Semi-Analytical Latency-Aware Distributional Inference for Limit Order Books

Semi-Analytical Latency-Aware Distributional Inference for Limit Order Books

• Autor
• Bernardo Teixeira
• Co-autores
• Renan Avila , Oswaldo Costa , Leandro Maciel
• Resumo

• In modern electronic markets, execution at millisecond horizons requires

  fast, state-conditional probabilities for price moves and passive fills. This creates a

  need for models that are both computationally tractable and directly interpretable in

  execution settings. We study whether the semi-analytical framework of Cont, Stoikov,

  and Talreja (CST) can provide such signals in Brazilian equities. First, we present the first

  CST-based analysis of latency effects in this market, incorporating reaction delays into

  market-making probabilities. Second, we assess CST adherence by comparing model-

  implied and realized event distributions. We also introduce the BTG-OBD-A26 dataset,

  a reproducible message-level order-book-dynamics dataset for Brazilian equities. Our

  results show that the CST framework captures directional and fill probabilities reasonably

  well, whereas the latency extension is less accurate because the model does not capture

  temporal clustering in order flow.

• Palavras-chave
• Financial engineering, Limit order book, Market microstructure, Fill probability, Mid-price direction, Continuous-time Markov chain, Latency, Market making
• Modalidade
• Comunicação oral
• Área Temática
• Engenharia Financeira

Voltar Download