I want to test CVaR using the Berkowitz test (focus on the left tail). I have a couple of doubts:
Do I need to transform only actual losses that are above CVaR;
In the first transformation, whether the mean and variant parameters of the F (L) of the risk model are used or whether these two parameters are calculated on the basis of the variables being transformed;
in the LR, whether the mean and variance of the new normal variables are used or are data related to the F (L) generated by the risk model.