CPU/GPU-Accelerated Jump Diffusion HJB Equations: A Comparative Study for Low-Latency Crypto Market Making with Order Flow Toxicity Tracking
A comparative analysis of CPU and GPU implementations for solving jump-diffusion Hamilton-Jacobi-Bellman equations in high-frequency cryptocurrency market making. Our approach incorporates inventory risk management, market impact modeling, and order flow toxicity tracking, demonstrating up to 43% higher profitability in volatile market conditions.
