"GPU-Accelerated Research in Quant Finance: Using CUDA to Speed Up Backtests and Analytics"
This book is for quantitative researchers, systematic portfolio managers, and technologists who want to turn GPUs from a buzzword into a practical edge. It bridges the gap between theoretical quant finance and high-performance computing, showing how to move real research workloads—backtests, risk engines, and pricing libraries—from CPU-bound prototypes to production-ready GPU pipelines.
Readers will learn the mathematical and statistical foundations most relevant to GPU acceleration, then build a rigorous research and backtesting methodology that survives contact with real markets and regulators. The core chapters develop a working mental model of modern GPU architectures and the CUDA programming model, before introducing powerful patterns and libraries for Monte Carlo, PDE/FFT pricing, portfolio optimization, and risk analytics. Throughout, the focus is on trustworthy speedups: performance engineering, profiling, validation, and reproducibility.
The book assumes comfort with Python and basic quantitative finance, but no prior CUDA experience. All examples are designed for implementation in a modern research stack, with LaTeX-quality formulas and code that map cleanly onto Python/CUDA tooling. The result is a practical, end-to-end guide to designing faster research loops and more ambitious models without sacrificing transparency or control.
