New neural operator slashes financial model calibration time

Researchers have developed a novel neural operator designed to significantly speed up the calibration process for local-stochastic volatility (LSV) models in quantitative finance. This new method, implemented using Deep Operator Networks (DeepONet) and Fourier Neural Operators (FNO), can perform calibration in milliseconds, a drastic reduction from the previous 98.5 ms. The projection-consistent operator ensures static-arbitrage constraints and improves accuracy, reducing local-volatility root-mean-square error by 36% and leverage root-mean-square error by 7-16% in synthetic tests. AI

IMPACT Accelerates complex financial modeling, enabling faster risk assessment and trading strategy development.

RANK_REASON Academic paper detailing a new methodology and benchmark results. [lever_c_demoted from research: ic=1 ai=0.7]

Read on arXiv cs.LG →

• Deep Operator Network
• Fourier Neural Operator

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →