Network Cohesion and Predictive Power of Social Media in the Bitcoin Market
46 Pages Posted: 9 Jan 2017 Last revised: 12 Feb 2019
Date Written: February 2019
Prior studies have shown that social media discussions can be helpful in predicting stock price movements. Given the large amount of social media data, how to effectively distinguish value-relevant information from noise remains an important question. We study this question from a network perspective and investigate the role of discussion network cohesion in affecting its return predictability. Building on theoretical arguments of information correlation, information free-riding, and hidden profile effect, we hypothesize that less cohesive social media discussion networks are more accurate in predicting future price movements. We empirically test our hypothesis using data from the Bitcoin market. Due to its highly speculative nature, social media discussions are likely to play a more significant role, and it is thus an ideal setting for our analyses. We find results consistent with our hypothesis, and they are robust after controlling for traditional media coverage and topic coverage in social media discussions.
Keywords: financial technology, social media analytics, network cohesion, Bitcoin, topic modeling
Suggested Citation: Suggested Citation