This study analyzes the application of Value at Risk (VaR) in estimating the risk of investment in banking stocks and the formation of an optimal portfolio using the Mean-VaR method based on the Markowitz approach. Many studies show that market data are often abnormal and make the assumption of normality considered irrelevant. This is the background of research on VaR using the historical simulation method, which is a method that moves away from the concept of normality. In addition, the crisis due to the Covid-19 pandemic makes the market difficult to predict. The period used in this study is during a normal market and a crisis (covid-19 pandemic). VaR is calculated with a holding period (t) of one week and a confidence level of 95%. Based on the backtesting test, the historical simulation method is accepted as an accurate method in estimating the VaR value in both normal and crisis periods. The optimal portfolios formed based on the mean-VaR are Portfolio-1 (normal period) and Portfolio-2 (crisis period). The composition of Portfolio-1 is BBRI, BBCA, BNLI, BTPN, and BNBA with the optimal proportion of each share sequentially of (18.35%), (23.90%), (11.39%), (18.63 %), and (27.73%). The VaR value of Portfolio-1 is -0.0107. The composition of Portfolio-2 is BNII and BNBA with optimal proportions of each share (22.71%) and (77.29%). The VaR value of Portfolio-2 is -0.0354. The results of this study can be used by investors as a reference in making investment decisions that focus on downside risk.