The Performance of Asset Allocation Strategies Across Datasets and Over Time

Abstract

This paper evaluates the ex-ante performance of popular asset allocation strategies including the classical mean variance rule, the widely used naive diversification and several newly sprouted risk-based techniques. In terms of risk-adjusted return, I find that there are no significant differences between these strategies for country- and industry-based portfolios. For the individual stock portfolio and multi-asset portfolios (stocks and bonds combined), the differences between strategy performances are relatively large, and the superior strategy changes over time. With respect to strategies’ risk loadings, the mean variance rule leads to the most fluctuating exposures in size, value and momentum factors for all four datasets, the other allocation methods are less so. This paper indicates that currently-used allocation strategies are all far from ex-post optimal.