Relationship between Bitcoin’s Tail Risk and Attention Shocks

Abstract

This paper examines whether investor attention helps predict Bitcoin tail risk and whether derivatives-market conditions help transmit this relationship. Using daily data from 1 January 2021 to 1 January 2026, it combines BTC spot prices, Google Trends attention, funding rates, and liquidation data. Tail risk is described using a binary extreme-event indicator, a peaks-over-threshold (POT) model, and realized downside and upside semivariance. Extreme-return days are preceded by higher attention and much larger liquidation activity, while downside semivariance rises more sharply than upside semivariance in tail regimes. Logit estimates show that lagged attention remains a positive predictor of next-day extreme events after controlling for lagged returns, realized volatility, trading volume, funding, and liquidations. The paper also implements a partial identification step by constructing orthogonalized lagged attention shocks. These remain significant, although the full preferred design with explicit macro-news and regulatory-event controls is left for future work. Overall, the results suggest that investor attention is linked not only to average Bitcoin market behaviour but also to market fragility.