18 October 2024

SUERF: Large Language Models and the Future of Financial Analysis

by Kim, Muhn, Nikolaev: provided standardized and anonymous financial statements to GPT4 and instruct the model to analyze them to determine the direction of future earnings.... the LLM outperforms financial analysts in its ability to predict earnings changes.

Abstract
We investigate whether an LLM can successfully perform financial statement analysis in a way similar to a professional human analyst. We provide standardized and anonymous financial statements to GPT4 and instruct the model to analyze them to determine the direction of future earnings. Even without any narrative or industry-specific information, the LLM outperforms financial analysts in its ability to predict earnings changes. The LLM exhibits a relative advantage over human analysts in situations when the analysts tend to struggle. Furthermore, we find that the prediction accuracy of the LLM is on par with the performance of a narrowly trained state-of-the-art ML model. LLM prediction does not stem from its training memory. Instead, we find that the LLM generates useful narrative insights about a company’s future performance. Lastly, our trading strategies based on GPT’s predictions yield a higher Sharpe ratio and alphas than strategies based on other models. Taken together, our results suggest that LLMs may take a central role in decision-making.

Introduction

The recent emergence of Large Language Models (LLMs) has sparked intense debate about the potential of generative AI. Although such tools have been widely adopted for various tasks in the financial domain, such as summarization, information extraction, or report writing, it is still unclear to what extent LLMs can play a critical role in financial markets. Therefore, in our recent research project, we investigate whether an LLM can effectively replicate or even surpass human financial analysts (as well as narrowly trained machine learning models) in predicting firms’ future performance. Based on our recent SSRN working paper, this policy brief highlights the performance of LLMs in this challenging quantitative task and discusses the broader implications for financial markets and decision-making processes.

Context of the Earnings Prediction Task

Earnings prediction is a cornerstone of financial analysis and a critical input for valuation models or stock recommendations. Financial analysts forecast future earnings by scrutinizing a company’s financial statements and assessing its financial health and growth potential. This process often involves a blend of quantitative analysis, contextual interpretation, and professional judgment, relying on a wealth of domain-specific knowledge and experience.

Our study examines whether an LLM like GPT-4 can perform this complex task on par with, or better than, human analysts and specialized machine learning models. We focus on a scenario where GPT is provided with only anonymized and standardized financial statements. We then aim to assess its capability to generate meaningful insights from purely numerical data....

 more at SUERF

© SUERF