I am interested in analysis, probability theory and generative machine learning algorithms. In my honours project I will be using deep-learning to train a hedging strategy on generated stock sample paths. A VAE will be trained on path signatures of existing stock paths, which will then generate new path signatures, from which novel sample paths will be recovered.

 

The project will allow me to explore the use of path signatures in data generation, which can be especially useful when there is minimal data available for more traditional deep-learning techniques.

 

Co-supervised by Mr Yuri Robbertze.