Alpha Go Everywhere: Machine Learning and International Stock Returns

We apply machine learning techniques and use stock characteristics to predict stock returns in 31 markets. We conduct an out-of-sample test by training the models with past U.S. data to predict international stock returns. Neural networks (NNs) and regression trees (RTs) outperform linear models in forming profitable portfolios and predicting return rankings. When the models are trained separately for each market, NNs achieve even stronger results, but RTs underperform linear models when the number of observations is low. Finally, we show that U.S.-based variables can further enhance the return predictability of NNs globally, suggesting that the markets are integrated.

Keywords: Neural Networks, Regression Trees, Overfitting, Cross-section of Stock Returns, International Asset Pricing

JEL Classification: C52, G10, G12, G15

Suggested Citation: Suggested Citation

Choi, Darwin and Jiang, Wenxi and Zhang, Chao, Alpha Go Everywhere: Machine Learning and International Stock Returns (May 18, 2022). Available at SSRN: https://ssrn.com/abstract=3489679 or http://dx.doi.org/10.2139/ssrn.3489679