The traditional portfolio value-at-risk (PVaR) estimation method commonly used in current practice exhibits considerable biases due to model specification errors. This paper attempts to improve PVaR estimation by relaxing the conventional assumption of no
