Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models

We examine the potential of ChatGPT, and other large language models, in predicting stock market returns using sentiment analysis news headlines. use ChatGPT to indicate whether a given headline is good, bad, or irrelevant for firms' prices. then compute numerical score document positive correlation between these scores subsequent daily returns. Further, outperforms traditional methods. find that more basic models such as GPT-1, GPT-2, BERT cannot accurately forecast returns, indicating return predictability an emerging capacity complex models. Our results suggest incorporating advanced into investment decision-making process can yield accurate predictions enhance performance quantitative trading strategies.