From words to returns: sentiment analysis of Japanese 10-K reports using advanced large language models
A major study finds that large language models (LLMs) like ChatGPT, Claude, and Gemini can extract sentiment from Japanese 10-K financial reports to predict future stock returns. Using data from over 11,000 firm-years, researchers discovered that negative sentiment in reports signals better future performance, challenging the efficient market hypothesis. Traditional sentiment analysis methods failed to show this link. The study highlights the power and complexity of advanced natural language processing for financial forecasting.
What it examines
This study uses advanced language models like ChatGPT, Claude, and Gemini to analyze sentiment in Japanese 10-K financial reports. By examining data from all Tokyo Stock Exchange companies (2014-2023), it aims to see if these models can predict future stock returns better than traditional methods.
What it concludes
The research finds that large language models can predict stock returns from report sentiment, unlike older methods. This challenges the efficient market hypothesis and suggests new ways to forecast stocks. Applications include improved financial analysis and investment strategies. Future work may refine models and explore other financial documents.
Evidence objects
A major study used advanced language models like ChatGPT, Claude, and Gemini to analyze sentiment in over 11,000 Japanese 10-K reports, covering 70 million words from 2014 to 2023.
key_findings bullet 1 · key_findings · validation V0
Surprisingly, LLM-derived sentiment scores showed a significant negative correlation with future stock returnsmore negative sentiment predicted better performancechallenging the efficient market hypothesis and outperforming traditional dictionary-based methods.
key_findings bullet 2 · key_findings · validation V0
The research highlights how sophisticated NLP tools can uncover predictive signals missed by older models, but notes practical challenges in real-world trading and the risk of overfitting remain unresolved.
key_findings bullet 3 · key_findings · validation V0
This paper uniquely applies advanced LLMs (ChatGPT, Claude, Gemini) to sentiment analysis of Japanese 10-K reports, an underexplored dataset. Its findingthat LLM-derived sentiment predicts future stock returns while traditional methods do notchallenges the efficient market hypothesis, offering novel insights and significant implications for both academic research and investment practice.
key_findings bullet 4 · key_findings · validation V0
Raw abstract and provenance
The advent of advanced natural language processing techniques and large language models (LLMs) has revolutionized the analysis of qualitative financial data. This research harnesses the capabilities of LLMs—specifically ChatGPT, Claude, and Gemini—to extract sentiment from Japanese 10-K reports, aiming to predict future stock returns. By analyzing an extensive dataset encompassing all companies listed on the Tokyo Stock Exchange from 2014 to 2023—a total of 11,135 firm-years and over 70 million words—we conduct the first comprehensive study of its kind in Japan. Comparative analyses are performed using traditional dictionary-based methods and a DeBERTaV2-based model to evaluate efficacy in information extraction. Our findings reveal substantial differences in the models’ abilities to predict stock performance. Notably, while dictionary-based methods show no significant relationship between sentiment and subsequent stock returns, LLM-derived sentiments exhibit a significant negative correlation with future returns. These results challenge the efficient market hypothesis by demonstrating that sentiment extracted from publicly available reports can predict stock performance. This study reveals the transformative potential of advanced Natural Language Processing (NLP) technologies in financial analysis, highlighting how sophisticated language models can uncover predictive signals previously undetected by traditional methods. The article details the methodologies employed, the challenges encountered, and the implications for integrating advanced sentiment analysis into financial forecasting.
Source row: 946 · abstract type: unknown