X(推特)社群媒體貼文與個股報酬

Abstract

本篇論文探討了從社群媒體平台 X（前身為 Twitter）貼文匯聚的情緒分數對每週公司層面股票回報的預測能力，並探討網路平台作為關鍵資訊來源的影響。本次研究針對標記公司上市代號的推文的資料集進行分析，並使用包括 VADER 演算法在內的情緒分析技術，通過統計回歸分析衡量社群媒體對股市表現的影響。研究結果表明以週為單位，情緒、極性和相異詞彙量的增加與股票正回報相關，證明了在社群媒體上表達的投資者情緒的強大預測價值。這項研究有助於理解社群媒體如何影響資本市場，並指出使用更多指標進行網路分析的潛力，才能探討人們被社群媒體所影響投資行為的見解。

This thesis investigates the predictive power of aggregated sentiment from the social media platform X (formerly known as Twitter) on weekly firm-level stock returns and highlights the increasing relevance of online platforms as essential sources of information. The study uses a comprehensive dataset of tweets labeled with firm-specific symbols and applies advanced sentiment analysis techniques, including the VADER tool, to examine the impact of social media sentiment on stock market performance using regression analysis. The findings suggest that positive shifts in sentiment, polarity, and unique word counts correlate with higher weekly stock returns, emphasizing the significant predictive value of investor sentiment expressed on social media. This research enhances the understanding of the influence of social media on capital markets and explores the potential of engagement metrics for conducting network analysis to further explore investment behaviors driven by social media.