9709 · How we rank

How we rank Mathematics

Guess papers claim to know which topics will be on your exam. That's not possible for Cambridge International: the board designs its papers to test almost the entire syllabus every series. Across the subjects we've measured, between 86% and 95% of syllabus content is tested each sitting (89% for 9709 Mathematics). So instead of guessing which topics appear, we ran the test guess papers never do — see below.

What genuinely varies from series to series is how many marks each topic is worth. That's what we rank: mark share measured from verified past papers, combined with how ready you are on each topic — so you revise where the marks are, not where you'd guess a topic might appear.

The experiment behind this

Before we ranked by mark weight, we tested topic prediction directly: could a model beat a naive "topics that appear most often" baseline, scored against exam series it never saw? Here's the record for Mathematics, kept as evidence — not a live accuracy claim about what we publish today.

Coverage — model weight-0.1

Enough tagged series to publish

Retired experiment: topic prediction

Before we ranked by mark weight, we tested whether topic prediction could beat a naive baseline, measured only from official Cambridge International past papers. For this subject it did beat that baseline — but topic prediction failed across the subjects we tested overall, and mark weight plus your own mastery is the more useful signal for revision priority either way, so that is what we rank by. These numbers are that experiment's result, not a live accuracy claim about what we publish today.

Hit rate: 39%
Naive baseline: 14%
Lift: +25%

Evaluated on 3 held-out series the model never saw during fitting, generated Aug 2026.

Limitations

Mark share is measured from a limited number of past series per subject — it gets more reliable as more series are added, not less.
A topic's mark share is typically what it has been worth, not a guarantee for the next sitting — any single topic can be worth more or fewer marks than its measured average.
Ranking by weight is one input; your own mastery matters at least as much, which is why the study plan combines both.