Finance-Insurance
Time: 15.00 am
Date: 24th of April 2025
Room 3001

Nicolas BARADEL (Ecole Polytechnique & INRIA) “Deep learning for pricing and hedging European options in incomplete markets.”

Abstract : We propose a deep learning framework for pricing and hedging European options in incomplete financial markets characterized by stochastic volatility and correlation. We introduce a single deep neural network that simultaneously estimates the option price and its hedging strategy, leveraging the self-financing condition. The network’s output provides the price, while its gradient with respect to the underlying assets yields the hedging strategy. This direct approach delivers results comparable to applying Black-Scholes hedging. We propose an approach with a control variable on the terminal condition, making it endogenous. Numerical experiments demonstrate that our method outperforms the Black-Scholes hedging strategy, achieving more concentrated profit and loss distributions. While inspired by finance, these methods can be applied to asset-liability management for life insurers, particularly for valuing their guarantees.

Organizers:  Jean-David FERMANIAN