THE POTENTIAL APPROACH IN PRACTICE

Abstract

This paper studies the fitting of Markov chain potential models to interest-rate derivative prices in four currencies simultaneously, using sequential Monte Carlo methodology (particle filtering). The potential approach starts from some Markov process which is supposed to drive the random observations, and there have been many studies where this Markov process is taken to be a diffusion; fewer studies have worked from a finite-state Markov chain, and this seems to be the first study to attempt to fit such models to data. With the available data, we show impressive agreement of the fitted models with market prices.