An Improved Reinforcement Learning Model Based on Sentiment Analysis

The progress of artificial intelligence technology has led to the rise of quantitative trading systems, particularly those using reinforcement learning, in the stock trading market.In this article, the deep Q network in reinforcement learning is combined with the emotional quantitative indicator ARBR to develop a high-frequency stock trading model specifically for the Chinese A-share market.In order to improve the model's performance, this study employs the PCA algorithm to reduce the dimensionality of the feature vector and incorporates the impact of market sentiment on long-short power into the spatial state of the trading model.Furthermore, the LSTM layer is used in place of the fully connected layer to overcome the limitations of empirical data storage in the traditional DQN model.This paper analyzes the model's performance based on cumulative income and the Sharpe ratio, while also comparing it to other approaches like the double moving averages strategy.The data indicates that the suggested model exceeds the comparison model in return rates, with a maximum annualized return of 54.5%, emphasizing its ability to substantially improve Reinforcement Learning results in stock trading.