This study examines the application of Large Language Models (LLMs) in fundamental analysis, focusing on company reports and documents describing macroeconomic conditions such as GDP and inflation changes. This data includes filings submitted to the U.S. Securities and Exchange Commission (SEC), which are accessible in the EDGAR database.
We preprocessed this data and sent it via API to the gpt-4o model in a Retrieval-Augmented Generation (RAG) style. Additionally, we prepared a document outlining an exemplar investor knowledge based on Kitchin cycles. Over a four-week period, we scanned important data for nine companies.
Using LLMs, we produced automatic briefs about these companies and sent them to nine participants, who are individual investors, to evaluate the effectiveness of this data analysis approach.
Blogger's Review: This study showcases the potential of Large Language Models in fundamental analysis, particularly in automating investor brief generation. The use of the RAG mechanism enhances data processing efficiency, providing investors with more accurate information to aid their decision-making process. Effective preprocessing and model application are key to success.