TEDAS - Tail Event Driven Asset Allocation

Abstract

In this study, we develop a two-step asset allocation strategy that identifies the risk of a benchmark asset and uses multi-moment dynamic portfolio selection to account for possible conditional non-normality of portfolio returns. The TEDAS - Tail Event Asset Allocation strategy is based on the non-positive Lasso adaptive quantile regression method which captures negative tail events for selected benchmark assets. Dynamic conditional multi-moment investor risk and utility measures are introduced and used to perform portfolio selection. This procedure assumes neither joint nor marginal normality of assets' returns and incorporates dynamic multivariate portfolio skewness and kurtosis statistics into portfolio optimization. The TEDAS strategy is tested for major international markets and demonstrates superior performance compared to the market benchmark and naive allocation.