Diverging Roads: Theory-Based vs. Machine Learning-Implied Stock Risk Premia

Abstract

We assess financial theory-based and machine learning methods to quantify stock risk premia and investigate the potential of hybrid strategies. The results indicate that at the one-month investment horizon, a theory-based approach using option prices is preferable, especially if risk premium estimates get updated at high frequencies. At the one-year horizon, a random forest with sufficiently long training delivers a better performance than option-based models. The integration of machine learning procedures to address the approximation errors of a theory-based approach is identified as a novel and promising hybrid strategy.

Keywords: stock risk premia, option prices, machine learning

JEL Classification: C53, C58, G12, G17