Credit VAR (value at risk) models measure credit risk exposures by estimating a metric called “unexpected credit losses,” defined as the difference between the mean and an extreme loss percentile of a credit portfolio’s profit-and-loss distribution for a selected horizon.