Value at Risk is one measure used to measure the maximum potential loss that will be experienced in the period of a day, five days and twenty days. In this study, the methodology used is Variance Covariance models and Historical models. Object of research include 6 shares for 3 consecutive years (2012 to 2014) was recorded on LQ 45. Potential losses are measured on a confident level of 95%. And this validity model was tested by conducting back testing with Kupiec Test, where the potential maximum loss calculation results compared with a loss actually occurred. Back test measurement results indicate that the potential loss of Variance Covariance model is smaller than the Historical models, but both models are declared valid in measuring the potential maximum loss of LQ 45.