Machine Learning in Hedge Fund Classification: Systematic vs. Discretionary Strategies and Their Performance Implications

34 Pages Posted: 27 Sep 2021 Last revised: 15 Apr 2024

See all articles by Hui-Ching Chuang

Hui-Ching Chuang

National Taipei University - Department of Statistics

Chung‐Ming Kuan

National Taiwan University

Date Written: March 25, 2021

Abstract

This paper applies machine learning to classify hedge funds into systematic and discretionary
categories. Leveraging textual analysis and advanced methods, our approach eliminates subjective judgment in analyzing investment strategies. We find that systematically classified funds, on av-
erage, yield higher excess returns than discretionary ones. Additionally, after applying the false discovery rate test for linear asset pricing models, a higher portion of positive alpha is observed in
the systematic category. The alpha average for outperforming systematic funds surpasses that of
discretionary funds across various risk factor models.

Keywords: False discovery rate, Random forest, Machine learning, Textual analysis

JEL Classification: C63, G11, G14, G23

Suggested Citation

Chuang, Hui-Ching and Kuan, Chung‐Ming, Machine Learning in Hedge Fund Classification: Systematic vs. Discretionary Strategies and Their Performance Implications (March 25, 2021). Available at SSRN: https://ssrn.com/abstract=3912348 or http://dx.doi.org/10.2139/ssrn.3912348

Hui-Ching Chuang (Contact Author)

National Taipei University - Department of Statistics ( email )

151, University Rd., San Shia District,
New Taipei City, 23741
Taiwan

Chung‐Ming Kuan

National Taiwan University ( email )

1 Sec. 4, Roosevelt Road
Taipei 106, 106
Taiwan

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
136
Abstract Views
656
Rank
385,631
PlumX Metrics