Stock Price Change Prediction Using Prompt-Based LLMs with RL-Enhanced Post-Hoc Adjustments

DOI

10.2991/978-94-6463-742-7_46How to use a DOI?

Keywords

LLM; reinforcement learning; stock price prediction; time series forecasting

Abstract

While most studies apply large language models (LLMs) to financial sentiment analysis, this research explores a prompt-based approach using LLMs, including Deepseek, Gemma, LLaMA, and Qwen, to estimate stock price percentage changes, also known as simple returns. The models are evaluated on a custom dataset comprising historical stock prices and news articles from both target companies and their related entities, such as competitors and business partners. The results indicate that although LLMs can generate reasonably accurate estimates of stock price fluctuations (actual values), they struggle with forecasting the correct price movement direction (i.e., increase or decrease). To address this, this paper proposes a reinforcement learning framework of using Proximal Policy Optimization to improve directional accuracy of LLMs. This study highlights the potential and limitation of LLMs in stock price forecasting and provides insights into possible solutions to improve the LLM predictive performance.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Qizhao Chen
PY  - 2025
DA  - 2025/05/31
TI  - Stock Price Change Prediction Using Prompt-Based LLMs with RL-Enhanced Post-Hoc Adjustments
BT  - Proceedings of the 2025 4th International Conference on Bigdata Blockchain and Economy Management (ICBBEM 2025)
PB  - Atlantis Press
SP  - 475
EP  - 483
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6463-742-7_46
DO  - 10.2991/978-94-6463-742-7_46
ID  - Chen2025
ER  -