Research work · 2021

A Research Agenda

Stefan Nagel

Machine Learning in Asset Pricing

Abstract

This chapter focuses on the on cross-section of stock returns in research opportunities and related potential applications in other areas of asset pricing. It begins with empirical applications of machine learning (ML) tools, noting that part of the challenge in empirical work is to sort out how to select and tweak ML methods to be most useful in asset pricing applications. Some progress has been made, but many questions remain about economically appropriate methods of regularization, the role of nonlinearities, and how to deal with structural change in the underlying data-generating process. The chapter discusses areas where application of ML methods seems promising and fall into the area of asset demand analysis. It offers some suggestions on how asset pricing theory could advance by adopting ML as a model of investor belief formation in high-dimensional environments.

DOI: 10.23943/princeton/9780691218700.003.0006