In this paper we have analyzed the performance of some Value at Risk models through the quadratic loss function backtesting approach. In order to highlight the differences among VaR models we have calculated the risk measure through Historical Simulation, EWMA, GARCH and EVT models. VaR was calculated on daily data of five Eastern and Central European main indices: BET (Romania), PX50 (Czech Republic), BUX (Hungary), SOFIX (Bulgary) and WIG20 (Poland) from 30.09. 2004 to 30.09. 2010. In order to highlight different behaviors in the crisis period we have divided the data into two samples and found that only advanced VaR models such as Extreme Value Theory or GARCH models can adequately measure the risk of the capital markets and satisfy the requirements of the investors in periods characterized by extreme events.