Extreme risk estimation if often associated with the calibration of fat tailed distribution and the sizing of low probability events. Along with Nassim Taleb, we tend to dissociate the estimation of the “size” – or the “damage” – and that of the probability, shedding a reasonable doubt on the estimation, not to say the meaning, of very low probabilities. In this talk, we expose the pitfalls of blind or myopic statistics for the measure of extreme risk and pinpoint the importance of scenario analysis. In one word, narrow-minded statistics tend to undervalue the importance of current available information and its relation with past events, whereas the more extreme the risk, the lesser the weight of statistics and the higher the influence of current specific scenario analysis. We also show the impact of some flawed extreme risk approaches on systemic risk, through the phenomenon of pro-cyclicality. We propose a solution to the question of a global extreme risk measure, the “Stress VaR”, using “poly-model” theory, and analyze its efficiency through 2008 crisis.
Raphael Douady is a French mathematician and economist, specialised in financial mathematics and chaos theory. With more than fifteen years experience in the banking industry (risk management, option models, trading strategies) and thirty years research in pure and applied mathematics, Dr Douady is renowned for his highly sophisticated quantitative solutions and statistical analysis.
A former fellow of Ecole Normale Supérieure in
Raphael Douady is one of the founders and the research director of Riskdata, a market-leading provider of risk management tools for investors, asset managers, hedge funds, fund of funds, and pension funds. He has been appointed as academic director of the French “Laboratory of Excellence” devoted to financial regulation (LabEx ReFi). He is also a member of the Praxis Club, a New York based think tank advising the French government on its economic policy and other related topics and on the “risk committee” of Finance Innovation, a French official entity supporting innovation in financial software.