Deep Hedging: Learning to Simulate Equity Option Markets

TL;DR

This paper introduces a novel approach using GANs to simulate equity option markets, enabling better training data for trading strategies and demonstrating the first successful application of GANs to multivariate financial time series.

Contribution

It presents the first successful application of GANs for generating multivariate financial time series, specifically equity option market simulations, outperforming classical methods.

Findings

GAN-based generators outperform classical methods on benchmark metrics

Adversarial training achieves the best simulation performance

Recurrent and temporal convolutional architectures are effective

Abstract

We construct realistic equity option market simulators based on generative adversarial networks (GANs). We consider recurrent and temporal convolutional architectures, and assess the impact of state compression. Option market simulators are highly relevant because they allow us to extend the limited real-world data sets available for the training and evaluation of option trading strategies. We show that network-based generators outperform classical methods on a range of benchmark metrics, and adversarial training achieves the best performance. Our work demonstrates for the first time that GANs can be successfully applied to the task of generating multivariate financial time series.