See, I wouldn't say that a kid who was a level 7 in Y8 was 'on course' for a grade 9. Loads of kids are on a level 7 in Y8. You'd expect them to get a level 8 in Y9, then an A* in Y11. But that's the top 6-odd percent.
The problem here, I think, is that the school are using FFT computer generated targets for these kids, and that these are being given to the parents (which they shouldn't be, but that's another moan.). The teachers are then being asked to predict grades, which will appear alongside the target. Who wants to give a kid who has a target of a 9 and is working really hard a predicted grade that is less than their target? The teacher may also assume that the FFT know more about the kid than they do and think that if FFT think their kid is capable of a 9, then they should probably get one.
One problem is that FFT targets are usually aspirational, the grade that the top 25% of kids who have a similar profile to that kid will get. So most won't reach it. My school uses the FFT target for the top 10%. These are shared with parents, it really makes me cross.
Another problem is that FFT use average points score at KS2 to generate their targets. A kid who does well in English but badly in maths at KS2 will have a higher GCSE maths target than a kid who does badly in English and in maths, even though ability in maths and English aren't particularly linked. You'd think that they'd just use the KS2 maths score to set targets for GCSE maths grades but they don't.