Title: Cross-section of expected cryptocurrency returns
Authors: Milan Ficura - University of Economics in Prague (Czech Republic) [presenting]
Gonul Colak - Hanken School of Economics (Finland)
Abstract: The predictive power of over 100 factors for the modelling of weekly cryptocurrency returns is analysed on a comprehensive dataset of almost 20 000 cryptocurrencies. The analysed factors include measures of momentum, illiquidity, investor attention, volatility, downside risk and systematic risk. The employed tools include univariate and multivariate portfolio-sorts, time-series regressions and Fama-MacBeth cross-sectional regressions. We show that in contrast to small and illiquid cryptocurrencies, the behaviour of large and liquid cryptocurrencies is increasingly similar to the behaviour of stocks. We identify a strong negative impact of past volatility on future cryptocurrency returns, similar to the low-volatility anomaly observed on the stock market. We further confirm the short-term momentum, long-term reversal, and short-term max-reversal effects on the cryptocurrency market, as well as the positive effect of S&P500 betas on future cryptocurrency returns. Nevertheless, the significance of these effects drops significantly in the multivariate regression tests once the low-volatility effect is taken into account.