Why BKT — isn't this just percent correct?

Percent correct doesn't tell you whether the pupil knows it because they understand it, or because they got lucky. BKT estimates the underlying probability that the pupil actually knows the skill, adjusted for slip (bad luck) and guess (lucky guess).

How does BKT relate to the math bank's theta?

Theta (1-PL Rasch) says how hard an item is and where the pupil sits on the same scale. BKT uses that signal to update kendskabsgrad per skill — giving us both 'what's the pupil's level' and 'how confident are we'.

Mastery engine

Per-skill mastery probability — BKT + IRT in one

Star ratings don't tell you how sure a pupil is. The mastery engine layers Bayesian Knowledge Tracing on top of the math bank's 1-PL Rasch theta and says it plainly: 'You're 78% confident on fractions'. The same picker works across maths, grammar, languages and FNFT.

Contact

BKT per skill

Each skill carries its own known-level, updated after every attempt with the classic BKT parameters (p_init, p_learn, p_slip, p_guess). No black box.

Wilson confidence

We also show how confident we are in the confidence. One correct attempt doesn't 'master' a skill — the Wilson lower bound prevents false positives.

Shared picker across subjects

The engine is subject-agnostic: maths, grammar, the language tutor and FNFT all call the same system. Pupils always get the skill with the lowest known-level that matches their year.

Mastery map, not stars

Per-domain progress bars with bands Mastered (≥85%), On track (≥40%), Struggling and Untested — visible to pupil, parent and teacher.

Built on top of the math bank

The 1-PL Rasch theta is still the signal — BKT layers interpretation on top. The adaptive picker for maths doesn't regress.

Same map for pupil, parent and teacher

Visible on /student/mastery, /parent/children/:id/mastery and the teacher's pupil page. Everyone sees the same percentage — no interpretation drift between home and school.

In the classroom

The pupil who wants to know 'where am I'

Opens the mastery map and sees '78% confident on fractions, 41% on decimals'. Clicks straight into the next item in the lowest-weight skill.

The parent who wanted stars explained

Gets a concrete number and a concrete skill — instead of '3 out of 5 stars'. Easier to talk about at home, easier to see progress over time.

The teacher differentiating per skill

Sees which skills the whole class is struggling with and which are ready for the next level. Differentiation becomes data-driven — not guesswork.

Works with

Adaptive math bank Adaptive assessment Personalised learning Always-on AI

Frequently asked questions

Ready to try it?

Get a free demo for your school — we'll show you the whole platform in 30 minutes.