So, a classical theoretical model tailored to exactly match known results is validated. And a brand-spankin-new theoretical model based on a different approach

So, a classical theoretical model tailored to exactly match known results is validated. And a brand-spankin-new theoretical model based on a different approach is invalidated. This means all assumptions of the classical model are correct and all those of the new model are incorrect? Doesn't sound right to me. These tests are validating / invalidating the predictive power of specific models, not testing their underlying assumptions. Issues of supervenience will not be worked out for certain until we have models of which we are more confident.