investment games- hec lausanne finance forum singapore 2015
TRANSCRIPT
Investing in a behavioral world Perspectives of performance measurement and risk
management through Investment Games
By Guillaume Holmberg-PérouxAmbassador HEC Lausanne – swissnex Singapore
Disclaimer : This presentation is my personal view and any mistakes are my own.
If markets were random and rules didn’t change
Simulation of daily Geometric Brownian motion generated prices, variance=3%
If markets were random and rules didn’t change
Simulation of daily Brownian motion generated returns, variance=3%
Tools for this world
• CAPM, Market Beta (performance)• Normal Distributions (risk management)• Black Scholes Option pricing/hedging• Modern Portfolio Theory and efficient frontiers• Value at Risk
Focus on volatility, expected returns, correlation
=> tools of the 20th century
(Almost) Newer tools
• GARCH models (risk management)• Regime switching / Jumps• Non-normal distributions modeling
=> Patches on the old models
Practitioners’ tools
• Fama-French: Size (B/M), Value (P/E) Momentum
• Fundamental analysis (cash flows, comparables, ratios)
• Technical Analysis (head and shoulders, ..)
=> doesn’t tell the whole story
Extracting aggregated expectations
Probability density function that matches aggregated stock market long term expectations and the correspondent risk profile (using monthly NYSE data, since 1963)
New Insights
• Behavior and mismatch => shape risk profile and risk premiums, not the reverse
Problem: assumptions of the investment game itself, not just plain optimism
The right Investment game?
• Non (pure) randomness of human behavior• Chaotic (simple rules but complex results)• Non probabilistic (buy/sell/do nothing)
Non randomness of human networks
Preferential attachement : sharholding, credit networks, financial advice, .. => fat tails
Games on Financial networks : Investment Strategies (simplified)
• New restaurant that opens, with no reservation system, and it’s saturday night.
You go, but it’s crowded (costly) You go but some seats are free (Good? Bad?) You don’t go but some seats are free (Good? Bad?) You don’t go but it’s crowded (good)
Investment Game: Strategic complements and substitutes on information
• Characteristics :
- Herding vs Contrarian- Non-probabilistic views
+ (random) news
Models really well reality.
Non random but chaotic : good news
The system as a whole is partially predictable (long term memory and critical points)
Source : Sornette, 2013
Now what? The new tools for the 21st Century (work in progress)
• Multi-fractal models, Fractional BM, M Cascades• Agent Based Models, Dynamic Systems• Strategic formation, Behavior and Networks• DATA and technology
Volatility is not enough