AI Models Enhance Understanding of Stock Price Movements

Recent research from Chicago Booth highlights a significant advancement in understanding stock price movements following earnings announcements, utilizing artificial intelligence (AI) to enhance predictive accuracy. Traditionally, the earnings surprise metric—comparing reported earnings per share to analysts’ forecasts—has dominated this analysis, explaining only about 5 percent of same-day stock price changes. The new findings from researchers Ralph S. J. Koijen and Bradford Levy indicate that AI models can explain approximately 17 percent of these movements, marking a notable improvement in market analysis.

The implications of this research are profound. By employing three frontier AI models in a live testing environment, the researchers analyzed nearly 2,000 earnings announcements from late 2025. This approach mitigated the risk of lookahead bias, which can distort results when models are trained on historical data that includes information not available at the time of the events being studied. The AI models not only predicted stock movements more effectively but also generated human-readable explanations that illuminate the reasoning behind their predictions. This capability allows analysts and investors to glean insights from earnings calls that go beyond mere numerical data, incorporating elements such as executive commentary and forward guidance.

The findings suggest that seasoned financial professionals may need to recalibrate their understanding of market reactions. For instance, the AI models indicated that strong earnings results, such as record profits or significant growth, often do not lead to stock price increases if the good news is already priced in. This insight challenges traditional assumptions and underscores the need for a more nuanced approach in analyzing market sentiment and stock valuation.

Koijen and Levy's research opens new avenues for further exploration. Their human-readable explanations can serve as a foundation for researchers and analysts to develop more specific hypotheses regarding stock price movements. As AI technology continues to evolve, there is potential for these insights to be integrated into smaller, cost-effective models, thereby democratizing access to advanced predictive capabilities across the investment landscape.

Despite these advancements, the researchers caution that a substantial portion of stock price variation remains unexplained, with their models only accounting for 20 percent of the movements. To further enhance predictive accuracy, they have launched a competition, sponsored by trading firm Optiver, inviting participants to submit their own models aimed at predicting stock responses to earnings-call information. This initiative could spur innovation in financial modeling and broaden the understanding of market dynamics.

Looking ahead, the integration of AI into financial analysis is likely to reshape competitive dynamics within the investment sector. Firms that adopt these advanced models may gain a significant edge in market forecasting, enabling them to make more informed decisions based on nuanced insights. As AI continues to refine its predictive capabilities, it is poised to transform not only how investors interpret earnings announcements but also how they strategize around market movements. The potential for AI to enhance understanding of stock price behavior signals a shift towards more sophisticated, data-driven investment strategies, which could redefine success in the financial markets.