Game Value · player
Does it see more than plus‑minus?
Each participant rates every appearance in five leagues. The ratings freeze at the round’s deadline and are then asked to order football that had not been played yet — against the same question answered by each player’s own on-pitch goal difference. What is published is the increment, and a rating that quietly follows the scoreboard scores zero.
Player Value ranking
Did the goals follow your players?
Week ending 2026-W43, 2026/2027. Each match is cut into three segments, so a substitution compares the same side with and without a player against the same opponent on the same afternoon; what is published is the increment over each player’s own on-pitch goal difference. The second ranking is never blended with this one.
| Rank RankOrdered by the skill score, because a table has to be in some order. The ordering is not the finding — two rows a hundredth apart are not separated by anything this board can measure. Read the standing under each number, which is the interval's answer, before the position. | Participant ParticipantThe name this row is published under, and how it got here. A row marked 'entered by the Arena' is a published rating the Arena reads and enters on the vendor's behalf: it did not submit, and it is scored exactly like one that did. A row marked 'scored under an exception' rests partly on segments played before it certified, under a stated reason — open the badge for the reason. | Skill score Skill scoreThis participant's Somers' D minus naive plus-minus's, over the identical segments. Zero means adding nothing to what each player's own on-pitch goal difference already says, and a rating that quietly follows the scoreboard scores zero by construction. The interval is paired — participant and baseline are resampled on the same fixtures, so their errors cancel — and a row counts as separated only when it excludes zero. The words beneath the number are that interval's reading. | Somers' D Somers' DHow often the rating orders a pair of segments the way the goals did, over the pairs the goals themselves order: (C − D) ÷ (n₀ − n_Y). Pairs form only within a stratum — one segment index, one score state — so a signal cannot earn credit for knowing which side was already ahead. Beneath it, naive plus-minus's D on the identical segments: each player valued by the goal difference accumulated while he was on the pitch before this match. The difference between the two is the skill score. Plus-minus differs slightly between rows because it is recomputed over each participant's own scored segments — the same signal, different football — which is what makes the difference between rows a difference about ratings. τ-b, with ties kept, is on the participant page. | Lineup-blind floor Lineup-blind floorA signal that knows only each club's record: points per match before the match, home minus away, no lineups and no ratings at all. It is published and ranks nothing. Beating plus-minus is the claim; clearing the floor is the precondition, because a rating that merely encodes which team is stronger scores about what the league table scores. | Coverage CoverageSegments this participant was scored on, over the segments the board could score. A segment needs prior ratings for at least nine of the eleven on the pitch on both sides, so early rounds — where nobody has a prior appearance yet — are scorable for nobody. Beneath it, the share of on-pitch players carrying a prior rating; where it says 'ours', part of the gap is football the Arena could not put in front of the participant because a player identifier would not cross to ours — our gap, not theirs. | Right winner Right winnerDescriptive, and nothing is ranked by it. Of the segments with a goal difference, the share where this participant's signal pointed at the side that won the segment — and beneath it, how many more segments that is than plus-minus called on the same ones. Goalless and level segments are dropped rather than counted as misses: a signal never predicts one. It reads only which side the signal favoured, where Somers' D ranks every pair of segments by margin, so the two can disagree; a few segments either way is noise. |
|---|---|---|---|---|---|---|
| 1 | FotMobEntered by the Arena · not submitted by the vendorScored under an exception Scored under an exceptionPublished a rating per appearance since the season began; delivered on enrolment in September, when the board began accepting them. The Arena confirmed no leakage: the ratings are the source's own published figures, unchanged, and were fixed before any result they are scored against. — 750 segment(s) played before this participant certified, scored under an exception granted 2026-09-27 reaching back to 2026-09-29 Weeks and pair counts | +0.030−0.022 to +0.083Level with plus-minus | +0.232plus-minus +0.203 | +0.138ranks nothing | 69.9%524 of 750 segments · 94.9% rated | 58.2%−5 vs plus-minus |
| 2 | WhoScoredEntered by the Arena · not submitted by the vendorScored under an exception Scored under an exceptionPublished a rating per appearance since the season began; delivered on enrolment in September, when the board began accepting them. The Arena confirmed no leakage: the ratings are the source's own published figures, unchanged, and were fixed before any result they are scored against. — 750 segment(s) played before this participant certified, scored under an exception granted 2026-09-27 reaching back to 2026-09-29 Weeks and pair counts | +0.016−0.035 to +0.067Level with plus-minus | +0.220plus-minus +0.205 | +0.141ranks nothing | 73.1%548 of 750 segments · 95.7% rated | 58.3%−2 vs plus-minus |
The floor is the column to read first. It is what a signal knowing only each club’s record would score on the identical segments — no lineups, no ratings. In both complete rehearsal seasons it landed about 0.004 below plus-minus, which means a rating that merely knows which team is better has already been priced in, and only what sits clear of both is a claim about players.
2026-09-game-value-player-v0Skill score is the participant’s Somers’ D minus naive plus-minus’s on the identical segments, so zero means adding nothing to what each player’s own on-pitch goal difference already says. The interval is paired, and a row reads level with plus-minus whenever it contains zero — the expected first-season result. Coverage is segments scored over segments scorable, with the share of on-pitch players rated beneath it. Right winner ranks nothing: it is how often the signal pointed at the side that won a segment, and can be read segment by segment under Players. How it is computed.How to read it
Level with plus-minus is the honest first-season result
A skill score is a difference between two concordance statistics measured on the same segments. Over three rehearsed seasons the two published rating products separated from each other by about 0.015 — while against plus-minus itself one was flat throughout and the other cleared zero in a single season. So most rows will carry an interval that contains zero, and the board says level with plus-minus in words rather than implying an ordering its own numbers refuse. A row that stays a little clear of the baseline for two seasons has shown something a row sitting at +0.05 for one week has not.
The two rankings are not a first and second opinion. One asks whether the goals followed the players a rating liked; the other asks whether the coach did. A product can be strong on either without being strong on the other, and that is a finding rather than a contradiction. There is no combined figure on this page because the Arena has no defensible weighting between them — inventing one would make up the most important number here.
The selection column also resolves several times coarser than the score, so it separates two products far more rarely. Read the board column by column, or the protocol for the segments, the deadlines, the stratification and how each interval is resampled.