If the NCAA bracket model is an indication, Nate Silver does not intend to disclose the actual model used to make his predictions, just like his election models

If the NCAA bracket model is an indication, Nate Silver does not intend to disclose the actual model used to make his predictions, just like his election models. He tells you what elements he's mixing but not how. I expect he will also continue to fail to quantify the performance of his models against other benchmarks. In the case of the NCAA how much value does his more complex model add compared to each of the rankings he uses as inputs alone?