Past papers as diagnosis, not practice

Do a past paper early, under something close to real conditions, before you feel ready. It feels awful and it is the most useful thing in the fortnight, because it tells you what you cannot do while there is still time to fix it. Doing them at the end tells you the same thing too late to act on.

Mark it against the real scheme where one exists. The gap between what you wrote and what earns marks is usually not knowledge at all — it is form: not answering the question asked, not showing method, not managing the clock. Those are fixable in days.