Andreas Neuhierl (Purdue): The Uncertainty of Machine Learning Predictions in Asset Pricing

Please join us for a seminar of the Advanced Financial Technologies Laboratory (AFTLab):

Time: Thursday, October 9, 2025, 5:00pm – 6:00pm

Location: Huang 305

Speaker: Andreas Neuhierl (Purdue University)

Title: The Uncertainty of Machine Learning Predictions in Asset Pricing

Abstract: Machine learning in asset pricing typically predicts expected returns as point estimates, ignoring uncertainty.  We develop new methods to construct forecast confidence intervals for expected returns obtained from neural networks. We show that neural network forecasts of expected returns share the same asymptotic distribution as classic nonparametric methods,  enabling a closed-form expression for their standard errors.  We also propose a computationally feasible bootstrap to obtain the asymptotic distribution. We incorporate these forecast confidence intervals into an uncertainty-averse investment framework. This provides an economic rationale for shrinkage implementations of portfolio selection.  Empirically, our methods improve out-of-sample performance.

Bio: Andreas Neuhierl is an Associate Professor of Finance at Purdue University. His research lies at the intersection of empirical asset pricing, financial econometrics and machine learning. His research has been published in the leading journals in finance and economics, such as the Journal of Financial Economics, Review of Financial Studies and the Journal of Monetary Economics. He joined the Purdue faculty in 2025 after serving as an Assistant Professor of Finance at Washington University in St. Louis. Prior to Washington University, he was on the finance faculty at the University of Notre Dame and held visiting appointments at the University of Chicago Booth School of Business. He received his PhD in Finance from Northwestern University and a doctoral degree in Statistics from the University of Augsburg.