Forecasting is difficult, partly because all models are wrong (or because we are wrong about models). And while we usually don’t know—with certainty anyway—what’s wrong, we often can know what’s not quite right. Specifically, we may not know the best model that can generate the most accurate forecast—not until we tried them all, so never, really. But we can know which of the models generates an