Every summer the same beliefs circulate, and every summer they are wrong.
“Only a fixed percentage can get a 9.” “They lower the boundaries if everyone does badly.” “It is marked on a curve, so you are competing with everyone else who sat it.” “The boundaries were low this year so the grade is worth less.”
None of those describe what happens. The real process is more interesting, and knowing it changes how you should think about a mock result, a raw mark, and the gap between the two.
The order of events is the key to all of it
This is the fact that makes the rest make sense.
Papers are marked first. Boundaries are set afterwards.
Your script is marked against the mark scheme by an examiner who has no idea what the boundaries will be, because nobody does yet. They award marks. That is the whole job.
Only once a large volume of scripts has been marked do the boards hold awarding meetings, where the boundaries for that particular paper, in that particular year, are decided.
So there is no curve applied to your script. There is no moment where your paper is compared to the one next to it. Your raw mark is your raw mark, and then a line is drawn.
Why the line moves
The boundaries move because papers are not equally hard from year to year, and the whole point of moving them is to stop that mattering.
Exam boards try to make each year’s paper as similar in difficulty as the last, and they cannot fully succeed. A paper with an awkward question in it produces lower raw marks from a cohort that is exactly as capable as last year’s. If the boundary stayed fixed, that cohort would get worse grades for no reason other than the accident of which questions were written.
So the boundary moves to absorb the difference. A harder paper gets lower boundaries. That is not generosity and it is not grade inflation. It is the correction that keeps a grade 7 in 2026 meaning what a grade 7 meant in 2025.
Which disposes of the “low boundaries mean the grade is worth less” idea entirely. Low boundaries mean the paper was hard. The grade means the same thing either way. That is the point of moving them.
What happens in an awarding meeting
Two kinds of evidence go on the table.
Statistical evidence. The board has predictions for how this cohort should perform, based on what this same group of students achieved earlier. For GCSE that anchor is their Key Stage 2 results; for A-level it is their GCSEs. This is the part usually called comparable outcomes, and it is why a cohort that arrived with stronger prior attainment is expected to leave with stronger results.
Examiner judgement. Senior examiners look at real scripts at the marks around each proposed boundary and ask whether the work in front of them is genuinely the standard of that grade. They compare them with scripts from previous years held as reference points, called archive scripts.
Where the two disagree, there is a discussion, and the judgement matters most at the key boundaries. In England, Ofqual sets the rules the boards work within, and since 2023 the standard has been the pre-pandemic one, held steady since.
The output is a set of boundaries, published on results day, usually on the board’s website within a couple of hours.
The things students get wrong about this
“Only the top 5% get a grade 9.” No. There is no quota, and nothing stops a cohort from producing more 9s than the last one if it is genuinely stronger. What is true is that grade 9 is set by a formula rather than by examiner judgement in the same way as the other boundaries, so it moves somewhat mechanically once grade 7 is fixed. But it is not a cap.
“I am competing against everyone else in the country.” Not in any direct sense. Your script is marked on its own. The cohort influences where the line goes, slightly, through the statistical evidence, but you cannot lose a grade because the person next to you did well.
“Boundaries are set before the exam.” They cannot be. The paper has to be sat and marked before anyone knows how hard it turned out to be.
“My mock boundaries are the real ones.” This is the one that causes actual harm, and it deserves its own section.
Why your mock grade is less solid than it looks
When your school marks a mock, it has to apply some boundary to turn your raw mark into a grade, and it will normally use the published boundaries from whichever year that paper is from.
Those boundaries belong to a different paper, sat by a different cohort, in a different year. They are the best available estimate and they are still an estimate. A mock grade is a raw mark that has had a borrowed line drawn through it.
Schools also, quite deliberately, sometimes set mock boundaries higher than the real ones. There are decent arguments for it. It means your mock grade is not a prediction, it is a signal.
The practical consequence: pay attention to your raw mark and where the marks were lost, not the letter or number on the front. The raw mark is real data about your performance. The grade attached to it is an inference built on a borrowed assumption, and it is the part most likely to change.
What this means for how you revise
A few things follow directly.
- Track marks, not grades, across the year. A move from 54 to 63 raw marks is unambiguous progress. A mock grade that stays at a 6 across three papers may hide exactly that improvement, because the boundary is a step function and you can gain nine marks without crossing one.
- The boundary you are near matters. If you are three marks below a boundary, three marks is the entire task. Find out which questions are closest to being recoverable and go there first. This is the single highest return revision decision available to most students and it requires knowing your paper question by question.
- Do not revise for the boundary being kind. You cannot know it in advance, nobody can, and building a plan around a hard paper this year is building on nothing.
- A near miss on results day has a specific route. If you are one or two marks off, that is exactly the situation where a review of marking is worth considering, through your school. We wrote about how reviews of marking actually work, including when it is not worth the fee.
The short version
Your paper is marked against a scheme, on its own, by someone who does not know the boundaries. The boundaries are set afterwards, by people looking at real scripts and at how this cohort was expected to perform, and they move to cancel out how hard the paper turned out to be.
Which means the only variable you control is the raw mark, and the only useful question after any practice paper is where the raw marks went.
A note for departments
A department that keeps its own boundary data across several years, per paper, calibrates mocks better and has a much easier conversation with students about what a mock grade is and is not. For the run-up to the day itself, see the GCSE and A-level results-day guides.
Marky holds boundaries as department data, set per subject, paper, tier and exam series rather than assumed, and records every paper at question level, so the raw marks stay visible underneath whichever boundary was applied. See how it works.