This paper proposes a method of estimating Value-at-Risk by combining asymmetric multivariate GARCH models and filtered historical simulation (Barone-Adesi et al., 1999). Next, incremental VaR is implemented in order to decompose the portfolio and assess the risk of every individual component. Ten competitive models were estimated and subsequently back tested using five techniques. All methodologies were applied on a sample of 11 financial assets from Bucharest Stock Exchange between 2014-07-08 and 2019-10-04. The results indicate that the method using filtered historical simulation in combination with multivariate GARCH models that account for asymmetry of financial returns lead to good VaR estimates. The methods discussed in this paper could help an investor to create a better risk-optimized portfolio, but could also be used by a regulatory authority in order to impose restrictions regarding risk.