Coskewness and the short-term predictability for Bitcoin return

The topic is of particular interest given the asymmetric relationship documented by numerous studies between the stock market and Bitcoin, yet the quantification of this asymmetry and its implications for Bitcoin excess return have been less explored. This study investigates the role of coskewness (asymmetry) between the stock market and Bitcoin returns in predicting the excess returns of Bitcoin. Drawing on data from January 2014 to March 2021, this study uses empirical analysis to demonstrate that the daily coskewness between Bitcoin excess returns and stock market index excess returns significantly forecasts Bitcoin's following-day excess returns. Significantly, this is true for both the in-sample and out-of-sample performances. For the forecast evaluation, we employed the Roos2 criterion, and the test used for its evaluation was based on the methodology proposed by Clark and West (2007). The robustness tests further validate the importance of coskewness as an informative indicator for forecasting excess Bitcoin returns. The subsequent analysis illustrates that incorporating coskewness as an added predictor in the univariate factor model can improve the forecasting of returns and economical significance. Consequently, these findings offer insightful implications for investors who aim to hedge risks and develop informed investment strategies. For example, considering coskewness can lead to a utility gain of 0.128% in a univariate factor model.