A recent draft paper by researchers at the University of Chicago reveals that generative AI surpasses financial analysts in predicting future earnings direction by 8 percentage points, showcasing the potential of AI in financial forecasting.
A recent draft paper by Alex Kim, Maximilian Muhn, and Valeri Nikolaev from the University of Chicago indicates that generative AI has outperformed financial analysts in predicting future earnings direction by 8 percentage points. The study highlighted that the AI model was only fed standardised balance sheets and income statements with anonymised company details, disallowing contextual information like sectoral issues or economic backdrop.
The key to ChatGPT’s accuracy is its “chain of thought” reasoning, which enables it to perform mathematical calculations and spot trends effectively. Despite AI’s strong performance, analysts have resources like earnings calls, management discussions, and industry knowledge that still play a crucial role in financial forecasting.
Currently, banks, hedge funds, and wealth managers are leveraging AI for routine tasks such as transcribing earnings calls and summarising data. These tools have proven beneficial in reducing human error and saving time. Nonetheless, the study suggests a hybrid approach, combining human expertise with AI capabilities, is likely to yield the most accurate forecasts in the financial sector.
