Why per 90 minutes matters for player props

Per 90 is the number that lets you compare players who don’t play the same amount. Two forwards can both have eight goals and be completely different players once you account for how long each spent on the pitch to get there.

What per 90 actually is

Per 90 takes any counting stat, shots, tackles, key passes, touches in the box, and expresses it as a rate over a full match. The calculation is simple: divide the stat by the minutes the player was on the pitch, then multiply by 90.

A midfielder with 40 shots across 2,000 minutes has 40 / 2,000 x 90, which is 1.8 shots per 90. A striker with 50 shots in 2,500 minutes has 50 / 2,500 x 90, which is the same 1.8 per 90. Their raw totals differ by ten shots, but they’re shooting at the same rate once you account for the time each spent playing. That’s the whole point of the number: it puts players on a per-match footing regardless of how many matches, or how much of each match, they actually played. FootyMetrics tracks minutes alongside every counting stat, so the per-90 rate is just the ratio between the two.

Why raw totals mislead when minutes differ

Raw season totals reward availability, not rate. A player who starts every week will out-total a player who comes off the bench even if the substitute is more productive per minute, simply because the starter has far more minutes to accumulate in.

Say two forwards have both scored eight goals this season. On the surface they look identical. But if one started 30 matches for 2,700 minutes and the other managed eight goals in 900 minutes off the bench and in the odd start, they aren’t the same player at all. The first is scoring at 0.27 per 90; the second at 0.80 per 90. Raw totals hid that completely. The bench player is finishing at nearly three times the rate, and that only shows up once you rescale for minutes.

This is exactly why per 90 is the right lens for a lot of props work. A prop line is set for a single match, so a per-match rate is closer to what you’re actually betting on than a season total that lumps 30 games together.

A worked example: the sub who looks electric

Here is the trap. A starter plays the full 90 minutes and has 3 shots, a per-90 of 3.0, which is a strong return for a genuine goal threat. A substitute then comes on for the last 30 minutes of the same match and has 2 shots. Run the sub through the formula: 2 / 30 x 90 = 6 shots per 90, double the starter’s rate.

Six shots per 90 is an elite rate. Almost no established starter sits that high across a season. If you sorted a props table by shots per 90 and this player had a minutes filter switched off, they’d top it, ahead of every genuine high-volume shooter in the league.

But it’s one cameo. Thirty minutes is a third of a match. The two shots might have come because the player’s team was chasing a goal late and throwing bodies forward, a situation that inflates shot counts for anyone on the pitch. Extend that same player over a full 90 and they’re far more likely to regress toward a normal rate than to keep firing off six shots a game. The per-90 figure took a single burst and presented it as a sustained rate it has done nothing to earn.

The problem isn’t the arithmetic. Two shots in 30 minutes genuinely is six per 90 at that moment. The problem is treating a rate built on 30 minutes as if it carries the same weight as one built on 2,500.

Why per 90 needs a minutes threshold

A rate is only as trustworthy as the sample it’s built on. The fewer minutes behind a per-90 number, the more a single good or bad spell swings it, which is the same small-sample problem that governs any short betting record. One hot cameo can double a low-minutes player’s per-90; one quiet one can halve it. Over 2,000 minutes, no single match can move the rate much at all.

There’s no single official cut-off, and different analysts use different ones, but the working principle holds generally: apply a minutes filter before you trust a per-90 rate, and be more sceptical the lower the minutes behind it. A few hundred minutes is often treated as a rough floor for taking a rate seriously, and a full season’s worth is more reliable again. Below the floor, read the per-90 as a hint about a player worth watching, not as a settled rate you can bet into.

A per-90 built on 30 minutes and one built on 2,500 are the same arithmetic and nothing like the same evidence.

FootyMetrics applies minutes filters on its trend and props tools for this reason, so a low-minutes player’s flattering rate doesn’t sit at the top of a table unqualified. This is the same reasoning behind variance and sample size in betting: a small sample can fake a signal in either direction, and only more of it tells you whether the rate is real.

The props angle: minutes are priced in

Here is where per 90 meets the bet slip. A bookmaker doesn’t price a shots or tackles line off a player’s per-90 rate in isolation. It prices for the minutes that player is expected to get in that specific match. A nailed-on starter expected to play 90 gets a line built for a full shift. A rotation risk or a likely early substitution gets a shorter line, because the book has already discounted for the minutes it thinks the player will actually be on the pitch.

That matters two ways. A strong per-90 rate is worth far less if the player isn’t expected to play a full match. A forward shooting at 3.0 per 90 who gets hooked on the hour isn’t a 3.0-per-90 bet on the night; they’re more like a 2.0-shot expectation, and the line will reflect that. The flip side is that the real value often sits in the gap between the minutes the market expects and the minutes you think the player will get. If you have a read that a player predicted to be rotated will actually start and go the distance, a line priced for a cameo can be genuine value.

This is why minutes risk and per 90 have to be read together. The rate tells you how productive a player is per match; the expected minutes tell you how much of that match you’re actually getting. A high per-90 from a player at real rotation risk is the props equivalent of the low-minutes trap: an attractive number attached to a situation that may never let it pay out. Most of that minutes risk sits in the gap between a guessed XI and the one the club actually names, which is the subject of predicted vs confirmed lineups.