A User-Centric Analysis of Social Media for Stock Market Prediction

• DOI: 10.1145/3532856
• Publication Date: 2022-09-29T11:47:05Z
• Authors: Mohamed Reda Boua...; Scott Sanner; Ga Wu
• Keywords: Stock (firearms); Sentiment Analysis; User-Generated Content

Social media platforms such as Twitter or StockTwits are widely used for sharing stock market opinions between investors, traders, and entrepreneurs. Empirically, previous work has shown that the content posted on these social can be leveraged to predict various aspects of performance. Nonetheless, actors may not always have altruistic motivations instead seek influence trading behavior through (potentially misleading) information they post. While a lot sought analyze how market, there remain many questions regarding quality predictions active users platforms. To this end, article seeks address number open research questions: Which platform is more predictive performance? What actually predictive, over what time horizon? How does posting vary among different users? Are all trustworthy do some user’s consistently mislead about true movement? answer questions, we analyzed data from covering almost 5 years messages spanning 2015 2019. The results large-scale study provide important insights which present following: (i) source than Twitter, leading us focus our analysis StockTwits; (ii) StockTwits, users’ self-labeled sentiments correlated with but only slightly in aggregate short-term; (iii) at least three clear types temporal 144 days horizon: short, medium, long term; (iv) incorrect who reliably wrong tend exhibit conjecture “botlike” post their removal tends improve content.