient Markets
Statistical Arbitrages
Damien Challet
Inefficient Markets
Damien Challet
Inefficient Markets
t = : EMH OK
t = {ptk }: momentum/reversal
(taken from M. K. Brunnermeier)
Damien Challet
Inefficient Markets
Event studies
Dividend changes
Split
Merger/Take over
Index inclusion/removal
Important news
Damien Challet
Inefficient Markets
Events
1
N
[ri,t + t rm,t + t ]
i
i=1
Plot
Damien Challet
Inefficient Markets
Take-over
Damien Challet
Inefficient Markets
CEO death
(taken from M. K. Brunnermeier)
Damien Challet
Inefficient Markets
Splits
Fama-French-Fischer-Jensen-Roll (1969)
Damien Challet
Inefficient Markets
Summary
Damien Challet
Inefficient Markets