This study utilizes intraday price data of Bitcoin, Ethereum, and Ripple to investigate how sensitive cryptocurrency returns are to higher-order realized moments (i.e., variance, skewness, kurtosis, hyper-skewness, hyper-kurtosis), and whether such sensitivity, if any, varies across bear and bull market conditions. We also evaluate the forecasting power of higher-order moments for future cryptocurrency returns. The empirical analysis draws on a quantile regression approach, after orthogonalizing raw returns with respect to a diverse set of global influences and risk factors. The results reveal that all moments up to the fifth order are generally relevant to explaining cryptocurrency returns, but with different degrees, depending on both the type and state of the cryptomarket. Moreover, both skewness and hyper-skewness show statistically significant predictive capabilities, whether in-sample or out-of-sample, for subsequent returns. Our evidence provides practical implications for asset pricing and risk management decisions.
Read full abstract