Technology and Innovation Community

Recent paper focusing on social media, retail investors, herding, large language models, AI, technical analysis and fundamental analysis written by:

Shuauyu Chen

Purdue University - Mitchell E. Daniels, Jr. School of Business

Lin Peng

City University of New York, Baruch College - Zicklin School of Business - Department of Economics and Finance

Dexin Zhou

City University of New York, Baruch College - Zicklin School of Business - Department of Economics and Finance

Abstract

Using large language models, we analyze trading strategies expressed in over 77 million messages on a leading investor social media platform. We find that stocks experiencing bullish sentiment in technical analysis (TA) posts tend to have lower future returns and a higher likelihood of buy herding on Robinhood. In contrast, sentiment extracted from fundamental analysis (FA)-related posts positively predicts future returns. More intense TA posting is associated with less informative retail order flows, whereas FA posting is positively linked to flow informativeness. We further show that social media TA sentiment tends to contradict signals derived from a state-of-the-art AI-based technical strategy, and the profitability of the AI strategy largely stems from exploiting the TA sentiment.  Our findings provide insights into the investment approaches of retail investors, the role of social media, and the interactions between different market players in the era of social media and AI-powered trading.