Any Role for Mean Reversion in Short Term Asset Allocation?

Abstract

This paper investigates whether improving the estimation of the expected returns from simple historical moments to the use of predictable variables, mean reversion or both, mean-variance optimal portfolio strategies are able to perform statistically better than the 1/N portfolio. Our analysis shows that, even for the short term horizon, exploiting the mean reversion and predictability feature present in some of the datasets allow mean variance strategies to perform statistically better than the 1/N strategy.

Keywords: mean reversion, strategy preference, 1/N, predictability, testing Sharpe equivalence

JEL Classification: G11