The problem of evaluating models built for forecasting is that one cannot get around invariance. There must be a stable bridge that connects the past and pres. ent with the future … To evaluate invariance its domain should be considered. In other words, one has to investigate what the list of all relevant non-negligible potential […]