Mathematical Finance, Stochastic Analysis, and Machine Learning Seminar by Thorsten Schmidt: Affine models under uncertainty and robust deep hedging

Speaker

Thorsten Schmidt, Department of Mathematical Stochastics, University of Freiburg

Title

Affine models under uncertainty and robust deep hedging

Abstract

Model uncertainty is certainly present everywhere, particularly in finance. We develop a tractable class of models with uncertainty based on affine processes. The approach is based on Knightian uncertainty and derives the associated non-linear expectations and a variational Kolmogorov equation. For practical applications, we discuss the hedging problem: first, we show how to estimate confidence intervals and incorporate these results into pricing. Regarding hedging, we develop a deep neural network that solves the problem efficiently in a Bayesian setting.

Mathematical Finance, Stochastic Analysis, and Machine Learning