New arXiv Paper on No-Arbitrage Implied Volatility Surface Generation
Get up to speed
Researchers including Oscar Brooks have released version 3 of a no-arbitrage implied volatility surface generation framework based on latent flow matching. The paper has been accepted for presentation at the 7th ACM International Conference on AI in Finance (ICAIF '26), held in Milan from November 14 to 17, 2026. The framework first compresses high-dimensional volatility surfaces into a low-dimensional latent space using a variational autoencoder regularized with differentiable penalties for calendar spreads, call spreads, and butterfly arbitrage. A flow matching model then learns to map a Gaussian prior to the empirical latent distribution, which is finally decoded back to volatility surfaces. The paper reports that this method generates 90.8% of surfaces satisfying all tested static no-arbitrage conditions, performs best in the extreme Q99 range, and balances distribution similarity, tail preservation, and financial consistency without post-sampling reweighting.
AI-generated from reports · 3 hr agoUpdated
Coverage timeline
Follow the coverage to understand different sides of the event.
- arXiv q-finA Latent Flow Matching Method for Arbitrage-Aware Implied Volatility Surface Generation
This paper proposes an arbitrage-aware latent flow matching framework for unconditional generation of implied volatility surfaces. The framework first compresses high-dimensional surfaces into a low-dimensional latent space using a variational autoencoder with differentiable penalty regularization for calendar spreads, call spreads, and butterfly arbitrage. A flow matching model then learns to map a Gaussian prior to the empirical latent distribution, which is finally decoded back to a volatility surface. This method can generate 90.8% of surfaces that satisfy all tested static no-arbitrage conditions, performs best in the extreme Q99 range, and balances distributional similarity, tail preservation, and financial consistency without post-sampling reweighting.
Trend of this event
Not enough consecutive observations yet; trend is not drawn.