A Portfolio Model with Risk Control Policy Based on Deep Reinforcement Learning

It was shown that deep reinforcement learning (DRL) has the potential to solve portfolio management problems in recent years. The Twin Delayed Deep Deterministic policy gradient algorithm (TD3) is an actor-critic method, a typical DRL method continuous action space. Despite success of financial trading, surprisingly, most literature ignores element risk control. research proposed combine long- and short-term (LSTR) control with TD3 build model capabilities. Using Chinese stock data from Shanghai Stock Exchange, we train validate model. Performances were compared without results indicated our proposal offers better investment returns.