Deep combinatorial optimisation for optimal stopping time problems: application to swing options pricing. (Q2094859)

In this paper, the authors propose an algorithm using reinforcement learning in order to solve optimal stopping times problem seen as an combinatorial optimisation problem. The method models directly the policy and does not need the derivation of a dynamic programming principle nor a backward stochastic differential equation. The stochastic optimization framework considered in this paper is described in Section 2. Numerical tests covering Bermudan and swing options are proposed in Section 4 and show good results in the pricing of 10 underlyings Bermudan option and also on 5 underlyings swing options having up to l = 6 exercise dates.