How Optimal are Risk-based Portfolios?

Abstract

Since the early 2000s, interest in risk-based rules has grown as the asset management industry adopted these allocation methods to maximize diversification benefits, with the hope of achieving better out-of-sample performance compared to traditional methodologies (e.g., mean-variance). The theoretical optimality of risk-based portfolios is often an afterthought. In this article, we suggest that perhaps their theoretical optimality warrants increased attention especially given their common acceptance. We provide the necessary and sufficient conditions for a class of risk-based rules (e.g., inverse volatility portfolios) to be considered theoretically optimal in a mean-variance framework, and we show that such conditions are met empirically using many common datasets. This finding supports the empirical observation that these portfolios outperform out-of-sample both mean-variance and equally weighted portfolios. Our results suggest that risk-based rules have merits from the perspective of optimality, beyond their robust characteristics.

Keywords: Portfolio Choice, Asset Allocation

JEL Classification: G11