We are going to present a method for valuing American options using Monte Carlo simulation. This method will allow us to implement more complex option payoffs with greater flexibility, even if the payoffs are path-dependent. Specifically, we use the Least-Squares Method of Longstaff and Schwartz in order to take into account the early exercise feature. The stock price is assumed to follow the Geometrical Brownian Motion.

http://tech.harbourfronts.com/derivatives/valuing-american-options-using-monte-carlo-simulation-derivative-pricing-python/