An analysis of a least squares regression method for American option pricing (Q1424693)

The authors analyze the least squares regression method computing the American option prices proposed by \textit{F. A. Longstaff} and \textit{E. S. Schwartz} [Rev. Financial Stud. 14, 113--148 (2001)]. Two types of approximation are involved in the proposed algorithm. Approximation one: replace the conditional expectations in the dynamic programming principle by projections on a finite set of functions. Approximation two: use Monte-Carlo simulations and least squares regression to compute the value function of approximation one. The almost sure convergence of the complete algorithm is proved under some general conditions. A type of the central limit theorem for the rate of convergence of the Monte-Carlo procedure is proved, thus providing the normalized error is asymptotically Gaussian.