Accurate Inference from TAQ using RNBBO

Abstract

We present the Relative National Best Bid and Offer (RNBBO), a method for achieving precise trade-quote alignment in TAQ data by incorporating participant timestamps and accounting for inter-exchange latencies. Broadly applicable with minimal computational overhead, the RNBBO dramatically improves trade-signing precision, achieving over 92% accuracy on major U.S. exchanges, including Nasdaq, Cboe, and NYSE. By eliminating significant biases in liquidity metrics, our method challenges key conclusions in recent studies. These findings highlight the urgent need to revisit existing TAQ-based research and adopt the RNBBO in future studies to ensure more accurate and reliable insights.

Keywords: TAQ, Relative National Best Bid and Offer (RNBBO), Trade Classification, Liquidity, Replication

JEL Classification: G10, G18, K22, C15, G12, G20