How the rank is calculated.

Ratings, like chess

Each subject has a rating from the Glicko-2 system, a well-known method also used for chess and online games. Every ranked question is treated as an opponent with a rating of its own: its difficulty, measured from how real students of that age answer it.

Getting a hard question right lifts a rating more than getting an easy one right, and missing an easy question lowers it more than missing a hard one. Ratings start at 1500 and are brought up to date once a day.

Certainty counts

A new rating is a guess. Glicko-2 keeps a second number, the rating deviation, that says how sure the system is. It shrinks as a student plays and grows if they stop.

The rank uses the rating minus its deviation, a cautious estimate, so one lucky week can’t make a Legend. A subject is marked provisional while its deviation is above 150.

The Ado Rank score

A student’s score is the average of their best 3 subjects, each counted as rating minus deviation. The score sets the title, and the order of scores sets each student’s place in their age group.

Who is ranked, and when

A student is placed after 3 dailies and stays on the boards while they have played in the last 28 days. Places are worked out every night for the whole age group, and “#96 of 1,204” counts every student ranked that night.

Fair play

Each age group is ranked on its own: Sprout (6–8), Spark (9–11), Surge (12–14), Apex (15–17), and 18–22. Ranked questions are ones every syllabus covers, so a student’s syllabus never moves their rank. Every ranked question is timed, with more time for younger age groups.

Ranked questions are written with AI help, checked by a second model and reviewed by people before they reach ranked play. The 60-second sprint is the exception: its questions are computed by code, so every answer is exact, and they are not reviewed by a person.

Nothing bought changes a rank. Optional paid contests, when they open, never count towards it.

What it doesn’t measure

The rank is earned at home, on the student’s own device, not in an exam hall. That is why it is built from many days and uses a cautious estimate: one very good or very bad day moves it only a little.

It measures accuracy and speed on short questions across subjects, over time. It doesn’t measure creativity, long problem solving, or anything a timed quiz can’t see. Read it as evidence of steady work, next to school results and everything else a student does.

One rulebook, versioned

Every rating is calculated under a numbered version of these rules, and stored history replays exactly under the version it was played with. If the rules ever change, the version number changes too, so a past rank is never quietly rewritten.

Titles and the scores that earn them

Stars in each subject

Each subject also earns 1 to 5 stars from the same cautious estimate.