Victor Duarte - Gies College of Business UIUC - Champaign Illinois
MACHINE LEARNING FOR CONTINUOUS-TIME FINANCE
This paper proposes an algorithm for solving a large class of nonlinear continuous-time models in finance and economics. First, I recast the problem of solving the corresponding nonlinear partial differential equations as a sequence of supervised learning problems. Second, I prove that the computational cost of evaluating the exact continuous-time Bellman residuals does not increase in the number of state variables, allowing for the solution of high-dimensional problems. To illustrate the method, I solve canonical asset pricing models featuring recursive preferences, endogenous labor supply, irreversible investment, and state spaces with up to ten dimensions.