Per 90

Minutes, not matches, sit on the bottom of the fraction

Per 90 takes a season total and rescales it to a common unit: what the total works out to for every ninety minutes a player was actually on the pitch.

The formula is total divided by minutes played, multiplied by 90. A player with 12 goals in 2,700 minutes converts to 12 ÷ 2,700 × 90 = 0.4 per 90.

The minutes are the part that trips people up. A 38 match season does not equal 38 lots of ninety minutes. Stoppage time pushes a full season past 3,420 minutes (38 × 90) for anyone who plays every minute of every match. The real divisor lands closer to 42.5 ninety-minute blocks than 38. Divide by matches instead of minutes played and the answer drifts further from the true rate the more added time gets played across the season.

That is why a per match figure and a per 90 figure for the same raw total will not agree. Per 90 comes out slightly lower, because its divisor, total minutes, is always slightly larger than matches multiplied by 90.

The gap widens as a season goes on. Early in a campaign, matches played and ninety-minute blocks played are close enough that the two figures barely separate. By the closing months, a player can have featured in several matches running five or more minutes of added time. A meaningful gap opens between their per match and per 90 numbers for the same raw total.

Three scenarios where the gap shows up

The hot substitute. A forward comes on for the final 15 minutes and scores once. Scaled to a full ninety, that is 6.0 per 90, a number nobody sustains across a season. The rate is real. The sample behind it is not.

The nailed-on starter. A forward on 3,000 minutes at 0.4 per 90 has scored around 13 goals across real, repeated exposure to matches. That rate has already survived cold games, tough opponents and quiet weeks.

Starter against substitute, judged fairly. Ranking both players by season total favours whoever has played more, not whoever is more productive when selected. Per 90 removes the volume advantage and compares them on the same footing, over whatever minutes each has actually had.

Trust the rate once the minutes back it up

A per 90 figure is only as reliable as the sample behind it. Above roughly 450 minutes, a rate has usually survived enough variation to mean something. Below that, treat it as a description of a hot streak, not proof of a new level.

Check the minutes played column before acting on any per 90 number. A rate with a large total behind it and a rate from one good half deserve completely different weight, even when the number itself looks identical.

Common misreadings

"Per 90 is the same as per match." Close, but not identical, and the gap widens across a season because of stoppage time. The two will not line up exactly for the same raw total.

"A high per 90 rate proves a new level." Not on a small sample. Sixty excellent minutes can put a squad player top of a leaderboard for a week. The number needs minutes behind it before it says anything about what comes next.

"A season total is close enough, per 90 is not necessary." Only when every player being compared has played close to the same minutes. The moment sample sizes diverge, and they almost always do, per 90 is the only fair unit to line them up on.