- BIMSA 数字经济实验室讨论班

Neural Importance Sampling for Option Pricing with Normalizing Flows

组织者

彼尔·约翰逊 , 高瑞泽 , 韩立岩 , 李振 , 刘瑾 , 龙飞 , 史冬波 , 汤珂 , 严兴 , 张琦

演讲者

严兴

时间

2026年06月12日 15:00 至 16:00

地点

A3-2-303

线上

Zoom 815 762 8413 (BIMSA)

摘要

The estimation of expectations for pricing complex financial derivatives presents a fundamental challenge for Monte Carlo methods, which often suffer from high estimator variance. To address this issue, we propose a novel importance sampling method for variance reduction that uses conditional normalizing flows to learn highly complicated proposal distributions. Our method is designed to approximate the theoretically optimal proposal density, while being explicitly conditioned on certain parameters. We construct a two-stage flow architecture featuring a deep conditional Inverse Autoregressive Flow (cIAF). The model is trained effectively using a loss function that is entirely unaffected by the unknown normalizing constant. We demonstrate the effectiveness of our approach on a suite of challenging exotic option pricing problems, including barrier, Asian, and basket options. Experiments show that our method substantially reduces estimator variance and significantly improves sample efficiency.

演讲者介绍

I have been an Associate Professor at BIMSA since 2025. Prior to this role, I was an Assistant Professor at the Institute of Statistics and Big Data, Renmin University of China. My research lies at the intersection of AI and finance/business, focusing on FinTech and Business Analytics through innovative machine learning and data science methodologies. My interests include tail risk management, empirical asset pricing, portfolio optimization, derivatives, consumer credit, and related areas. Recently, I have also developed an interest in out-of-distribution (OOD) generalization and uncertainty quantification in machine learning. I publish in both finance/business and machine learning academic journals and conferences.

I am seeking highly self-motivated Postdoctoral researchers or research interns to conduct high-quality research in the areas of AI, digital economy, or applied mathematics. If you are interested, please feel free to contact me.