"Error Estimates for the Longstaff-Schwartz Least-Squares Monte Carlo Algorithm" by Daniel Zanger (UCSB)

Event Date: 

Monday, November 30, 2020 - 3:30pm to 4:30pm

We discuss error estimates for the well-known Longstaff-Schwartz algorithm, a least-squares Monte Carlo regression algorithm for American/Bermudan option pricing. The overall goal is to establish the most general such estimates, presupposing the fewest hypotheses possible on the hypothesis sets (approximation architectures) used in the regression as well as the underlying distributions. Both the sample and approximation errors are addressed.