> What is true is teachers overrate girl's performance earlier in school compared to boys, then girls catch up and overtake, often approaching their teacher's predictions. This might well be a case of matching expectations. Another, to me reasonable, explanation for the results is that teacher's grading includes potential, based on their experience. It's hard to know, and would be very hard to ethically test.

Girls overtake boys in terms of grades assigned by teachers, but they don't in terms of scores on standardized tests. You can make the argument that the grades are just another method of overrating girls' performance.