An optimization process in Value-at-Risk estimation

A new method is proposed to estimate Value-at-Risk (VaR) by Monte Carlo simulation with optimal back-testing results. The Monte Carlo simulation is adjusted through an iterative process to accommodate recent shocks, thereby taking into account the latest market conditions. Empirical validation covering the current financial crisis shows that VaR estimation via the optimization process is relatively reliable and consistent, and generally outperforms the VaR generated by a simple Monte Carlo simulation. This is particularly true in cases when the out-of-sample evaluation sample spans a lengthy period, as the traditional method tends to underestimate the number of extreme shocks.