A NEW SEMIPARAMETRIC MIRRORED HISTORICAL SIMULATION VALUE-AT-RISK MODEL

In this paper, the authors have developed and presented a new semiparametric value-at-risk (VaR) model for the assessment of market risk. The model is based on the theoretical foundation of the Historical Simulation (HS) method. The basic intention was to develop a new model that would be easy to implement and able to envelop the empirical features of returns, such as leptokurtosis, asymmetry, autocorrelation, and heteroscedasticity, and also to improve risk estimation in the tail distribution for the sample size and the confidence level prescribed by the Basel III standard. To obtain the answers to the question of whether the new model is an improvement against the popular improvements of the HS method, its performances were tested in terms of adherence to the backtesting rules of the Basel Accord and also compared with the backtesting results of the popular improvements of the HS method. The backtesting results justify the expectations of the new model.