Not so fast: Identifying and remediating slow and imprecise cryptocurrency exchange data

Abstract

Using raw tick-level trade data from 17 major cryptocurrency exchanges, we show that heterogeneity in matching engines can affect the computation of various liquidity and trading metrics. Using simple analytical techniques, we generate an algorithm to identify exchanges with slow matching engines or imprecise timestamps. Having identified problematic exchanges, we propose tractable techniques which can remediate the bias in metrics generated by problematic exchanges. Our techniques and exchange classifications are useful for academic and industry-based users of cryptocurrency exchange data to identify and remediate problematic trade-level data.