Explainable Deep Reinforcement Learning for Portfolio Management: An Empirical Approach

TL;DR

This paper presents an empirical method to interpret deep reinforcement learning strategies in portfolio management by comparing feature importance with a linear hindsight model, revealing DRL's superior multi-step prediction power.

Contribution

It introduces an explainability approach for DRL in portfolio management using integrated gradients and linear hindsight models, providing insights into multi-step prediction capabilities.

Findings

DRL agents show stronger multi-step prediction power than other machine learning methods.

The approach effectively quantifies feature importance and prediction power in portfolio strategies.

Empirical analysis on Dow Jones 30 stocks demonstrates the method's practical applicability.

Abstract

Deep reinforcement learning (DRL) has been widely studied in the portfolio management task. However, it is challenging to understand a DRL-based trading strategy because of the black-box nature of deep neural networks. In this paper, we propose an empirical approach to explain the strategies of DRL agents for the portfolio management task. First, we use a linear model in hindsight as the reference model, which finds the best portfolio weights by assuming knowing actual stock returns in foresight. In particular, we use the coefficients of a linear model in hindsight as the reference feature weights. Secondly, for DRL agents, we use integrated gradients to define the feature weights, which are the coefficients between reward and features under a linear regression model. Thirdly, we study the prediction power in two cases, single-step prediction and multi-step prediction. In particular, we quantify the prediction power by calculating the linear correlations between the feature weights of a DRL agent and the reference feature weights, and similarly for machine learning methods. Finally, we evaluate a portfolio management task on Dow Jones 30 constituent stocks during 01/01/2009 to 09/01/2021. Our approach empirically reveals that a DRL agent exhibits a stronger multi-step prediction power than machine learning methods.