A percentile without a named cohort is theatre
"Top 8%" is not a fact about a player until you say top 8% of whom. How PlayMaker picks the comparison group, what a league coefficient actually is, and why the platform sometimes refuses to give you a percentage at all.
David Avramovic 4 min
Ranking a Regionalliga striker against a district-league under-17 on one scale is an error a scout spots in thirty seconds, and it is the default behaviour of any platform that computes a percentile over "all players". The fix is not a better model. It is naming the group.
The cohort is chosen, and the choice is printed
Every percentile on PlayMaker says which population produced it. The selector tries the player’s own competition first; if there are not at least twelve comparable players in it, it falls back to the same tier across competitions, and finally to everyone with league coefficients applied. Whichever rule fired is stated in the label, so "Top 8% of 61 midfielders across all competitions" and "Top 8% in this league" are never mistaken for each other.
Below twelve players the platform shows a rank instead of a percentage. Nine players in a cohort makes every one of them a multiple of eleven percentage points apart, and dressing that up as "top 22%" is a decimal point doing rhetorical work it has not earned.
Age-relative standing is a second sentence, not a fallback
Youth recruitment runs on one sentence — how does this player compare with others born the same year — and a fallback chain can only ever produce one answer. So the age-relative percentile is computed separately and reported alongside: a scout wants both "top 11% in this league" and "top 4% of seventeen-year-olds", and those are different facts.
It widens from the player’s exact age to ±1 and then ±2 years, and then refuses. It does not quietly become an everyone-percentile while the sentence around it still says an age, which is the specific way this figure normally goes wrong.
What a league coefficient is, and why none of them are typed
A coefficient is the multiplier that makes two competitions comparable. The tempting way to build the table is to type numbers into it. The problem with typing them is that the ordering then depends on whoever was typing: nothing stops a youth league ending up rated above the senior side beside it, or a fourth tier above a third.
So a competition declares three things — its country, its age group and its rung in the pyramid — and the coefficient is the product of three ladders. That makes three orderings structural rather than clerical: a higher rung always beats a lower one, a youth competition is always easier than the first team beside it, and a stronger country always beats a weaker one. Country strength applies in full at the top of a pyramid and tapers to nothing at the grassroots, because the gap between two national federations is made of professional infrastructure and none of that reaches the seventh level.
A transfer restates the number immediately
A rating is stated on the terms of the league the player is actually in. A ledger records which competition each stretch of minutes came from, so a promotion restates the figure the day it happens, a relegation restates it back, and the discount fades as minutes are earned at the new level. A player who has never changed level is untouched by any of it.
That also fixes a question recruiters ask constantly and most platforms answer wrongly: has this player improved? Progression is measured on the level-invariant figure rather than the displayed one, because a player who held his level while moving up two divisions has improved, and on the displayed number that reads as flat.
The scales say whether anybody measured them
The 0–99 scales each metric is mapped onto have to come from somewhere. Ideally they are measured from the population — the 5th and 95th percentiles observed in a strength band, for a specific position group, above a minimum sample. On a young database most of them are not measured yet, and are the constants the platform shipped with.
The profile says which. It is a fair question — was this number produced by measuring my league, or by an assumption — and on a young database the honest answer is often "none of it". Calibration is also keyed by position group, not by strength band alone: a save percentage averaged across a band that is nine-tenths outfielders describes how many goalkeepers are in the band, not how well anybody keeps goal.