Multilevel nested simulation for efficient risk estimation

TL;DR

This paper introduces an adaptive multilevel Monte Carlo method for efficiently estimating nested expectations, such as risk measures in finance, achieving near-optimal computational complexity.

Contribution

It develops and analyzes an adaptive MLMC algorithm for nested expectations, providing theoretical complexity bounds and practical algorithms for risk measures like VaR and CVaR.

Findings

Achieves $ ext{O}( ext{ε}^{-2}| ext{log} ext{ε}|^2)$ complexity for nested expectation estimation.

Provides a stochastic root-finding algorithm for VaR and CVaR with $ ext{O}( ext{ε}^{-2})$ complexity.

Validates the theoretical results with numerical experiments on a model problem.

Abstract

We investigate the problem of computing a nested expectation of the form $\mathbb{P}[\mathbb{E}[X|Y] \!\geq\!0]\!=\!\mathbb{E}[\textrm{H}(\mathbb{E}[X|Y])]$ where $\textrm{H}$ is the Heaviside function. This nested expectation appears, for example, when estimating the probability of a large loss from a financial portfolio. We present a method that combines the idea of using Multilevel Monte Carlo (MLMC) for nested expectations with the idea of adaptively selecting the number of samples in the approximation of the inner expectation, as proposed by (Broadie et al., 2011). We propose and analyse an algorithm that adaptively selects the number of inner samples on each MLMC level and prove that the resulting MLMC method with adaptive sampling has an $\mathcal{O}\left( \varepsilon^{-2}|\log\varepsilon|^2 \right)$ complexity to achieve a root mean-squared error $\varepsilon$. The theoretical analysis is verified by numerical experiments on a simple model problem. We also present a stochastic root-finding algorithm that, combined with our adaptive methods, can be used to compute other risk measures such as Value-at-Risk (VaR) and Conditional Value-at-Risk (CVaR), with the latter being achieved with $\mathcal{O}\left(\varepsilon^{-2}\right)$ complexity.