Analyzing Cascading Outbreak of GameStop Event: A Practical Approach Using Network Analysis and Large Language Models

Shengyuan Lin, Keyi Wang, Xiao-Yang Liu

Abstract: The GameStop short squeeze in January 2021, driven by collective online actions in the Reddit//WallStreetBets community, has shown a surprising impact on financial markets. In this paper, we take a practical approach using network analysis and large language models (LLMs) to investigate such a cascading outbreak phenomenon. Utilizing GPT-4o for data preprocessing tasks, including text cleaning, relevance scoring, and duplicate removal, helps curate a high-quality dataset. We apply models, such as the Virus-on-a-Network Model and the Community-Affiliation Graph Model, to study the underlying mechanisms of user activity and information dissemination. Our analysis shows that a low information spread threshold and influential users contribute to the rapid dissemination of information. Our findings provide insights into the critical role of online communities in the financial market. In this practical approach, LLMs show great capabilities in processing and analyzing social platform speech. This study highlights the importance of online community behaviors and their potential to influence financial markets, offering a novel analytical tool for regulators, financial institutions, and investors.