DeepTrader: A Deep Reinforcement Learning Approach for Risk-Return Balanced Portfolio Management with Market Conditions Embedding

• DOI: 10.1609/aaai.v35i1.16144
• Publication Date: 2022-09-08T17:51:00Z
• Authors: Zhicheng Wang; Biwei Huang; Shikui Tu; Kun Zhang; Lei Xu
• Keywords: Risk–return spectrum; Macro; Drawdown (hydrology)

Most existing reinforcement learning (RL)-based portfolio management models do not take into account the market conditions, which limits their performance in risk-return balancing. In this paper, we propose DeepTrader, a deep RL method to optimize investment policy. particular, tackle balancing problem, our model embeds macro conditions as an indicator dynamically adjust proportion between long and short funds, lower risk of fluctuations, with negative maximum drawdown reward function. Additionally, involves unit evaluate individual assets, learns dynamic patterns from historical data price rising rate Both temporal spatial dependencies assets are captured hierarchically by specific type graph structure. Particularly, find that estimated causal structure best captures interrelationships compared industry classification correlation. The two units complementary integrated generate suitable fits trend well strikes balance return effectively. Experiments on three well-known stock indexes demonstrate superiority DeepTrader terms risk-gain criteria.