Design choices, machine learning, and the cross-section of stock returns

Abstract

We fit over one thousand machine learning models for predicting stock returns, systematically varying design choices across algorithm, target variable, feature selection, and training methodology. Our findings demonstrate that the non-standard error in portfolio returns arising from these design choices exceeds the standard error by 59%. Furthermore, we observe a substantial variation in model performance, with monthly mean top-minus-bottom returns ranging from 0.13% to 1.98%. These findings underscore the critical impact of design choices on machine learning predictions, and we offer recommendations for model design. Finally, we identify the conditions under which non-linear models outperform linear models.

Keywords: Machine learning, stock returns, design choices, non-standard errors, portfolio performance

JEL Classification: C52, C55, G11, G17