Download
Abstract
This paper examines the cross-sectional predictive ability of text-based fundamental factors in the cryptocurrency market. Using BERT topic modeling, we construct a Fundamental Sentiment Index (FSI) from news articles related to cryptocurrencies. We find that cryptocurrencies with high exposure to fundamental sentiment, typically payment and platform tokens, earn a risk premium, while governance tokens hedge against fundamental risks. Currency betas correlate with blockchain value metrics, such as the ratio of users and transactions to market cap. A long-short strategy based on FSI exposure delivers statistically significant returns, highlighting the importance of fundamental sentiment in explaining cryptocurrency returns beyond traditional factors.