Adaptive Monte Carlo variance reduction for Lévy processes with two-time-scale stochastic approximation (Q931375)

An approach to a two-fold optimal parameter search for a combined variance reduction technique of the control variates and importance sampling in a pure-jump Lévy process framework is proposed. More precisely, Section~2 recalls in brief the principle of the control variates and the importance sampling. Section~3 formulates combined control variates and importance sampling and discusses the optimality with the control variates component held fixed. Moreover, the almost sure convergence of a two-time-scale algorithm to a unique root of the gradients of the variance is proved. In Section~4 the parameter search using the two-time-scale algorithm is embedded in an adaptive Monte Carlo procedure in the case of the gamma distribution, and the strong law of large numbers for an empirical mean and the central limit theorems for the empirical variance are proven to hold. Section~5 presents and discusses results of numerical experiments of the DCO tranche pricing with the gamma copula model and the intensity gamma model. Finally, Section~6 concludes this study with some remarks.