Stock Price Prediction Based on Large Language Models and Reinforcement Learning

Stock prediction has always been a key and difficult point of research in the financial field. For a long time, prediction models mostly relied on structured data such as historical market data and technical indicators, making it difficult to fully utilize the rich textual information contained in news, financial reports, and social media, and also difficult to adapt to the dynamic changes of the market. In recent years, the rapid development of artificial intelligence technology, especially the rise of large language models (LLMs) and reinforcement learning (RL), has provided new ideas for stock prediction. This paper systematically reviews the application status of large language models and reinforcement learning in stock prediction and explores the potential of their combination. The article first introduces the basic concepts of large language models and reinforcement learning respectively, then analyzes the advantages of large language models in semantic understanding, context modeling, interpretable prediction, and factor mining, then summarizes the advantages of reinforcement learning in sequential decision-making, return optimization, and risk control, and finally summarizes typical frameworks and synergistic effects of their integration. The purpose of this work is to offer theoretical references and useful advice for developing financial analysis systems that are more intelligent, comprehensible, and self-adaptive.