CVA and IM: welcome to the machine

Henry-Labordere proposes a neural networks-based technique to price counterparty risk and initial margin

CLICK HERE TO VIEW THE PDF

Building heavily on recent work, Pierre Henry-Labordère introduces a primal-dual method for solving backward stochastic differential equations based on the use of neural networks, stochastic gradient descent and a dual formulation of stochastic control problems. The algorithm is illustrated using two examples relevant to mathematical finance: the pricing of counterparty risk and the computation of initial margin

Solving numerically high-dimensional, non-linear