Extreme Investor Sentiment and Herding in the Cryptocurrency Market: Empirical Insights from Music-Based Sentiment Data

Abstract

Purpose: This study analyzes herding behavior in the cryptocurrency market and the moderating role of extreme investor sentiment.
Design/methodology/approach: Investor sentiment is measured using emotional data derived from 200 popular songs on Spotify, combined with market data from 361 cryptocurrencies during the 2017–2023 period. Herding behavior is identified through the Cross-Sectional Absolute Deviation (CSAD) index, employing both static regression and regime-switching models.
Findings: The findings reveal a significant presence of herding, especially during periods of high market volatility. The static model shows that negative sentiment amplifies herding behavior, whereas positive sentiment has no clear effect. In the regime-switching model, sentiment appears statistically insignificant, implying that market volatility is the dominant factor.
Originality/value: The use of music-based emotional data offers a novel approach for capturing investor sentiment and irrational behavior in digital asset markets.