Deep Learning in Characteristics-Sorted Factor Models

52 Pages Posted: 23 Sep 2018 Last revised: 11 Apr 2023

See all articles by Guanhao Feng

Guanhao Feng

City University of Hong Kong (CityU)

Jingyu He

City University of Hong Kong (CityU)

Nick Polson

University of Chicago - Booth School of Business

Jianeng Xu

University of Chicago, Students

Date Written: October 1, 2021

Abstract

This paper presents an augmented deep factor model that generates latent factors for cross-sectional asset pricing. The conventional security sorting on firm characteristics for constructing long-short factor portfolio weights is nonlinear modeling, while factors are treated as inputs in linear models. We provide a structural deep learning framework to generalize the complete mechanism for fitting cross-sectional returns by firm characteristics through generating risk factors -- hidden layers. Our model has an economic-guided objective function that minimizes aggregated realized pricing errors. Empirical results on high-dimensional characteristics demonstrate robust asset pricing performance and strong investment improvements by identifying important raw characteristic sources.

Keywords: Cross-sectional Returns, Deep Learning, Latent Factors, Pricing Errors, Security Sorting.

JEL Classification: C1, G1

Suggested Citation

Feng, Guanhao and He, Jingyu and Polson, Nick and Xu, Jianeng, Deep Learning in Characteristics-Sorted Factor Models (October 1, 2021). Available at SSRN: https://ssrn.com/abstract=3243683 or http://dx.doi.org/10.2139/ssrn.3243683

Guanhao Feng (Contact Author)

City University of Hong Kong (CityU) ( email )

83 Tat Chee Avenue
Hong Kong

Jingyu He

City University of Hong Kong (CityU) ( email )

83 Tat Chee Avenue
Hong Kong
Hong Kong

Nick Polson

University of Chicago - Booth School of Business ( email )

5807 S. Woodlawn Avenue
Chicago, IL 60637
United States
773-702-7513 (Phone)
773-702-0458 (Fax)

Jianeng Xu

University of Chicago, Students ( email )

Chicago, IL
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
2,317
Abstract Views
8,314
Rank
10,251
PlumX Metrics