How Quanta marks your papers and finds your weak skills

Every paper starts as two documents: Cambridge’s question paper and its published mark scheme. Each question is split into its parts, and each part into the individual marks the scheme awards:

• Method, accuracy and independent marks (M, A and B in Cambridge’s notation), each with the marks it carries and what earns it, including whether a later mark can follow through from an earlier error. What those codes mean.
• The syllabus skills the part tests — one primary skill and any secondary ones, taken from Cambridge’s own syllabus content. That tag is what lets a lost mark count against the right skill.
• A checkable final answer for each maths part, and the official scheme wording kept beside it so you can compare your work after marking.

The totals are checked against the paper: criteria add up to each part, parts to each question, questions to the paper. Maths papers are mapped from the printed pages rather than extracted text, because PDF text extraction silently drops minus signs.

• Maths (0580, 9709). Final answers are checked automatically. Parts that cannot be checked by machine — “show that” proofs, sketches, constructions, answers in words — are marked by you against the exact criteria for that part, so those marks still count.
• Business (9609). Written answers are marked by an AI marker that is given that question’s published mark scheme — its indicative content and accept/reject guidance — and marks to Cambridge’s grid for the tariff: knowledge, application, analysis and evaluation for 3- to 20-mark questions, point-marking for 1- and 2-mark ones. You see the marks and the reasons straight away; a teacher can adjust them. How Business answers are marked.
• Economics and Chemistry multiple choice (9708, 9701). Answers are checked instantly against the key, and each one is explained once you commit to it.

Every marked part updates the skills it is tagged to. A few rules decide how:

• A wrong answer moves a skill further than a right one. One mistake is stronger evidence of a gap than one success is of mastery.
• Careless slips count for less. A miss on a skill you are strong on, mid-streak and answered fast, looks like a slip rather than a gap, so it moves the score less.
• Scores do not decay with time. A skill you have not practised for a while is flagged as stale instead, so you can confirm it rather than watch it fall.
• Four bands, fixed thresholds: weak below 40, developing 40–64, strong 65–84, mastered 85 and above.

The result is not “you scored 68%” but “you keep losing the reason mark on circle theorems” — a pattern no single paper shows.

Quark, Quanta’s study coach, reads that skill picture and points you at what to practise next — a short set of questions on one weak skill, or the next full paper. To see which topics carry the most marks across real papers, the most common topics pages count them.

• It does not publish Cambridge’s questions or mark schemes on public pages — those are Cambridge’s copyright and sit behind a free login. Public pages describe papers with facts: codes, marks, topics, grade boundaries.
• It does not predict what will be on your paper. It measures what has been.
• It does not replace your teacher. A teacher can join you, see the same skill picture and set you papers.