Portfolio Value-at-Risk Approximation for Geometric Brownian Motion

Abstract

Value-at-risk (VaR) serves as a measure for assessing the risk associated with individual securities and portfolios. When calculating VaR for portfolios, the dimension of the covariance matrix increases as more securities are included. In this study, we present a solution to address the issue of dimensionality by directly computing the VaR of a portfolio using a single security, therefore requiring only one variance and one mean. Our results demonstrate that, under the assumption of Gaussian distribution, the deviation between the computed VaR and actual values is relatively small.