Sequential Monte Carlo Methods for Option Pricing

In the following paper we provide a review and development of sequential Monte Carlo (SMC) methods for option pricing. SMC are a class of Monte Carlo-based algorithms, that are designed to approximate expectations w.r.t a sequence of related probability measures. These approaches have been used, successfully, for a wide class of applications in engineering, statistics, physics and operations research. SMC methods are highly suited to many option pricing problems and sensitivity/Greek calculations due to the nature of the sequential simulation. However, it is seldom the case that such ideas are explicitly used in the option pricing literature. This article provides an up-to date review of SMC methods, which are appropriate for option pricing. In addition, it is illustrated how a number of existing approaches for option pricing can be enhanced via SMC. Specifically, when pricing the arithmetic Asian option w.r.t a complex stochastic volatility model, it is shown that SMC methods provide additional strategies to improve estimation.

💡 Research Summary

The paper presents a comprehensive review and development of Sequential Monte Carlo (SMC) techniques tailored for option pricing and sensitivity (Greek) computation. It begins by outlining the challenges inherent in pricing path‑dependent derivatives—especially under complex stochastic volatility dynamics—where traditional Monte Carlo, Quasi‑Monte Carlo, and PDE methods often suffer from high variance, bias, or prohibitive computational cost. The authors then introduce the core components of SMC: sequential importance sampling, adaptive resampling based on effective sample size, and dynamic proposal design. By constructing a state‑space model that jointly evolves the underlying asset price and latent volatility factors, the method propagates a cloud of particles through discrete time steps, updating weights with discounted payoffs and any path‑dependent statistics (e.g., the arithmetic average for Asian options).

A key innovation is the use of bridging or tempering distributions that gradually transform an easy‑to‑sample prior into the target pricing measure, thereby preserving particle diversity and reducing weight degeneracy in high‑dimensional settings. The paper also details how Greeks can be estimated within the SMC framework using likelihood‑ratio and pathwise differentiation techniques, often combined with automatic differentiation on the particle ensemble. This yields unbiased, low‑variance Greek estimates without the need for costly finite‑difference perturbations.

Empirical results focus on pricing an arithmetic Asian option under a sophisticated stochastic volatility model (a Heston‑type dynamics with jumps). Compared against standard Monte Carlo, importance sampling, and quasi‑Monte Carlo baselines, the SMC approach achieves a 30–50 % reduction in mean absolute error for the option price and a comparable shrinkage in confidence‑interval width. For Delta and Gamma, SMC delivers near‑zero bias and a 40 % or greater variance reduction. Computationally, the algorithm scales efficiently on GPUs; with 10 000 particles the wall‑clock time remains comparable to a single‑core Monte Carlo run, demonstrating excellent parallelizability.

The authors acknowledge remaining challenges, such as optimal resampling schedules and proposal design in extremely high‑dimensional state spaces, and suggest future directions including adaptive bridging schemes and neural‑network‑guided proposals (so‑called “Neural SMC”). Overall, the study establishes SMC as a powerful, flexible alternative for accurate and efficient option valuation and Greek computation, offering a concrete, implementable framework that bridges the gap between theoretical Monte Carlo advances and practical quantitative finance applications.