How this data is compiled
Everything on this site is derived from documents the awarding bodies publish themselves. No boundary is estimated, interpolated or carried over from another year.
Sources
- AQA Education —AQA's published boundaries
- Pearson Education Ltd —Edexcel's published boundaries
- Oxford Cambridge and RSA Examinations —OCR's published boundaries
How the parsing works
Each board publishes the same facts in a different shape, so each gets its own adapter.
- AQA publishes spreadsheets at content-hashed URLs that change whenever they republish. Rather than hard-code those links, we read the archive index and then take the exam series and qualification from inside each workbook, so the ingest survives AQA reorganising their site.
- Pearson Edexcel publishes PDFs. Their CDN answers every URL with HTTP 200 — serving an HTML error page for names that do not exist — so each download is verified against the PDF magic bytes and then against the qualification and series printed on its own cover page. A file that is not what we asked for is discarded rather than filed under the wrong year.
- OCR publishes PDFs going back to 2017, with component and qualification rows interleaved in one table. Tables are reconstructed from the text coordinates rather than a flattened text dump, because the column position is what distinguishes "no grade 9 available on Foundation tier" from "this row is shifted".
What we drop
An ingest that guesses is worse than one that omits. Records are discarded when the subject cannot be identified, when no specification code is present, or when a row carries no published boundaries at all — which boards emit for papers nobody entered in a given series.
Double-award combined science, which is graded on paired grades such as 9-8 rather than a single ladder, is not yet included: the calculators assume a single grade per award and would give misleading answers.
Which awards have a paper-by-paper breakdown
About nine in ten awards do. Where one is missing, the page says so and offers only the total-mark converter.
The gap is concentrated in one genuine ambiguity rather than a parsing failure. OCR describes some A-level awards as a range of permitted units — for exampleY100 + (Y101 to Y113) + (Y201 to Y224, except Y213, Y219) — which names the units you were allowed to choose from, not the ones you sat. There is no determinate paper set to calculate against, so we do not invent one.
Where the calculators are approximate, and where they are not
Converting a total subject mark to a grade is exact — it is a table lookup against the published boundary.
Working back from individual paper marks is exact when the papers' maximum marks sum to the subject maximum, which covers most subjects. It is an approximation in two cases, and the page says so both times: where a board scales a component before adding it to the total, and where a subject offers a choice of options and the board does not publish which option code belongs to which award. In the second case we ask you which papers you sat rather than guessing.
Current coverage
| Measure | Count |
|---|---|
| Awards | 4,039 |
| Component records | 21,712 |
| Exam series | 17 |
| Subjects | 57 |
| Awarding bodies | 3 |
| Years covered | 2017–2026 |
Corrections
If a figure here disagrees with the board's own document, the board is right and we have a bug. The citation under every table links to the source document so you can check.