Deep stochastic optimization in finance

This paper outlines, and through stylized examples evaluates a novel highly effective computational technique in quantitative finance. Empirical Risk Minimization (ERM) neural networks are key to this approach. Powerful open source optimization libraries allow for efficient implementations of algorithm making it viable high-dimensional structures. The free-boundary problems related American Bermudan options showcase both the power potential difficulties that specific applications may face. impact size training data is studied simplified Merton type problem. classical option hedging problem exemplifies need market generators or large number simulations.