Pressing numbers flatter the team that is losing

Every popular pressing metric is contaminated by the scoreline, and nobody adjusts for it before publishing the league table of intensity.

A team goes two goals down at half time and comes out chasing everything. By full time their pressing numbers look magnificent. Nobody looks at those numbers and asks the obvious question, which is whether a side that had to chase for forty five minutes is really the more aggressive side, or just the one that was behind.

This is the contamination that sits underneath almost every pressing statistic in circulation. Passes allowed per defensive action, pressures in the final third, whatever your preferred version, all of them are ratios built on opponent behaviour. A team that leads by two invites pressure, so the opponent's press looks fierce. A team that leads by two also stops pressing itself, because there is no point, so its own numbers collapse. The metric records a difference in philosophy where what actually happened was a difference in the scoreline.

Game state adjustment exists. It is not hard mathematics. Split the match by score and weight accordingly, or restrict to the periods where the game was level, which is usually most of the first half. Very little published analysis does it, and I think the reason is unglamorous. Splitting by state shrinks your sample and shrinking your sample makes the pretty league wide chart look uncertain, and an uncertain chart is harder to write a confident headline about.

The consequence is that we misdescribe teams. A mid table side that defends deep while ahead and presses like a maniac when behind will average out as moderately aggressive, which is a description of a team that does not exist. Their two modes are further apart than anyone's average suggests.

There is a defence of the raw numbers, and it is not nothing. Over a full season, if a side spends similar time in each game state as everyone else, the noise partly cancels. True for a mid table team. Completely false for the sides at either end of the table, which is to say the teams everybody actually writes about.

So the metric works best on the clubs nobody argues about and worst on the ones we argue about constantly.

If you only make one adjustment before quoting a pressing figure at somebody, make it this. Ask what the score was.