A Latency Commentary: Why Dynamic RBBO Outperforms Fixed Latency Adjustment

Abstract

Researchers face two competing methods for adjusting latency in NYSE TAQ: Holden et al. (2023)'s dynamic Relative Best Bid and Offer (RBBO) and Schwenk-Nebbe and Thimsen (2024)'s fixed adjustment. We show that exchange latency is dynamic, making fixed adjustments unreliable and prone to trade-signing errors. Correctly benchmarking the two methods confirms RBBO's superior accuracy. Moreover, Holden et al. (2023) offers a simplified Latency Timestamp Adjusted (LTA) method, making dynamic adjustments more accessible. Fixed-latency approaches require constant revisions and introduce biases, underscoring the necessity of dynamic adjustments for robust market microstructure research.