Funding-Aware Optimal Market Making for Perpetual DEXs

Battle Count: 80

Directly addresses optimal market making in perpetual futures, one of the most actively traded crypto derivatives. The funding-aware HJB provides a concrete, implementable quote policy that improves mean performance and reduces inventory risk versus classical Avellaneda-Stoikov. The methodology is immediately applicable to liquidity provision on perpetual DEXs (Hyperliquid, dYdX, GMX) and CEX perpetual markets. The paper bridges stochastic control theory with practical DeFi market structure.

Requires implementing a two-dimensional (inventory x funding) finite-difference HJB solver with monotone birth-death discretization, CFL condition enforcement, backward time-stepping, and quote recovery from value-table differences. Additionally requires: (1) OU calibration with MLE on hourly funding data, (2) exponential fill-intensity calibration from crossed fills joined to L2 snapshots, (3) event-driven one-minute backtest simulator with hourly funding application, (4) multi-seed Monte Carlo evaluation with paired comparisons. Numba-accelerated bilinear interpolation is used for real-time quote lookup. The mathematical derivation is non-trivial but well-documented.

The paper provides detailed parameter specifications (alpha, phi per asset), grid dimensions (Nt=2048), CFL conditions, calibration procedures, and holdout window (26 Nov 2025 - 31 Dec 2025). However, no explicit GitHub repository or code link is provided. The methodology is fully described mathematically, and Hyperliquid data is publicly accessible, but reproducing the exact backtest requires implementing the HJB solver, fill calibration, and simulation pipeline from scratch.