Information leakage between on-chain and off-chain markets can be measured with methods that capture direction, magnitude, and timing of information flow. Quantifying leakage helps regulators, exchanges, and researchers detect frontrunning, wash trading, or faster signal propagation that affects market fairness and environmental costs from increased transaction activity.
Statistical causality and information measures
Core techniques ask whether past variation in one market predicts future variation in the other. Granger causality developed by Clive W. J. Granger at the University of California, San Diego tests predictability in vector autoregressions and is widely used to detect lead-lag relationships. Mutual information traces back to Claude Shannon at Bell Telephone Laboratories and measures nonlinear dependence without assuming linearity. Transfer entropy introduced by Thomas Schreiber at the Max Planck Institute for the Physics of Complex Systems refines this by estimating directed, nonlinear information transfer and is robust to non-Gaussian returns, useful when transaction timestamps differ or distributions are heavy-tailed.
Market microstructure and price discovery
At the microstructure level, methods quantify which venue incorporates new information. Information share proposed by Joel Hasbrouck at New York University Stern School of Business decomposes contribution to price discovery using vector error correction models and is common in studies comparing centralized exchanges and on-chain decentralized venues. Variance-based spillover metrics such as the Diebold-Yilmaz spillover index by Francis X. Diebold at the University of Pennsylvania measure how shocks in one market transmit volatility to another, giving a system-wide leakage estimate. Hawkes-process models of order arrival and quoting intensity capture excitation and clustering of trades across markets and highlight mechanisms behind leakage, revealing whether on-chain events trigger concentrated off-chain reactions or vice versa.
Relevance, causes, and consequences
Causes of leakage include latency differences, privileged access to off-chain order books, algorithmic trading, and arbitrage through bridges and centralized custodians. Consequences range from faster concentration of liquidity and amplified volatility to cultural effects on participant trust and territorial regulatory responses when cross-border off-chain actors influence local on-chain prices. Empirical application requires careful timestamp alignment, microsecond-level data when possible, and robustness checks across linear and nonlinear measures to avoid spurious detection. Combining complementary methods yields the strongest evidence: statistical causality for timing, information-share or spillover indices for magnitude, and point-process models for mechanism. This multi-method approach supports transparent, evidence-based policy and platform design to reduce harmful leakage.