Valuing American Options by Weighted Least-Squares Quasi-Monte Carlo

American options pricing has the backward nature of iterative search feature. Based on the least-squares Monte Carlo(LSM), this paper employs Faure sequences and doubles the sample´s number by antithetic variate method to decrease the variance of simulation. Then, underling assets are valued. Thus, weighted least-squares quasi-Monte Carlo (WLSQM) is proposed by weighted least-squares regression. Comparing the two methods with option value, standard error and computation cost, WLSQM is better than LSM, which validates WLSQM is efficient on pricing American options.