There are plenty of data available on each students. The issue is how to normalise them in order to create a national ranking.

Mock exams may be useful because, arguably, the results should already be pretty normalised, but I'm not sure if these took place this year.

So what is left is to work with is the grades the students have had throughout the year and previous years and, importantly, also what school they are in.

The latter is heavily charged politically so they are wary of communicating on how that impacts their model.

It seems to me, though, that pupils from top private schools are not complaining too much about having their grades downgraded, so it seems that it did not happen too much for them.