The comparison of Value-at-Risk models performance

Abstract

This study to use the concept of Value at Risk (VaR) to compare the results to show which methods of modeling can be effective in describing the total loss of investment in SET index, SET50 index and mai index during January 2016 to December 2022. The three approaches which are Historical Simulation, Delta Normal and Monte Carlo Simulation are used to calculate VaR. The estimation effectiveness is measured by Likelihood Ratio (Kupiec, 1995).

The results are that Delta Normal and Monte Carlo Simulation showed no difference in efficiency at 95 and 99 percent confidence intervals. At 95 percent confidence intervals, Delta Normal and Monte Carlo Simulation are applicable in measuring VaR for SET index. Historical Simulation is the most applicable approach in measuring Value at Risk as compared to the other two methods at 99 percent confidence intervals. However, Historical Simulation may overestimate VaR and cannot capture the actual loss.