Relative age effect

Why are elite young players born disproportionately in the months just after the age-group cutoff?

Line up the birthdays of a professional academy's under-twelve squad and they are not spread evenly through the year. They pile up in the months immediately after whatever date the federation uses to divide age groups, and thin out toward the far end of the band.

The first instinct is to look for something in the season of birth — daylight, illness, a mother's winter. But the pile-up does not sit in a fixed month. It sits just after the cutoff, wherever the cutoff happens to be, and where federations have moved their cutoff date the peak has been documented moving with it. That kills the biological story cleanly: whatever is happening is done by the calendar of the competition, not the calendar of the year.

So the cause is administrative. That only sharpens the puzzle, because these are grown professionals. A few extra months of childhood should be worthless by twenty-five. Why does a difference that must have vanished still show up in the roster?

Start with what the cutoff does mechanically. Everyone inside a one-year band is treated as the same age, so two children in a squad can differ by nearly twelve months. At ten, that is a tenth of a lifetime. Around puberty it is far worse, because the spread in maturation opens much wider than the spread in birth dates — two boys the same age can be years apart in physical development.

Now put a selector in front of them. What does a coach at a trial observe? Speed, size, strength, coordination, composure — every one of them inflated by being older and further along. The coach is trying to measure talent and is handed talent plus maturity, with no way to separate them in an afternoon. So the older-in-band child is over-selected. That much is measurement error, and here is where most accounts stop.

But measurement error alone predicts something we do not see. If selection were merely noisy, the mistake would wash out: the wrongly-chosen would be found out, the wrongly-rejected would catch up and be picked next year, and by adulthood the birthdays would even out. They do not. So the error must be doing more than being wrong.

Ask what happens to a child the moment they are selected. More coaching, and better coaching. More competitive minutes against stronger opposition. Expectations placed on them, and a self-concept as one of the good ones. The unselected child gets less of every one. Within a season the two are no longer being measured differently — they are being developed differently.

That is the mechanism, and it inverts the usual reading. Selection did not discover a difference in ability; it created one. A spurious signal was converted into a real skill gap by the treatment that responding to the signal conferred. Run the trial a year later and the coach is now correct: the older-in-band child genuinely is better. The system has manufactured the evidence for its own earlier decision.

Then notice it repeats. Each round takes the output of the last round as its input, so the advantage does not decay — it ratchets. And the ratchet turns one way only, because falling out of a pathway is easy while re-entering one you left at eleven is very hard. By the time the maturity difference has genuinely disappeared, the developmental difference it seeded has had a decade to compound and the rejected have long since stopped playing.

There is a curious tail to this. Several studies find that among relatively younger players who do make it, career outcomes are better — longer careers, higher valuations, more appearances at the top. It is tempting to read this as adversity building character. The disciplined reading is statistical: a late-born player had to clear a higher bar to survive a biased filter, so conditional on surviving it, they must be better on everything the filter was not measuring. The finding is real in several datasets, contested in others, and its size is genuinely unsettled.

Three clarifications keep this honest.

First, this is not a claim that early-born players are frauds. After a decade of superior development, most really are the better players. The fault is not in the current squad but in the counterfactual: comparable late-born children with equal potential were filtered out before the development that would have shown it.

Second, it is not universal. The effect is strongest where physical maturity helps and selection starts early; in disciplines rewarding small, late-maturing bodies the skew can run the other way. Its size varies enormously by sport, country and level, so a single headline figure would mislead.

Third, this is close kin to what happens when schools sort pupils by ability, and the relationship is worth stating rather than blurring. Ability grouping's harm comes from downstream amplification: labelled groups get different curricula, expectations and hours. The relative age effect is the same amplification machinery running on an initial sort that is almost pure noise — which makes the sporting case close to a natural experiment showing the amplification is real even when the thing sorted on carries no information about ability at all. The same caution applies to the underdog tail above, and to survivorship reasoning generally: any group defined by having got through a filter is not a fair sample of who entered it.

• Whether bio-banding, which groups children by maturity rather than birth date, shifts the distribution or merely relocates it.
• Why the same skew appears in school outcomes and in diagnoses of the youngest children in a class.

Every term the collection defines is gathered in the glossary.