Determinants of price discovery in cryptocurrency option markets: an interpretable machine learning perspective

This paper provides the first systematic evidence on price discovery in cryptocurrency option markets. Using one-minute BTC and ETH options data, we estimate the Information Leadership Share (ILS) and analyze its determinants via interpretable machine learning. We find that option-implied prices drive approximately 97\% of price discovery for BTC and 96\% for ETH, significantly exceeding equity market benchmarks. Tree-based ensembles outperform linear and neural network models, revealing strongly non-linear determinants. Specifically, risk-neutral entropy, volatility-surface curvature, skew, and higher-order moments are primary drivers of ILS, while maturity and Greek-based leverage play minor roles. These results establish cryptocurrency options as the dominant venue for information incorporation, surpassing underlying spot markets. Our findings highlight the unique informational efficiency of crypto derivatives and identify key nonlinear factors governing their price leadership.