Skip to content

Our prediction record

Were we right?

Every score range we show is written down when we make it, then checked against the result a candidate reports. This page publishes what came back, including when it is bad.

The figure that matters

0

predictions checked against reported real results. We publish an exam figure only after 30 checks.

The promise

A range that has to face the result

We give a range, not a single flattering number. It is built so that 80 out of every 100 real results should land inside it. Whether they do is the measurement below.

Nothing has been checked yet. A prediction can only be checked once somebody has sat the real exam and told us what they got, so this page fills in after the first exam series and not before.

We are publishing it empty rather than waiting until it flatters us. When it has something in it, it will say what it says.

Limits stated plainly

What these numbers are not

Results are reported by candidates themselves. We are not an exam board and cannot read anyone's result, so these figures are only as good as what people told us — including the ones who did not tell us.

A prediction only counts here if we made it before the exam was sat. That is enforced in the database rather than trusted to us — a prediction recorded with the result already in hand cannot enter this page at all.

And a prediction is a description of the evidence, not a promise about the future. It answers “if you sat it today”. It will never tell you how much you are going to improve between now and the exam, because that is the most flattering thing we could say and the one thing nobody could check until it was too late to correct.

We publish the other numbers we would rather keep quiet too: every answer we got wrong.