Option Pricing Using Machine Learning

The thesis focuses on option pricing using neural networks. The goal is to usestate-of-the-art recurrent neural network for option pricing and comparing it’s performanceto Black-Scholes model and another neural network architecture. The readeris in theoretical part, provided with foundations of Black-Scholes model and machinelearning. We also introduce and describe in great detail feed-forward networks andrecurrent neural networks with Long short-term memory units. Subsequently, in thepractical part we present our data and describe training of neural networks. Furthermore,we present our proposed approach for pricing options using LSTM networks.In addition, we present detailed technical aspects and software use for training. Thelast part is dedicated to results obtained by all models. In conclusion, we suggestwide range of improvements and ideas for further research.