11th Online Seminar in Finance: Alejandro Lopez-Lira (University of Florida)
The Institute of Finance at PUEB is pleased to announce the 11th Online Seminar in Finance.
Speaker: Alejandro Lopez-Lira, University of Florida
Title: Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models
Abstract: We examine the potential of ChatGPT and other large language models to predict stock market returns using news headlines. We evaluate headlines with ChatGPT to determine if they are positive, negative, or neutral for companies’ stock prices. We find a significant positive correlation between ChatGPT scores and subsequent daily stock returns and that ChatGPT outperforms traditional sentiment analysis methods. Basic models such as GPT-1, GPT-2, and BERT cannot accurately forecast returns, indicating return predictability is an emerging capacity of complex language models. Long-short strategies based on ChatGPT-4 deliver the highest Sharpe ratio. Furthermore, we find predictability in small and large stocks, suggesting market underreaction to news. Predictability is stronger among smaller stocks and negative headlines, consistent with limits-to-arbitrage arguments. Only advanced language models effectively interpret signals in firms’ press releases. We propose a novel method to evaluate ChatGPT’s reasoning. Overall, incorporating advanced language models into investment decisions could improve prediction accuracy and quantitative trading performance.
Date: December 1st, 2023, 15:00 (Warsaw time)
Speaker’s bios: Alejandro Lopez-Lira is an Assistant Professor of Finance at Warrington College of Business, University of Florida, USA. He holds a PhD in Finance from the Wharton School, University of Pennsylvania. His research covers topics in empirical asset pricing, machine learning, and textual analysis.
To register, please click here: https://forms.gle/kWrzCExxA6kwBbZu9 *
Registration deadline: November 29th, 2023, 20:00 (Warsaw time). The invitation to the Teams webinar will be sent on November 1st, 2023.
*If the link does not work, please paste it into your browser.