Every number here came from a game someone played.
What is behind the figures on every other page: where the boards come from, what the floors are, and what Riot’s match data does not contain.
Rebuilt on every refresh. Updated 13 Sep 2026, 08:40 UTC.
How the numbers are made
TFT Codex samples ranked matches through Riot’s API and stores every final board — units, items, star levels and traits, whole. Compositions are clustered out of the boards nobody else’s list explains, and named from the traits and carries they actually run.
Unit, item and trait numbers pool across every board we sample, so they carry far larger samples than any single composition does. Everything is measured over a rolling 14-day window and rebuilt every two hours.
Every board kept
Units, items, star levels and traits, stored whole — so a question we haven’t thought of yet can still be answered later.
Comps nobody wrote up
About a third of boards match no published composition. Those are clustered into comps of their own, named from the traits and carries they run.
A tier has to be earned
Below the sample threshold a comp shows a dash, not a grade. We would rather say nothing than say something thin.
The floors
A comp shows an average at 50 boards and a tier at 400. A unit, item or trait row needs 400 boards, and the Master+ split on those pages needs 400 of its own — below that a placement average cannot separate one row from the next. Rows under a floor are kept and counted; they are shown without a number.
Placement has a standard deviation of about 2.29, so a 163-board average carries a 95% interval of roughly ±0.35 — wider than the gap between two tiers. That is the whole reason the floors exist.
What is not measured
Riot’s match data reports final boards only. It does not say which augments a player took, where units stood, or what a board looked like before the last round. So there are no augment win rates here, no positioning and no stage-by-stage boards, and the augment section on a comp page pairs what an augment grants with something measured about the comp instead.
The bracket shown on a comp is the one its match was discovered through, not each player’s own rank. Master+ is the finest split those labels can honestly carry.
Who is named
Riot IDs are public, shown in every lobby and on Riot’s own leaderboards, and we resolve them only for Challenger, Grandmaster and Master. Everyone else in our data is a one-way pseudonym we could not name if we wanted to. The privacy policy says exactly what is stored.
The overlay
The same comps and numbers, drawn over the game on Android. It names the comp your board is closest to and shows what the sample says about each augment offered. No account, and nothing uploaded from the phone.