The 75th Minute in Knockouts: How a Quiet PPDA Rise Exposes Tournament Squad-Depth Truth

On 11 July 2026, in Moscow, the Croatia-England semifinal had gone to extra time. I was pulling remote event data from an internet cafe in Rangpur. England had pressed superbly for the first half hour, but in extra time the picture flipped. Luka Modric covered 13.8 kilometres that night, and Croatia's pressing metric (PPDA) dropped to 8.7. Croatia were pressing more aggressively, and England's press was breaking down. The scoreline read 2-1. In my notebook I wrote: England did not lose on the quality of their passing, they lost on a timing error after the 75th minute. Eight years later, that line has become my most valuable lesson about tournament cycles.

Methodology box: Data source - tournament event data (pass logs, pressing triggers), sample of 14 knockout matches; model version xG v3.2; glossary - xG (expected goals), PPDA (passes allowed per defensive action; a lower value means more aggressive pressing), distance covered (kilometres). The sample is small, so this is a provisional verdict, not a final one; I record the review conditions below.

I open every tournament analysis with this box. A number without a source is a rumour, and you cannot pick a team with a rumour. From my years of watching matches, I can say this: in tournament football, the gap between what the spectator sees and what the data shows is widest in extra time. In the group stage, teams stay cautious and the margin for error is small; in knockouts, every error costs more, and fatigue sits like an invisible tax.

There is an arithmetic truth to a tournament cycle: in a 32-team tournament, a champion must play seven matches, denser than a 38-game club season. In club football, after a bad night you can correct things the following week; in a tournament there is no such chance. Every team therefore oscillates between two things - how much to press, and how much energy to conserve. My table shows that teams who survive into extra time bring their pressing intensity under control before the 75th minute.

Core analysis - the data evidence chain

When you lay out the event data of 14 knockout matches, a pattern becomes clear. In the first 60 minutes, teams' average PPDA sits between 9.2 and 11.5. After the 75th minute, that average jumps past 12.6. The press is gradually going passive. I have a rule I write in every tournament piece: if PPDA climbs past 12, the press is passive - that is not a tactic, that is fatigue.

But if fatigue were the only cause, the fix would be simple - run more. The reality is more complex. In that Moscow semifinal I isolated the pressing triggers. England's problem was not a lack of running; it was the decision of who presses when. Croatia played one extra pass in midfield, and England's front line caught the trigger late every time. Every late trigger means two or three extra metres of running. Across 90 minutes, those extra metres pile into a mountain.

I built Modric's press map, and that became the story. The map showed Modric was often pressing at the right moment but over a shorter distance - he ran less, but in the right place. Pressing is not a running contest; it is a positioning game. Whoever closes more space with less running buys more time. This map has a practical side: a pressing system rests on three pillars - the trigger (when to press), the coverage shadow (who closes which space), and transition risk (what lies behind if the press breaks). England's triggers were right, but their coverage shadows were messy, so the midfield emptied the moment the press broke.

This is where an old lesson returns. In 2026, as a junior analyst for a Dhaka-based new-media outlet, I built my first xG model in a Rangpur internet cafe. For the Bangladesh Premier League match between Abahani Limited Dhaka and Sheikh Russel KC, I logged 1,842 passes and 24 shots. The model showed Abahani's 2-1 win was flattered - xG read 1.7 to 0.9. I believed data never lied. I published a 900-word breakdown with raw event data. It was shared 3,400 times. I found the Rangpur spreadsheet did not lie; the derby chose chaos. The model was right, but the match chose a variable outside the model - emotion.

That lesson is even more relevant in tournament football. In a knockout match, teams' pressing behaviour changes abruptly after a goal. My data shows that in the ten minutes after conceding, the losing team's PPDA falls by an average of 2.3 points - they suddenly press more. That is a sign of courage, but also of risk, because more pressing means more space behind, and leaving that space open on tired legs is self-harm.

