Predicting abnormal trading behavior from internet rumor propagation: a machine learning approach

Abstract

In 2021, the abnormal short-term price fluctuations of GameStop, which were triggered by internet stock discussions, drew the attention of academics, financial analysts, and stock trading commissions alike, prompting calls to address such events and maintain market stability. However, the impact of stock discussions on volatile trading behavior has received comparatively less attention than traditional fundamentals. Furthermore, data mining methods are less often used to predict stock trading despite their higher accuracy. This study adopts an innovative approach using social media data to obtain stock rumors, and then trains three decision trees to demonstrate the impact of rumor propagation on stock trading behavior. Our findings show that rumor propagation outperforms traditional fundamentals in predicting abnormal trading behavior. The study serves as an impetus for further research using data mining as a method of inquiry.

Fake news in business and management literature: a systematic review of definitions, theories, methods and implications

“…Additionally, managers must address possible customer harm and take preventive efforts to protect them. Cheng et al (2023) predicts abnormal stock trading behavior using social media data (posts, likes and responses) and decision tree induction. They find that rumor propagation predicts abnormal trading behavior better than management shocks and other factors.…”