Implementable tail risk management: An empirical analysis of CVaR-optimized carry trade portfolios

Abstract

Although it is relatively easy to identify limitations of the mean-variance framework in managing tail risk, offering a coherent, implementable alternative is more difficult. The goal of this article is to propose an implementable solution to the significant puzzle of portfolio construction in non-Gaussian markets, that is, in markets with more rare events than expected in a mean-variance framework. In this article, we explain how we improved on traditional risk management approaches with heavy-tailed distributions tailored to take into account extreme comovements. Our findings show that this new quantitative asset allocation method, with non-Gaussian dynamic risk models, leads to enhanced downside protection without constraining upside potentials. Finally, we conduct out-of-sample tests to demonstrate these risk control capabilities on a currency carry portfolio allocation example. © 2011 Macmillan Publishers Ltd.