Risk-Averse Action Selection Using Extreme Value Theory Estimates of the CVaR

The Conditional Value-at-Risk (CVaR) is a useful risk measure in machine learning, finance, insurance, energy, etc. When the CVaR confidence parameter is very high, estimation by sample averaging exhibits high variance due to the limited number of samples above the corresponding threshold. To mitigate this problem, we present an estimation procedure for the CVaR that combines extreme value theory and a recently introduced method of automated threshold selection by Bader et al. (2018). Under appropriate conditions, we estimate the tail risk using a generalized Pareto distribution. We compare empirically this estimation procedure with the naive method of sample averaging, and show an improvement in accuracy for some specific cases. We also show how the estimation procedure can be used in reinforcement learning by applying our method to the multi-armed bandit problem where the goal is to avoid catastrophic risk.

[1]  S. Coles,et al.  An Introduction to Statistical Modeling of Extreme Values , 2001 .

[2]  Vartan Choulakian,et al.  Goodness-of-Fit Tests for the Generalized Pareto Distribution , 2001, Technometrics.

[3]  Bruno Rémillard,et al.  Statistical Methods for Financial Engineering , 2013 .

[4]  A. Frigessi,et al.  A Dynamic Mixture Model for Unsupervised Tail Estimation without Threshold Selection , 2002 .

[5]  C. Guedes Soares,et al.  An Application of the Peaks Over Threshold Method to Predict Extremes of Significant Wave Height , 1998 .

[6]  S. Uryasev,et al.  Calculating CVaR and bPOE for common probability distributions with application to portfolio optimization and density estimation , 2018, Annals of Operations Research.

[7]  R. Rockafellar,et al.  Conditional Value-at-Risk for General Loss Distributions , 2001 .

[8]  Eric P. Smith,et al.  An Introduction to Statistical Modeling of Extreme Values , 2002, Technometrics.

[9]  A. McNeil.  Extreme Value Theory for Risk Managers , 1999 .

[10]  Jun Yan,et al.  Automated threshold selection for extreme value analysis via ordered goodness-of-fit tests with adjustment for false discovery rate , 2018 .

[11]  Carl Scarrott,et al.  A Review of Extreme Value Threshold Estimation and Uncertainty Quantification , 2012 .

[12]  Sergio M. Vicente-Serrano,et al.  Mapping the hazard of extreme rainfall by peaks-over-threshold extreme value analysis and spatial regression techniques , 2006 .

[13]  Richard S. Sutton,et al.  Introduction to Reinforcement Learning , 1998 .

[14]  Christine M. Anderson-Cook,et al.  Book review: quantitative risk management: concepts, techniques and tools, revised edition, by A.F. McNeil, R. Frey and P. Embrechts. Princeton University Press, 2015, ISBN 978-0-691-16627-8, xix + 700 pp. , 2017, Extremes.

[15]  Alessandro Lazaric,et al.  Risk-Aversion in Multi-armed Bandits , 2012, NIPS.

[16]  L. Haan,et al.  Residual Life Time at Great Age , 1974 .

[17]  D. Tasche,et al.  On the coherence of expected shortfall , 2001, cond-mat/0104295.

[18]  R. Tibshirani,et al.  Sequential selection procedures and false discovery rate control , 2013, 1309.5352.

[19]  Weihu Cheng,et al.  Extreme tail risk estimation with the generalized Pareto distribution under the peaks-over-threshold framework , 2018, Communications in Statistics - Theory and Methods.

[20]  E. Simiu,et al.  Extreme Wind Distribution Tails: A “Peaks over Threshold” Approach , 1996 .

[21]  Sergey Sarykalin,et al.  Value-at-Risk vs. Conditional Value-at-Risk in Risk Management and Optimization , 2008 .

[22]  J. Pickands.  Statistical Inference Using Extreme Order Statistics , 1975 .

[23]  Jia Yuan Yu,et al.  Sample Complexity of Risk-Averse Bandit-Arm Selection , 2013, IJCAI.