Boolean Programming for Optimal Stopping in American Option Pricing with Reinforcement Learning

In this paper, we analyze the effective implementation of Reinforcement Learning for American Option Pricing. We treat American Option Pricing problems as Optimal Stopping Problems and apply Monte-Carlo Simulation. We solve the Stopping problems in American Option Pricing using Q-Learning and Reinforcement Learning. We analyze the details of Function Approximation and treat the Stopping Problem with Boolean Programming. By applying Boolean Programming, the problem can be treated with convex optimization, making the numerical algorithms more efficient. We conduct Numerical Experiments using the codes developed in the references using Matlab and Python. Through simulation experiments using our approaches of Reinforcement Learning and Boolean Programming, we obtain good results in finding solutions for Optimal Stopping in American Options Pricing.