Research On The Relationship Between Social Network Sentiment And Bitcoin Returns

With the development of technologies,the way of people how to get information and how to share their emotion changes a lot,social network has becomes a basic and essential part of our daily life.Therefore,try to measure investor sentiment though the social network sentiment is a trends in the study of behavior finance.Bitcoin,which always regard as the first token of the block chain industry,have the most powerful influence and important place.It has becomes a popular investment spot in recent years and contains both the attributes of internationalization and decentralization.This thesis' s purpose,is to discuss the relationship and granger causality between investor sentiment and bitcoin returns,whereas choosing social network sentiment represents the investor sentiment.We pick up DTH which means Daily Twitter Happiness as the index of investor sentiment data,using both the mean and quantile regression model to conduct our research,also comparing these two different models while doing the Granger causality tests.The empirical findings suggest that,despite the relationship between social network sentiment and bitcoin return is not significant under naive mean Granger causality test,but in some particular quantile intervals,especially the high quantile intervals,social network sentiment do have the relationship between bitcoin return.The conclusion is the same when conduct the granger causality test.The social network sentiment is the significant granger causality reason of bitcoin return,while the result can only be find in one-way,bitcoin return is not the granger causality reason of social network sentiment.When watching the regression coefficient,we can also find that the more unhappier the DTH is,which means negative mood such as fear and angry,the more possibility of bitcoin return increase is happen.The reason can be conclude to the hedge attribute of bitcoin.This results suggest that in the extreme emotion intervals,bitcoin returns also changes violently.That means the bitcoin returns can be influenced by the random emotional impact,therefore,the investment of bitcoin have a higher random risks.Quantile regression model keeps the extreme point,which means have a higher fitness with the characteristic of bitcoin price's high volatility.This model is a choice of methods which can be used in the research of other investment target that has similar attributes.