REPUTATION-BASED HERDING ON REDDIT AND ITS EFFECT ON STOCK MARKET REACTIONS

Social media platforms like Reddit have emerged as channels through which retail investor sentiment forms and spreads, influencing the behavior of actors in the stock market. This paper examines how Reddit sentiment affects market valuations and whether more reputable authors guide broader crowd behavior. The data covers Advanced Micro Devices (AMD) and Tilray (TLRY) in 2018. A large language model, Claude Haiku 4.5, was prompted to conduct sentiment analysis of subreddit posts. Authors are assigned an influencer label based on a set of interaction criteria. Using an OLS regression with log returns of AMD and Tilray as the outcome variable across three time intervals, I find that for a large-cap stock like AMD, the influencer signal and its interaction terms show no significance. For Tilray, the influencer interaction terms are significant at the five-minute interval, while positive posts from non-influencers are not. These findings suggest that reputation-based herding is most apparent in speculative stocks with weak fundamentals, and that credible Reddit authors carry differential price impacts relative to general crowd sentiment.