Reinforcement Learning for Portfolio Selection in the Vietnamese Market

Tóm tắt báo cáo

In this paper, we explore the application of reinforcement learning in the context of Vietnam’s rapidly growing financial market, where research on algorithmic trading, in general, remains limited. We implement and compare state-of-the-art reinforcement learning algorithms such as Proximal Policy Optimization (PPO), and Twin Delayed Deep Deterministic Policy Gradient (TD3); in an effort to improve trading strategies and decision-making. Additionally, we employ the Spectral Residual method to detect anomalies in sequence state spaces and mitigate potential risks. We conclude that Special Residual noise filtering delivers the best portfolio performance across the board, and the ActorCritic using Kronecker-Factored Trust Region (ACKTR) and the PPO attain dominance in portfolio performance in the training data and testing data respectively.