🤖 AI Summary
This paper investigates whether large language models (LLMs) can predict daily stock price movements using news headlines alone—without domain-specific fine-tuning on financial data.
Method: We conduct zero-shot and few-shot inference with models including ChatGPT, complemented by event-study analysis, panel regressions, and a novel theoretical model integrating information capacity constraints, irrational investor responses, and arbitrage limitations to characterize AI capability thresholds. We further propose the first interpretability evaluation framework for financial reasoning in LLMs, incorporating customized prompt engineering and reasoning-path attribution.
Contribution/Results: Empirical results show that LLM-derived signals significantly improve out-of-sample return prediction accuracy relative to conventional text-based models; exhibit heightened sensitivity to small-cap stocks and negative news; validate the existence of an AI capability threshold effect; and demonstrate that broader LLM adoption accelerates price discovery and enhances market informational efficiency.
📝 Abstract
We document the capability of large language models (LLMs) like ChatGPT to predict stock price movements using news headlines, even without direct financial training. ChatGPT scores significantly predict out-of-sample daily stock returns, subsuming traditional methods, and predictability is stronger among smaller stocks and following negative news. To explain these findings, we develop a theoretical model incorporating information capacity constraints, underreaction, limits-to-arbitrage, and LLMs. The model generates several key predictions, which we empirically test: (i) it establishes a critical threshold in AI capabilities necessary for profitable predictions, (ii) it demonstrates that only advanced LLMs can effectively interpret complex information, and (iii) it predicts that widespread LLM adoption can enhance market efficiency. Our results suggest that sophisticated return forecasting is an emerging capability of AI systems and that these technologies can alter information diffusion and decision-making processes in financial markets. Finally, we introduce an interpretability framework to evaluate LLMs' reasoning, contributing to AI transparency and economic decision-making.
