Neural Networks for Option Pricing

Sammanfattning

The task of pricing options is one with many different solutions, and overtime more complicated models of the markets have been developed in anattempt to replicate assets more accurately. In this thesis we investigate theuse of neural networks for pricing within these models. As speed is of equalconcern for option pricing, the offline training period of neural networks isa desirable property for this use. Implementing neural networks is not assimple as it may seem though and as we find out, many adaptations havebeen made to try and rectify the issues this implementation introduces. Welook more closely at methods where neural networks are used for calibrationof model parameters.