LLM Analysis of 10-K and 10-Q Filings: RAG Results

Shyam Balagurumurthy Viswanathan , Amritha Arun Babu Mysore , Arjun Karat , Raghavan Muthuregunathan

Artificial Intelligence, Machine Learning, LLM, Finance, Financial Reports

In the field of financial analytics, Large Language Models (LLMs) have become significant tools for analyzing and interpreting complex, unstructured text, like the content found in financial reports. This paper focuses on the use of LLMs to study 10-K and 10-Q filings, which are essential financial statements required from publicly traded companies. These reports thoroughly view a company's financial situation and future outlook, targeting a broad audience, including investors, analysts, regulators, employees, and the general public. The main part of this research involves combining Retrieval-Augmented Generation (RAG) models and knowledge graphs with LLMs to improve the analysis of these financial documents. A key challenge in understanding 10-K and 10-Q reports is figuring out which companies are competitors and how they fit into the market, something that regular LLM methods don't fully address. Our approach uses RAG models and knowledge graphs to overcome this issue, offering a better understanding of how companies compete and where they stand in their industry based on these reports. This study aims to show that integrating LLMs with RAG and knowledge graphs can significantly improve how we analyze financial data, leading to a deeper and more complete understanding of a company's position in the market, its competition, and potential risks. This research goes beyond just analyzing financial reports, suggesting a new way to use AI in business intelligence and market research.

Volume 9 Issue 1, January-2024

Pages : 594 - 602