Extreme Correlations and Optimizing for Stress

Abstract

Asset correlations often change significantly during periods of financial market dislocation. This article describes a rigorous way of deriving a covariance matrix for use in stressed market environments. This lets investors apply standard mean–variance techniques to analyze investment strategy under extreme conditions. Applications include risk management and constructing optimally defensive portfolios for stressed markets. The method is based on recent advances in multivariate extreme-value theory.