Paper: Forecasting Implied Volatility Surfaces with Latent Diffusion Models

chaumian · x · 2026-08-25

A new paper titled 'Arbitrage-Aware Multi-Step Forecasting of Implied Volatility Surfaces' proposes a conditional latent diffusion framework. It generates joint 30-step trajectories of implied volatility surfaces and underlying returns. An arbitrage-aware autoencoder learns a low-dimensional surface representation, while the diffusion model captures conditional joint evolution. Evaluated on SPX data, the framework generates realistic probabilistic scenarios while outperforming the persistence benchmark in point forecasting.

Original post →

More from Research

Research channel →