Home | Publications | WSS+23

Constrained Portfolio Management Using Action Space Decomposition for Reinforcement Learning

MCML Authors

Link to Profile Matthias Schubert

Matthias Schubert

Prof. Dr.

Core PI

Spatial Artificial Intelligence

Link to Profile Thomas Seidl

Thomas Seidl

Prof. Dr.

Director

Database Systems, Data Mining and AI

Abstract

Financial portfolio managers typically face multi-period optimization tasks such as short-selling or investing at least a particular portion of the portfolio in a specific industry sector. A common approach to tackle these problems is to use constrained Markov decision process (CMDP) methods, which may suffer from sample inefficiency, hyperparameter tuning, and lack of guarantees for constraint violations. In this paper, we propose Action Space Decomposition Based Optimization (ADBO) for optimizing a more straightforward surrogate task that allows actions to be mapped back to the original task. We examine our method on two real-world data portfolio construction tasks. The results show that our new approach consistently outperforms state-of-the-art benchmark approaches for general CMDPs.

inproceedings WSS+23

PAKDD 2023

27th Pacific-Asia Conference on Knowledge Discovery and Data Mining. Osaka, Japan, May 25-28, 2023.
Conference logo
A Conference

Authors

D. Winkel • N. Strauß • M. Schubert • Y. Ma • T. Seidl

Links

DOI

Research Area

 A3 | Computational Models

BibTeXKey: WSS+23

#p-schubert #p-seidl #p-tresp
Back to Top