# Practical Deep Reinforcement Learning Approach for Stock Trading

**Authors:** Xiao-Yang Liu, Zhuoran Xiong, Shan Zhong, Hongyang Yang, Anwar Walid  
**arXiv:** [1811.07522](https://arxiv.org/abs/1811.07522)  
**Date:** 2018-11-19  
**Summary page:** https://www.summarizepaper.com/en/paper/1811.07522v3/

## Key points

- Stock trading strategy is crucial for maximizing returns in investment companies
- The stock market's complex and dynamic nature makes it challenging to obtain an optimal strategy
- The study titled "Practical Deep Reinforcement Learning Approach for Stock Trading" explores the potential of deep reinforcement learning to optimize stock trading strategies
- The study uses 30 stocks as trading stocks and their daily prices as the training and trading market environment
- The authors train a deep reinforcement learning agent to obtain an adaptive trading strategy that maximizes investment return
- The proposed deep reinforcement learning approach outperforms both Dow Jones Industrial Average and traditional min-variance portfolio allocation strategies in terms of Sharpe ratio and cumulative returns
- This suggests that deep reinforcement learning can be a powerful tool for optimizing stock trading strategies in complex markets
- This study contributes to the growing body of research on using machine learning techniques for financial decision-making.

## Summary

In the world of investment companies, stock trading strategy plays a crucial role in maximizing returns. The complex and dynamic nature of the stock market makes it challenging to obtain an optimal strategy. In this study titled "Practical Deep Reinforcement Learning Approach for Stock Trading," authors Xiao-Yang Liu, Zhuoran Xiong, Shan Zhong, Hongyang Yang, and Anwar Walid explore the potential of deep reinforcement learning to optimize stock trading strategies. The study uses 30 stocks as trading stocks and their daily prices as the training and trading market environment. The authors train a deep reinforcement learning agent to obtain an adaptive trading strategy that maximizes investment return. To evaluate the agent's performance, they compare it with Dow Jones Industrial Average and traditional min-variance portfolio allocation strategies. The proposed deep reinforcement learning approach outperforms both baselines in terms of both Sharpe ratio and cumulative returns. This suggests that deep reinforcement learning can be a powerful tool for optimizing stock trading strategies in complex markets. This study contributes to the growing body of research on using machine learning techniques for financial decision-making. It highlights the potential of deep reinforcement learning as a promising approach for improving investment outcomes in dynamic markets by providing an effective way to optimize stock trading strategies while taking into account the complexities of the market environment.