Predictive role of online investor sentiment for cryptocurrency market: Evidence from happiness and fears

摘要

We examine the predictive ability of online investor sentiment for six major cryptocurrency returns. For this, we use two proxies, the FEARS index of Da et al. (2015) and Twitter Happiness sentiment, applying the bivariate cross-quantilogram of Han et al. (2016). Happiness sentiment index significantly predicts Bitcoin return as well as other major cryptocurrencies at the two extreme states of the market and for extreme levels of sentiment. Hence, investors should readjust their portfolios according to the market sentiment and limit their decision on the safe-haven property of Bitcoin. As to FEARS, predictability also exists but is rather pronounced for a low level of sentiment. Overall, Happiness sentiment reveals to be a persistent and robust predictor for most cryptocurrency returns. FEARS index also shows significant predictability of returns, but the predictability is weaker and mainly in the short-term. In summary, our findings provide evidence that online investor sentiment is a significant nonlinear predictor for most major cryptocurrencies returns, suggesting though the superiority of Twitter to Google-based online investor sentiment proxy. Moreover, cryptocurrency returns seem to be driven more by sentiment transmitted through social media than with macroeconomic news, which is in line with the nature of cryp-tocurrency participants, mainly young individuals computer enthusiasts.