GPT-4 Outperforms Human Financial Analysts in Predicting Earnings Growth




Bar graph showing GPT-4 outperforming human financial analysts in predicting earnings growth.

In a groundbreaking development, researchers from the University of Chicago have found that OpenAI’s GPT-4 large language model (LLM) outperforms human financial analysts in predicting future company earnings. This discovery could have significant implications for the future of financial analysis and investment strategies.

Key Takeaways

  • GPT-4 demonstrated a 60% accuracy rate in predicting earnings growth, surpassing the 53-57% accuracy range of human analysts.
  • The AI model was tested using anonymized balance sheet and income statement data without any additional context.
  • Researchers used a technique called “chain-of-thought” prompting to enable GPT-4 to mimic human reasoning patterns.
  • The study suggests that combining human and AI models could result in superior forecasts.

The Research

The study, titled ‘Financial Statement Analysis with Large Language Models,’ was published by researchers at the University of Chicago in May 2024. The researchers provided GPT-4 with standardized balance sheet and income statement data and used a detailed “chain-of-thought” prompt outlining various analytical techniques and metrics common in earnings forecast analysis.

Despite lacking industry context and specific company details, GPT-4 achieved impressive results. The model’s goal was to determine whether earnings would grow or decline in a subsequent period and to indicate broad magnitude confidence. The AI outperformed consensus forecasts and showed comparable, if not superior, performance to better public industry models.

Measured Performance

GPT-4’s overall accuracy in predicting whether earnings would increase or decline was approximately 60%. The model also produced similar results for 2023 using 2022 data, indicating it was not drawing on actual historical performance. Interestingly, GPT-4 outperformed its predecessor, GPT-3.5, and showed similar results to Google’s Google Gemini Pro 1.5, though the latter was tested over a more limited sample.

The researchers noted that the model’s superior performance likely stems from its ability to recognize business concepts and patterns and conduct intuitive reasoning even with incomplete datasets. However, they were unable to pinpoint exactly what the model was doing to achieve this accuracy.

Alpha Generation

The study found that investing based on GPT-4’s analysis could historically lead to stock market outperformance. The Sharpe ratio from this strategy was superior to that of an artificial neural net trained for earnings prediction on an equal-weighted basis. However, the artificial neural net outperformed GPT-4 in terms of Sharpe ratio on a value-weighted basis.

The bulk of the model’s returns, especially in recent history, came from its long positions rather than its short exposure. While forecasting accuracy has declined somewhat in recent decades, the results remain generally above that of human consensus forecasts.

Future Implications

The success of GPT-4 in predicting earnings directions with relatively limited financial data and its considerable improvement over GPT-3.5 is impressive. As in many fields, LLMs are likely to become increasingly disruptive and effective in financial analysis and prediction.

However, experts remain cautious. Some argue that the study’s benchmark used for comparison may not represent the capabilities of current financial analysis tools. Others highlight that while AI excels at data analysis and pattern recognition, it lacks the creativity and experience of human analysts.

The researchers believe that combining human and AI models could result in superior forecasts, as humans can bring additional insights that LLMs may not currently have access to, while LLMs can avoid common human biases and perform robust and comprehensive analysis.

In conclusion, while GPT-4’s performance in financial prediction is promising, it is likely to serve as a complementary tool to human analysts rather than a replacement. The future may hold a collaborative approach, where AI augments human capabilities, leading to more informed and potentially more successful financial decisions.


Share this content

Leave a Reply

Your email address will not be published. Required fields are marked *

Latest posts