Does It Pay to Follow Anomalies Research? Machine Learning Approach with International Evidence

Abstract

We study out-of-sample returns on 153 anomalies in equities documented in the academic literature. We show that machine learning techniques that aggregate all the anomalies into one mispricing signal are profitable around the globe and survive on a liquid universe of stocks. We investigate the value of international evidence for selection of quantitative strategies that outperform out-of-sample. Past performance of quantitative strategies in regions other than the United States does not help to pick out-of-sample winning strategies in the U.S. Past evidence from the U.S., however, captures most of the return predictability outside the U.S.

Keywords: Anomalies, International Finance, Machine Learning, Neural Network, Random Forest.

JEL Classification: G11, G12, G15