This study explores the existence and dynamic of herding behaviours in the cryptocurrency market by analysing the return dispersion of Bitcoin and major altcoins. It assesses whether herding intensifies during the period of extreme market returns or heightened volatility or elevated investor sentiment.
Using return data of Bitcoin and 25 major altcoins from 2020 to 2024, the study applies dispersion-based models, Cross-sectional Standard Deviation (CSSD) and Cross-Sectional Absolute Deviation (CSAD), to detect herding. It further incorporates regime-specific variables such as volatility and sentiment data from Google Trends and panel regression is used to distinguish between herding behaviour in Bitcoin and altcoins. It further extends the CSAD model with volatility- and sentiment-interaction terms, employs panel regressions to compare Bitcoin and altcoins and applies rolling-window estimations to capture dynamic herding intensity over time. Additionally, crisis-specific regressions test whether herding strengthens during major events in the research time period.
The results provide weak evidence of herding under the condition of extreme market states, opposite to classical hypotheses. Herding fails to intensify under high volatility but gets more dispersed. Sentiment proxies from Google Trends are positively associated with return divergence, indicating anti-herding behaviour. Cross-sectional evidence shows that altcoins are more heterogeneous and speculative than Bitcoin, while short-lived herding appears only during crisis episodes.
The finding challenges the conventional behavioural finance assumptions in the context of the digital asset market. It suggests that cryptocurrency investors behave more heterogeneously and divergently under pressures, which have implications for market risk assessment, algorithmic trading designs and regulatory oversights. This paper is among the few to empirically investigate herding behaviour in the cryptocurrency market using both the traditional dispersion approach and modern sentiment-based analysis. It contributes to the behavioural finance literature by highlighting anti-herding dynamic and heterogeneity among digital assets and offers new insights into investors' psychology in decentralized and speculative environment.
