Methodology · updated September 2026

How CrackTMUA works

We show you a lot of numbers: a predicted band, mock scores, question difficulty. You should know where they come from. This page explains how each one is worked out, how we tested it, and what we still can't know.

Predictions within 2 marks on real past papers
57% up from 35%
Real past-paper sittings it was tested on
84
Average miss on those sittings
1.9 marks out of 20

Your predicted band

Your predicted band is our estimate of the score you'd get if you sat the TMUA today. It comes with a range, because no estimate from practice can be exact, and the real exam has some luck in it too.

Predicted bandMedium confidence
6.2range 5.47.0
13579
The number is where you are today. About two times in three, the result lands inside the shaded range.

What counts, and how

Your first try at each question

Get a question wrong, read the solution, try again, and you'll probably get it right. That shows you remember the solution, not that you could do it cold. So only your first try counts.

Skips and unanswered questions

In a mock, a question you open and skip scores zero, as on the real test, and so do questions you didn't reach before time ran out. Leave a mock early and the ones you never opened are left out of your band, though your mock result still scores them zero.

How you're improving

The band follows your progress as you practise. Take a week off and it stays where it is, rather than drifting.

Both papers

Paper 1 and Paper 2 test different things. Until you've answered a few questions from each, your range stays wider.

Why it leans slightly low

We don't knock anything off to play safe. It sits a little low for two reasons. It shows where you are today, not on exam day, and most students keep improving right up to the test. And it won't jump high off easy questions alone: to show a high band, you need to get harder questions right.

How we tested it

We tested the band on 84 sittings of real past papers by CrackTMUA students. For each one, we took only the answers that student had given before it, predicted their mark, and compared it with the mark they actually got, counting only questions they had never seen. Each prediction came from a version of the band set up without that student's answers.

What we can't know yet

Whether the band matches real TMUA results. UAT-UK publishes how it scales scores, and we use that, but there's no official way to turn a practice mark into a TMUA score. The honest test is real results: after results day, we'll ask students to share their real score so we can check.

On real past papers

Mark predicted within 2 out of 20

Our previous method35%

Missed by 3.1 marks on average

How the band works now57%

Missed by 1.9 marks on average

84 sittings of official past papers by CrackTMUA students, measured September 2026. A small sample, so treat it as a guide. We update it as more come in.

How your mocks are scored

Your mock result comes from your mark and how hard those exact questions are, the way the real TMUA turns marks into a score. So a hard paper doesn't cost you.

A mock result is about one sitting. Your dashboard band uses your first try at every question you've answered, so the two won't always match. One tells you how that paper went, the other where you stand overall.

PaperMarkBand
2023 Paper 1 (easier)12/204.8
2021 Paper 1 (harder)12/205.7
Same 12/20. On the harder paper it's worth almost a band more.

How we rate difficulty

Every question has a difficulty from 1 to 9, the same range as TMUA scores. The higher the number, the fewer students get it right.

We rate questions blind: ours are mixed in with real past-paper questions whose difficulty is already known, and nobody rating them is told which is which. The real ones show how harshly the raters score: if a real 6.0 comes out as a 7.0, ours are corrected to match.

Then students' answers check the ratings. The rating you see doesn't change; the band makes the correction. Students find the official specimen easier than its ratings, most likely because many have seen it before, and our own Mock 2 plays almost a point harder than we rated it.

Q014?

A class of at least two students all sit the same tests. The class is called levelled if…

Q015?

The solutions to 7x⁴ − 6x² + 1 = 0 are ±cos θ and ±cos β. Which one of the following…

Q016?

f(x) is defined for all real x. Which one of the following is a sufficient condition for…

Q017?

A rectangular card ABCD has AB = √6 cm and BC = √3 cm. The card is folded along…

Ours and past-paper questions, rated without knowing which is which.

How we write and check questions

Every question we write follows the official TMUA specification and is checked against it. Since July, every new question goes through checks like these before it goes live. Older questions were re-rated blind in July, and every answer in Mocks 1, 4 and 5 was re-checked from scratch in September.

Our worked solutions explain why each step works, point out the trap most people fall into, and show why the other options are wrong. Most already do. We're rewriting the rest, starting with the specimen, then the older past papers.

Spot a mistake? The report button on every question comes straight to us, and we fix the question or take it down.

A real record

Mock 8 · Paper 2 · Question 11

  1. Written

    Written to the official TMUA specification.

  2. Answer checked

    Worked out again from scratch. It matched the key.

  3. Refereed

    The right answer could be spotted without solving it: it was the only option one small change away from all the others.

    One option replaced

  4. Rated blind

    Rated alongside real past-paper questions.

  5. Live

    Went live in Mock 8, with a report button for anyone who spots a problem.

Questions come back before you forget them

Every question you practise joins your review queue. It comes back when you're about 90% likely to still remember it: sooner if you got it wrong, and later each time you get it right. If you've set your exam date, nothing is scheduled after it. The timing uses FSRS, the open-source scheduler also built into Anki.

Your review queue

Due today
4
Tomorrow
2
In 3 days
5
Next week
9
Get one right and it comes back later. Get it wrong and it comes back sooner.

The same screen you'll see on the day

Mocks run in a replica of the Pearson VUE screen you'll use in the exam: the same timer, navigator and flagging, and no feedback until you finish. The first time you see that screen shouldn't be in the test centre.

Test of Mathematics for University Admission01:12:40
Flag
End Exam◀ Previous   Navigator   Next ▶

What we don't claim, and what we're still fixing

If we know about a weakness, you should too.

  • We aren't affiliated with UAT-UK or Pearson VUE.
  • Your predicted band is an estimate, not an official score.
  • Our mocks aren't past papers. We write them and rate them against real ones.
  • Students score about 1 mark higher on Paper 2 mocks than the band predicts. We're looking at why.
  • Our free score calculator uses a simpler table than the band, so the two can disagree by up to about 0.9 of a band. We're bringing them into line.

Things students ask

See it for yourself

Sit the official specimen free, under real exam conditions, and get a band for it.

Practise questions