Percentile cards · similarity
The Scouting Terminal
Type a name, read the whole player. Every figure is a percentile computed inside the player's position group across the Big-5 leagues — a full category card, a shape radar, the ten most statistically similar players, and a style map you can click through. Lamine Yamal, at 18, already rates 96.
Game model
The player read the way a coaching staff talks — one role per phase, with the evidence beside it. Opta event corpus, Big-5 2025/26 outfielders only. Thirds stand in for phases — the feed carries no possession-phase segmentation. Loss-window figures (counterpress, third man, one-twos) are minute-window proxies: the feed carries no seconds, so “within moments” means within the same or next recorded minute. Percentiles are within position cohort. Everything here explains the profile; none of it feeds the Statz Rating.
Lane occupancy
The vocabulary
The coaching terms this section uses, each with what is actually measured — no definition claims more than the data carries.
The print
The player's full-season Opta event print — every open-play pass, take-on, inferred carry and defensive action, browsable by concept and filterable on the pitch below. Post-recovery and carry figures are minute-window proxies (the feed carries no seconds); percentiles are within position cohort across Big-5 outfielders, 2025/26 only. No footedness or time-to-release figures — the event feed carries neither body part nor seconds; we don't fake them.
The print
Career trajectory
The selected player across the three seasons on file. The Statz Rating is cross-season comparable — its absolute-output benchmarks are pinned once and reused in every season — while percentiles, roles and archetypes are judged within each season's own position cohort. The player id is stable across seasons, so this is one player, three snapshots.
Loading the career index…
Most similar players
Ranked by SCORE — cosine similarity of the full z-scored per-90 profile (100 = statistically near-identical). ROLE compares only what a player does; RATING is the Statz Anchored Role Rating — absolute output blended with peer role quality. Click a row to load that player's card.
Find a younger version
The recruiter's question: who plays like the player above, only younger? The pool is every same-position qualifier in the selected season with a usable profile; each candidate is scored on stylistic match — cosine similarity of the standardised per-90 profile — blended with rating and age gap. Click a row to load that player, then run the search again.
Style match = cosine similarity of standardised per-90 profiles within position; younger = lower age in the selected season. The match is ranked on style, rating and age only — never fees or availability; the contract and market-value figures shown on a player's card are context, not inputs to this search.
Style map
Every qualifier projected onto two principal components of their per-90 profile. The selected player is the highlighted point; their ten most-similar players wear the paired series colour — the legend under the chart names both. Nearby dots play alike.
Where this player lives among 1,585 peers
Principal components of 22 per-90 metrics · PC1 42.9% of variance, PC2 18.4%
Notes & caveats
1 Percentiles are within position cohort (AM and W ranked separately), not possession-adjusted; defensive counts favour players on lower-possession sides.
2 FotMob leaderboards mix per-90 rates and season totals;
totals were converted with total × 90 / minutes after empirical unit classification
(median ratio 1.001 vs Understat-derived per-90 across 1,167 joined players).
3 The Aerial category is omitted (FotMob's 35-stat catalogue has no aerial-duel board); Progression therefore leans on xGChain, xGBuildup and long balls.
4 Every one of the 1,585 Big-5 players with ≥900 minutes this season is included; ratings, roles and archetypes are computed within position group across the full pool.