We apply the general method for hedging a portfolio of derivatives introduced by Buehler, Gonon, Teichmann, and Wood (2019) to the specific problem of hedging a short position in an Asian call option. In particular, we compare the performance of feedforward and recurrent neural network architectures in approximating the hedging strategy, through experiments with simulated data.
These experiments are part of an undergraduate dissertation (https://www.overleaf.com/read/pftggckgfzqg) with summary slides (https://www.overleaf.com/read/rwnnrdbxcgtv).