> The lowest performers are unlikely to become best, but the middle still changes significantly.

As long as the middle gets sorted into the Realschule, there's no problem ;-) But more seriously, according to wikipedia[0] IQ is relatively stable beginning at an age of approx. 11 years, so if we compensate for the measurement error with a permeable system I only see two problems: As you mentioned, at this age the brain devolops so rapidly that being the oldest in the class gives students a measurable advantage, but that is an issue for any school system, and as I said in a neighboring comment, overly ambitions parents are meddling too much, but that is a relatively recent phenomenon and a problem unto itself.

[0] https://en.wikipedia.org/wiki/Intelligence_quotient#Age

Be very careful extrapolating from Wikipedia on IQ. The science is subtle and not at all as stable as the encyclopedia makes it sound. One of the reasons people in the field talk about this stuff so gingerly is the proclivity laypeople have for taking half-baked science and turning it into public policy.

I know the difficulties surrounding IQ, I was just quoting it to show that predictions about future ability can be made at that age.

... and my point was, probably no they cannot.

I quoted a source for my claim, imperfect though it may be. Can you support your assertion?

Your source was Wikipedia.

More to the point: you can't in one breath say "I know the difficulties surrounding IQ" and in the next say "any source I cite about IQ must represent settled science". One or the other of those statements must not be true.

Would you kindly elaborate on what you perceive to be the subtleties about IQ? Besides the danger of mistakenly thinking an important factor is the only factor, which is of course a common problem. About predictive qualities in childhood, take a look at [0], maybe this study is more to your liking. But anyway, this discussion is about the predictability of academic success in young children, so please provide a source for your claim that such predictions are infeasible.

[0] Deary, I. J., Whalley, L. J., Lemmon, H., Crawford, J. R., & Starr, J. M. (2000). The stability of individual differences in mental ability from childhood to old age: Follow-up of the 1932 Scottish mental survey. Intelligence, 28, 49 -55

German system does not measure IQ. They split children based on second, third and partly fourth grade grades.

There is a lot more then IQ that goes into grades. All of it happens before 10.

I think I explained myself poorly. Of course the schools don't measure raw IQ, there are many factors like intrinsic motivation (problematic as a measure in young children) conscientousness (which is a relatively stable character trait) and of course parental pressure which distorts any measurement we can try to make. All I was trying to say that it is possible to judge the potential of the children with adequate accuracy.

Your claim the IQ does not change after 11 years old, even if true, is completely irrelevant then. It shows neither that it is good idea to split kids as 11 years old nor that you can say potential so soon.