Research

We propose a novel approach to improve short-term equity return predictability by analyzing truncated high-frequency return distributions. We segment returns into core and tail components, focusing on core and tail asymmetries — differences between `typical' or `extreme' upside and downside variances. Our empirical findings show that these predictors achieve an in-sample adjusted R^2 of about 7% and an out-of-sample R^2 exceeding 3% for one-month-ahead market return forecasts, outperforming traditional predictors such as valuation ratios and the variance risk premium. It is the core asymmetry, rather than the tail asymmetry that drives the predictive power.

The cryptocurrency market operates continuously, leading to frequent price fluctuations and information dissemination. This can hinder investors from reacting promptly to market changes, a phenomenon attributed to investors' limited attention. Research in traditional markets shows that the limited attention bias allows successful implementation of momentum strategies. However, past research on cryptocurrency markets finds mixed results. To resolve the puzzle, we utilize a survivorship bias-free dataset while accounting for variations in market capitalization and trading volume. This differentiation is crucial given young and tech affine retail investors' inclination toward smaller-capitalized cryptocurrencies, due to their higher risk tolerance and limited attention. More risk averse investors such as institutional investors, in contrast, focus more on top cryptocurrencies. In line with expectations, we find effective momentum strategies among larger-capitalized cryptocurrencies.

Crypto-assets constitute a large and heterogenous asset class. Only a minority of crypto-assets qualify as crypto-currencies, while the large majority provide services that go beyond Bitcoin’s peer-to-peer payment system pioneered by Nakamoto (2008). This paper introduces a comprehensive taxonomy spanning crypto-currencies, service crypto-assets, smart-contract platforms, distributed applications and tokenized assets (including both fungible and non-fungible tokens). To illustrate stylized facts and capture various risks of the cross-section of crypto-assets this paper relies on a novel dataset covering more than 14,000 crypto-assets.