Researchers at the University of Chicago have discovered that large language models LLMs, such as GPT-4, are capable of conduction financial statement analysis with an accuracy that rivals their human counterparts — and in some cases surpasses them.The group published their findings this week in a paper entitled “Financial Statement Analysis with Large Language Models.”For their research, the group tested the performance of GPT-4 in analyzing corporate financial statements to predict future earnings growth. “Even without any narrative or industry-specific information, the LLM outperforms financial analysts in its ability to predict earnings changes,” the researchers said. “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.”The group said that the LLM was able to generate “useful narrative insights about a company’s future performance” and that its trading strategies based on GPT’s predictions “yield a higher Sharpe ratio and alphas than strategies based on other models.”
Recommended by Our Editors
The researchers note that LLMs likely have an advantage because of their vast knowledge base as well as ability to quickly recognize patterns and business concepts, all things that allow it to make intuitive decisions even with incomplete information.While we’ll still likely have human financial analysts for the foreseeable future, the research points to LLMs becoming a potentially helpful tool for those analysts to help them streamline their work and ultimately make better and more informed decisions. Over time; however, the use of LLMs could potentially change how financial analysts work in the future.
Get Our Best Stories!
Sign up for What’s New Now to get our top stories delivered to your inbox every morning.
This newsletter may contain advertising, deals, or affiliate links. Subscribing to a newsletter indicates your consent to our Terms of Use and Privacy Policy. You may unsubscribe from the newsletters at any time.
