Deep learning of value at risk through generative neural network models: The case of the Variational auto encoder

We present in this paper a method to compute, using generative neural networks, an estimator of the “Value at Risk” for a financial asset. The method uses a Variational Auto Encoder with an 'energy' (a.k.a. Radon-Sobolev) kernel. The result behaves according to intuition and is in line with more classical methods. • Estimation of the Value at Risk with generative neural networks; • No a priori assumptions on the distribution of the returns; • Good practical behavior.