Hawkes-Driven OTC Market Making: Volterra–Riccati Approximation

Battle Count: 50

Directly addresses OTC market making, which is a core quantitative trading problem. The paper provides an implementable state-feedback quote rule that incorporates persistent RFQ memory, validated against exact benchmarks. The endogenous OTC market impact mechanism and inventory risk reduction are directly applicable to dealer pricing systems. The modular Riccati layer can be paired with various request-flow models. Particularly relevant for FX dealers, fixed-income market makers, and any RFQ-based venue. The practical message about conditioning quotes on recent RFQ history is immediately actionable.

Requires solving backward Riccati ODE systems driven by conditional Volterra forecast curves, computing Hawkes resolvent responses, implementing state-feedback updates after each RFQ observation, and handling multi-scale exponential mixtures or power-law kernels. The exponential benchmark is tractable (1D memory state), but general kernels require discretized forecast curves (N_f grid points) and careful handling of the Volterra integral equation. Monte Carlo validation with common random numbers adds computational cost. The quote optimizer tabulation and sigmoid win probability add moderate complexity. No open-source implementation provided.

Numerical parameters for validation experiments are fully specified (Table 2, Section 5.1). The exponential Hawkes benchmark is self-contained and reproducible. However, the empirical motivation uses proprietary HSBC spot-FX RFQ data (anonymized, >500K events per pair over 3 years) which is not publicly available. The power-law experiment parameters are given but require custom implementation of Volterra-Riccati solvers. No code repository is provided.