GPU-Accelerated KDE in Quantitative Finance

NVIDIA Developer · youtube · 2026-07-18

The video presented a use case in quantitative finance: stock return distributions are rarely normal, so fitting them using methods like KDE provides a more realistic approximation.

The focus isn't on the financial conclusions themselves, but rather the computational implementation. While calculating KDE on a CPU is relatively slow, moving the workload to a GPU significantly accelerates the process. A companion notebook is available at github.com/will-hill/GPU-Quant-Finance.

Original post →

More from Infra

Infra channel →