Benchmarking Robustness of Deep Reinforcement Learning approaches to Online Portfolio Management
Machine Learning
2024-01-17 v1 Portfolio Management
Abstract
Deep Reinforcement Learning approaches to Online Portfolio Selection have grown in popularity in recent years. The sensitive nature of training Reinforcement Learning agents implies a need for extensive efforts in market representation, behavior objectives, and training processes, which have often been lacking in previous works. We propose a training and evaluation process to assess the performance of classical DRL algorithms for portfolio management. We found that most Deep Reinforcement Learning algorithms were not robust, with strategies generalizing poorly and degrading quickly during backtesting.
Keywords
Cite
@article{arxiv.2306.10950,
title = {Benchmarking Robustness of Deep Reinforcement Learning approaches to Online Portfolio Management},
author = {Marc Velay and Bich-Liên Doan and Arpad Rimmel and Fabrice Popineau and Fabrice Daniel},
journal= {arXiv preprint arXiv:2306.10950},
year = {2024}
}
Comments
Submitted to INISTA 2023