dupire

Low-latency options pricing engine in Rust. BSM, Black-76, Heston, Bates (jumps), Local Vol (Dupire), Monte Carlo (Euler/Andersen QE). Adaptive Gauss-Kronrod CF pricers, full analytic Greeks, forward-mode AD (incl. jump sensitivities), Halley IV solver, LM/DE global calibration, no-arbitrage repair, Rayon parallelism. CI + clippy, 0 warnings.

Gaussian process PDE solvers that combine linear PDEs with linear inequality constraints and solve complex-valued Helmholtz field problems.

This project aims to strip the Equity Local Volatility surface and implement the associated PDE pricing method.

SVI vol surface, Dupire local volatility (PDE) and Heston calibration, with cross-model repricing on real SPY options

End-to-end implied and local volatility surface from live SPY options, with Dupire local vol and an empirical study of the VIX term structure as a predictor of realised volatility.

End-to-end implied and local volatility surface from live SPY options, with Dupire local vol and an empirical study of the VIX term structure as a predictor of realised volatility.

Implied and local volatility surface from live SPY options, with Dupire local vol and an empirical study of VIX term structure vs realised vol

Dupire local volatility from a real SPY surface, and what flat-vol pricing costs on a barrier option: 32bp of spot. Live in-browser PDE.

Local volatility calibration from implied volatility surfaces using Dupire’s formula, with Monte Carlo pricing of vanilla and exotic options.

Arbitrage-free implied volatility surfaces on SPY options: raw SVI, constrained SVI and SSVI, butterfly/calendar arbitrage audit, Dupire local vol and barrier pricing under local vs flat vol. Tested, CI on Python 3.11–3.13.