Deep Learning Approaches for Pricing Options in Stochastic Volatility Models

Efficient computation of option prices is essential for making quick trading decisions. This paper investigates the use of deep learning to expedite the accurate calculation of European option prices within a local volatility framework that utilizes five parameters. We compared the predictions of the deep neural networks against results obtained from the Monte Carlo method across various scenarios. Our numerical experiments indicate that the approximation network achieves satisfactory accuracy. The network performs exceptionally well within the core region of the parameter domain.