In 2026, when the pandemic halted the game, I built an empty-stadium model from Rangpur with no live matches. Using Bundesliga restart data, I analysed Bayern Munich versus Borussia Dortmund: the home team's xG fell from 2.1 to 1.4, and home advantage fell from 0.42 to 0.18 goals. For 47 days I published a daily data bulletin, and the outlet's traffic tripled. My editor called it the only reliable content of the shutdown. Since then I have moved from reactive match reports to predictive writing - not what happened, but what should happen if a given condition occurs.

That same logic works in tournament football. Consider an example. Say a team has played the group stage with a PPDA of 8.5 - highly aggressive. In the knockout, if the opponent can move the ball quickly through midfield, that same press boomerangs back. Because pressing on tired legs widens the gap in transition. In my calculation, the lower the PPDA, the higher the transition risk - but only when squad depth is thin.

This is where the real truth of a tournament cycle hides: success in knockouts depends on bench depth, not on the talent of the starting eleven. In a club season you can rotate; in a tournament every match is a final. The team that keeps two comparable players in every position is the one that survives into extra time. My data shows that among teams reaching the last eight, those whose substitutes played at least 25 percent of total minutes scored more late goals.

A subtle but important judgement hides here. Right now Europe's big clubs are busy in a brand war in the transfer market - they want to buy the biggest name, whatever the price. But my table shows real value is created at smaller clubs, who cheaply find suitable players for a specific role. The team that buys the right player for a specific trigger role is the one that can build bench depth. Buying a name and buying a role - the difference shows up in a tournament.

Another misconception also needs correcting. Many prioritise a goalkeeper's ability to hit long kicks, yet keepers whose basic shot-stopping numbers (save rate against post-shot xG) are declining still command big transfer fees. In my data, those saves matter far more in extra time than long kicks. In a knockout, each save is worth a goal.

That Moscow match proved bench depth. Croatia's substitutes added fresh energy in midfield in extra time, and England's tired midfield melted before them. The result - a goal and a final ticket. This is the succession protocol: who comes on and when is scripted in advance. A red-card script, an injury-cascade backup, a next-man-up plan - with these three, a team does not panic in a tournament.

Contrarian angle - correlation is not causation

A warning is necessary here, because the most dangerous error in the world of numbers is to assume that when two things happen together, one caused the other. My table shows PPDA rises in extra time and the probability of losing rises too. But that does not mean a team loses whenever PPDA rises. Fatigue, substitution timing, the scoreline and the opponent's tactics all work together.

I have often erred here. Once, seeing a team's PPDA rise across three straight matches, I concluded their fitness coach had failed. A video audit later showed the problem was the coach's substitution timing - he was making changes at 82 minutes instead of 70. The players could run, but no one was telling them when to stop and when to start again.

So a tournament's pressing decline is not a physical failure; it is a management failure. And a management failure is correctable - far more easily than a physical one. This is my contrarian call: those saying the team is tired are asking the wrong question. The right question is - who was on the bench, and when did the coach bring him on.

I also accept my sample is small, only 14 matches. So I am not giving a final verdict here; rather a provisional one, and I record its review conditions. If the next round shows the link between bench depth and extra-time performance is close to zero, I will treat my threshold idea as wrong and revise the model. The data monk's job is not to lock his verdict in place, but to release it when new evidence arrives.

Takeaway - the next round's signals

In the next round I will watch three things. First, each team's 75th-minute PPDA - if it climbs past 12, I will assume the press has broken. Second, the percentage of total minutes played by substitutes - if it is below 25 percent, the team is at risk in extra time. Third, pressing behaviour in the ten minutes after conceding - courage or self-harm, the numbers will tell.

Tournament football teaches us that a thin line runs between victory and defeat, and that line is often drawn at the 75th minute. The question is therefore not who played better. The question is who knows their own system after the 75th minute. Whoever has a deep bench, whose coach's substitutions follow a script, will have the last laugh. The rest will watch the final on television, a spreadsheet in hand, a question in mind - where was the mistake?