The purpose of this study is to evaluate the forecasting ability of GARCH-type models in estimating the Value-at-Risk (VaR) by introducing a new four-parameter distribution, called Exponentiated Odd Log-Logistic Normal distribution. The statistical properties of new heavy-tailed distribution are investigated and a simulation study is performed to assess the maximum likelihood estimations of introduced distribution. Then, the VaR is forecasted by using mean and volatility forecasts and quantile estimation of introduced distribution. Daily VaR forecasting ability of proposed two-stage model is compared with the GARCH models specified under heavy-tailed distributions by means of two backtesting methods. Empirical findings show that proposed two-stage model outperforms to well-known distributions such as normal, Student's-t, generalized error, and skewed generalized error distributions at high quantiles.