The main objective of this paper is to compare and evaluate the performance of the Value-at-Risk methodologies, paying particular attention to models that can capture skewness, kurtosis, fat-tailed characteristics or time-varying volatility. We examine the probability density function of Mixture Normal, Historical Simulation, BRW and Kernel Estimation of the non-parametric models, and the time-varying volatility models including EWMA, HW and conditional Extreme Value Theory (Conditional-GEV and GPD). The backtesting procedures include failure rate, binominal test, conditional coverage test and loss function. In general, models that can capture time-varying volatility perform better.