Strategic OTC market making with reputation feedback

Battle Count: 61

Highly relevant to quantitative trading, particularly for electronic OTC market making in FX, fixed income, and commodities. The paper directly addresses how reputation feedback (win ratios, fill ratios) affects optimal quoting strategies, creating trade-offs between spread capture and franchise value. The bistability results have direct implications for dealer strategy: the same technology and risk appetite can lead to very different outcomes depending on initial reputation. The last-look/rejection threshold optimization is directly applicable to streaming FX market making. The campaign vs. monetization phase distinction is practically actionable for algorithmic quoting systems.

High complexity. Requires: (1) solving a 2D HJB equation with fast inventory and slow reputation states; (2) implementing the adiabatic decomposition with frozen-score ergodic problems; (3) quadratic BEGV approximation with Riccati equation for inventory slope; (4) 2D policy iteration on a 100x100 score grid with 100-point threshold and spread grids; (5) computing invariant measures of the fast inventory process; (6) Lambert W function evaluation for RFQ controls; (7) numerical optimization over rejection thresholds for streaming. The full pipeline involves multiple nested numerical procedures. However, the paper provides closed-form expressions for key components (RFQ optimal quote, streaming conditional quote) that simplify implementation.

The paper provides detailed parameter tables (Table 1), explicit functional forms for win/fill probabilities, gate functions, and the full HJB formulation. Numerical methods (policy iteration, grid sizes, interpolation) are described. However, no code or repository is provided. The Lambert W closed-form solution for RFQ and the grid-based optimization for streaming are specified. Reproduction would require implementing the adiabatic decomposition, BEGV quadratic approximation, and 2D policy iteration from scratch.