Cryptocurrencies asset pricing via machine learning

In this paper, we study the cryptocurrency asset pricing via machine learning. The common pricing factors similar to equity asset pricing are constructed. The classical equity-based risk factors including size, momentum, and value to growth from the Fama-French three factor model are studied. For volatility risk factor category, we investigate realized volatility, skewness and jump. We also investigate liquidity factors including bid-ask, volume growth and Roll’s measure. We collect other crypto-unique factors. We show the machine learning models which could handle 30 characteristics together explain most of the excess return of cryptocurrencies.