Can ChatGPT Help Professional Analysts Pick Better Stocks? Testing Human AI Collaboration in Stock Market Predictions
Major
Finance
Submission Type
Poster
Area of Study or Work
Finance
Faculty Advisor
Bryan McCannon
Location
CNS Atrium
Start Date
4-11-2026 11:15 AM
End Date
4-11-2026 12:30 PM
Abstract
This study examines whether combining ChatGPT 4 recommendations with professional analyst ratings improves investment performance compared to using either source alone. I tracked 50 major U.S. companies across five sectors over seven weeks from October 17 to December 5 2025. Each week, I collected analyst consensus ratings from MarketBeat and asked ChatGPT for recommendations based on recent news headlines. I then constructed three hypothetical portfolios: an Analyst only portfolio, a ChatGPT only portfolio, and a Hybrid portfolio containing only stocks where both sources agreed on a "Buy" rating. The Hybrid portfolio generated a 7.8% total return compared to 5.2% for analysts alone and 4.1% for ChatGPT alone, while also showing lower volatility (1.6% vs. 2.3% and 2.5%). The Sharpe ratio of 0.36 for the Hybrid portfolio substantially exceeded both the Analyst portfolio (0.15) and ChatGPT portfolio (0.10). Regression analysis controlling for market conditions and past performance revealed that agreement between both sources predicted an additional 0.35% weekly return (p = 0.019). These findings suggest that human AI collaboration may produce better investment outcomes than relying on either source independently.
Can ChatGPT Help Professional Analysts Pick Better Stocks? Testing Human AI Collaboration in Stock Market Predictions
CNS Atrium
This study examines whether combining ChatGPT 4 recommendations with professional analyst ratings improves investment performance compared to using either source alone. I tracked 50 major U.S. companies across five sectors over seven weeks from October 17 to December 5 2025. Each week, I collected analyst consensus ratings from MarketBeat and asked ChatGPT for recommendations based on recent news headlines. I then constructed three hypothetical portfolios: an Analyst only portfolio, a ChatGPT only portfolio, and a Hybrid portfolio containing only stocks where both sources agreed on a "Buy" rating. The Hybrid portfolio generated a 7.8% total return compared to 5.2% for analysts alone and 4.1% for ChatGPT alone, while also showing lower volatility (1.6% vs. 2.3% and 2.5%). The Sharpe ratio of 0.36 for the Hybrid portfolio substantially exceeded both the Analyst portfolio (0.15) and ChatGPT portfolio (0.10). Regression analysis controlling for market conditions and past performance revealed that agreement between both sources predicted an additional 0.35% weekly return (p = 0.019). These findings suggest that human AI collaboration may produce better investment outcomes than relying on either source independently.