With the development of financial technology, quantitative trading has become an important trading method in the market with the help of data and algorithms. However, traditional quantitative trading models exhibit limitations in frequent fluctuations and nonlinear markets. Therefore, this paper reviews the applications of deep reinforcement learning in various areas of quantitative trading. And focuses on high-frequency trading, medium- and long-term asset allocation and timing strategies, as well as cryptocurrency trend forecasting and risk control. This paper's research demonstrates that deep reinforcement learning has strong adaptability and strategy optimization potential in complex and dynamic markets. It can potentially improve returns and risk management. But it still faces some challenges, such as poor interpretability, weak robustness and robustness, as well as high training costs. This paper compares the performance of deep reinforcement learning models with other models and summarizes their advantages and disadvantages. It also recommends that future work on deep reinforcement learning should focus on improving model transparency, establishing risk constraint mechanisms, and optimizing training efficiency to promote the more robust development of intelligent trading systems.
Bing Ren (Wed,) studied this question.