Review of Social Media Sentiment and Contextual Bandit Models in Stock Market Investment

College of Arts and Sciences, Wake Forest University, Winston Salem, NC, USA

^* Corresponding author: miaor22@wfu.edu

Abstract

In recent years, social media has become an important channel for investors to obtain market information, and investors are increasingly concerned about how emotional signals obtained from social media affect the stock market. With the profound impact of social platforms on market sentiment, studying how to effectively use these real-time emotional signals has become a hot issue in the financial field.This review explores the use of social media sentiment analysis and contextual bandit models in stock market investment. With platforms like Twitter and Reddit impacting market sentiment, incorporating these signals into investment strategies offers a dynamic edge. Traditional strategies rely on historical data, but real-time sentiment provides valuable insights into market fluctuations.This paper examines the application of multi-armed bandit (MAB) algorithms, particularly their contextual variant, in financial decision-making. By leveraging social media sentiment, contextual bandit models balance exploration and exploitation to adjust investment portfolios dynamically. The review also addresses key challenges, such as handling noise and non-stationarity in sentiment data, and discusses how these methods can improve decision-making in uncertain market environments, ultimately helping investors maximize returns and minimize risks.