The evidence surely has to have a main focus on mocks as the closest thing to the real assessment, guaranteed done in exam conditions.
What I’m imagining is that Johnny is predicted a 3 based on mocks, then I have to say ‘but he was sick in paper 1, here’s the evidence that he’s one of the better students in the class’ with attached test data, and he gets bumped to a 4. They won’t then go looking at the test data and say ‘but Jemima did badly in the tests even though she did well in the mocks so we will knock her to a 3’.
The mock results would be the prime evidence, with other evidence only coming in if a mock result needed amending. If you see what I mean.
Maybe that’s a maths thing. Our mock data is usually pretty sound.