Crypto factor zoo (.Zip)

Abstract

How many factors are genuinely needed to explain the cross-section of cryptocurrency returns? To answer this, we are the first to apply the alpha-based, iterative factor selection methodology of Swade et al. (2024), initially developed for equities, to the cryptocurrency market. Using a comprehensive set of 36 return-predictive factors, we find that just two to three factors can eliminate all significant portfolio alphas. The most influential factors include turnover volatility, bid–ask spreads, and blockchain-native metrics such as the new-address-to-price ratio. Liquidity-related variables dominate the selection process, appearing consistently across weighting schemes, model specifications, and periods.