The asymmetric influence of event-based news on cryptocurrency volatility: a survival analysis study of jump and co-jump behaviour in cryptocurrency markets

Tembo, Thokozile Mirriam, Ma, Tiejun, McGroarty, Frank and He, He (2025) The asymmetric influence of event-based news on cryptocurrency volatility: a survival analysis study of jump and co-jump behaviour in cryptocurrency markets. World Finance Conference, University of Malta, Valetta, Malta. 28 - 31 Jul 2025.

Abstract

This paper examines how cryptocurrency-specific news sentiment influences the timing and occurrence of volatility jumps in cryptocurrency returns, offering a behavioural framework that distinguishes between polarity-driven and state-contingent effects. Using ten years of hourly Event Sentiment Scores (ESS) from RavenPack, we analyse a panel of 87 cryptocurrencies and detect idiosyncratic and systemic jumps in volatility returns using the Bipower Variation method. These discontinuities are modelled using Cox Proportional Hazards estimation to assess how sentiment affects the hazard rate of both jump types.

Our results uncover a clear polarity-based asymmetry under calm conditions: mildly positive and negative sentiment both accelerate jump risk, but optimism exerts a stronger effect—consistent with behavioural theories of speculative momentum. However, this asymmetry disappears under systemic stress. During co-jump periods, sentiment effects are symmetrically amplified, suggesting that market synchronisation lowers behavioural thresholds and increases the volatility hazard irrespective of sentiment direction. At the systemic level, both positive and negative sentiment significantly raise the hazard of co-jumps, but without statistically significant asymmetry in polarity.

These findings introduce the concept of contagion asymmetry, where the amplification of sentiment-induced risk is not driven by emotional valence but by the system’s vulnerability to synchronised behaviour. Robustness checks across thresholds, coin tiers, time periods, and systemic conditions confirm that the strength and shape of sentiment effects are highly conditional. Methodologically, our use of survival analysis allows for more precise behavioural interpretation by capturing not only whether sentiment matters, but when. This study extends the sentiment-volatility literature by reframing co-jumps as state-contingent contagion events and highlights the importance of timing in modelling crypto-market instability.

This record has no associated files available for download.

Contributors

Author: Thokozile Mirriam Tembo

Author: Frank McGroarty

Download statistics