Iterative construction of the optimal Bermudan stopping time (Q2488505)

The central result of the paper is an iterative construction of the Bermudan Snell envelope via a sequence of stopping times which increases to the first optimal stopping time. At each iterative step a whole family of stopping times \((\tau_i\)) is introduced where \(i\) runs through the set of exercise dates and \(\tau_i\) is the stopping time for the Bermudan which is not exercised before date \(i\). The main point is that at each iteration step an improved approximation is obtained for the Snell envelope, which ranges over all exercise dates. The presented iterative method may be considered as a pseudo-algorithm since for an implementation conditional expectations need to be computed along trajectories. In a Markovian setting these conditional expectations may be computed (approximately) by Monte-Carlo and thus the resulting algorithm is spelled out. The method may be considered generic for all discrete optimal stopping problems. The power of the procedure is demonstrated for Bermudan swaptions in a full factor LIBOR market model.