Hmmm, this is semantics. The order of grades within a school will come from the school and should have a high confidence level, but if a school has an atypical cohort, the grades determined by the statistical modelling may or may not be described as estimates.
Big cohorts and selective schools where there is less variation will be OK, and I think GCSEs should be OK as they are likely to be within 1 grade of teacher estimates, but we just don't know how non-selective schools with outliers will fare.
Estimates are generally the result of a calculation, the prevailing view in the press is that around a third of CAGs will be changed. Of course it is also true that if people did exams a significant number would have had a bad exam and would not have done as well as expected.