Quantization based recursive Importance Sampling

TL;DR

This paper introduces a deterministic importance sampling method using quantization and Newton-Raphson, offering a robust alternative to traditional stochastic recursive importance sampling, especially for complex Monte Carlo simulations.

Contribution

It presents a novel quantization-based importance sampling algorithm that does not rely on simulations and is applicable to multi-dimensional and infinite-dimensional problems.

Findings

The method is consistent for multi-dimensional distributions and diffusion processes.

The error in the change of measure is controlled by quantization error.

Effective in pricing options in complex, high-dimensional settings.

Abstract

We investigate in this paper an alternative method to simulation based recursive importance sampling procedure to estimate the optimal change of measure for Monte Carlo simulations. We propose an algorithm which combines (vector and functional) optimal quantization with Newton-Raphson zero search procedure. Our approach can be seen as a robust and automatic deterministic counterpart of recursive importance sampling by means of stochastic approximation algorithm which, in practice, may require tuning and a good knowledge of the payoff function in practice. Moreover, unlike recursive importance sampling procedures, the proposed methodology does not rely on simulations so it is quite generic and can come along on the top of Monte Carlo simulations. We first emphasize on the consistency of quantization for designing an importance sampling algorithm for both multi-dimensional distributions and diffusion processes. We show that the induced error on the optimal change of measure is controlled by the mean quantization error. We illustrate the effectiveness of our algorithm by pricing several options in a multi-dimensional and infinite dimensional framework.