> There’s literally zero reason to invest any time into models using it whatsoever

This is dumb. There are plenty of cases where predicting the rational outcome and measuring an empirical gap from it reveals opportunity.