A Stock Trading Strategy Based on Deep Reinforcement Learning

The stock market plays a vital role in the overall financial market. Financial trading has been broadly researched over the years. However, it remains challenging to obtain an optimal strategy in an environment as complex and dynamic as the stock market. Our article is interested in solving a stochastic control problem that aims at optimizing the management of a trading system in order to obtain an optimal trading strategy that would enable us to make profitable decisions by interacting directly with the environment. To do this, we explore the power of deep Reinforcement Learning that differs from traditional Machine Learning by combining the task of predicting stock behavior and analyzing the optimal course of action in a single unit, thus aligning the Machine Learning problem with the investor's objectives. As a method, we propose to use the Deep Q-Network algorithm which is a combination of Q-Learning and Deep Learning. Experiments show that the approach proposed can learn the behavior to solve a stock trading problem by producing positive results in a complex dynamic environment.