Detecting Information Asymmetry in Dark Pool Trading Through Temporal Microstructure Analysis

Abstract

This research introduces a novel methodology for detecting information asymmetry in dark pool trading environments through temporal microstructure analysis. Dark pools, as non-displayed trading venues, create unique challenges for information asymmetry measurement due to their inherent opacity and lack of pre-trade transparency. Our approach applies a multi-dimensional framework that integrates heterogeneous autoregressive (HAR) modeling with behavioral autoregressive conditional duration (BACD) components to identify distinctive temporal signatures associated with informed trading. Analysis of 2.7 million dark pool transactions across five major venues reveals significant autocorrelation structures in trade clustering, order size distribution, and execution timing that correspond with subsequent price movements. The dual-stage attention-based neural network implementation demonstrates 91.6% accuracy in identifying information asymmetry events, outperforming traditional detection methods. Cross-venue information flow analysis reveals bidirectional but asymmetric information transfer between dark and lit markets, with approximately 37.2% of price discovery occurring in dark venues despite their lower trading volume. Principal dark pools exhibit consistently higher asymmetry levels compared to agency models, suggesting architectural influences on information environments. These findings provide valuable insights for regulatory frameworks, market design optimization, and trading strategy development while acknowledging limitations in participant identification and behavioral attribution.