Over the last three matchweeks, the PPDA of a mid-table club has fallen from 8 to nearly 13 — meaning each of their defensive engagements now waits for the opponent to complete almost five extra passes. On the broadcast graphic, that figure glows red like a drained health bar, and the presenter calls it a sign of collapse. I stayed behind after the shoot, rewound the tape, and saw something else entirely: a data table that is empty in the most dangerous place — the denominator.

Three matchweeks is too small a sample to conclude anything. But inside the churn of the regular season, three matchweeks are just enough to spawn a narrative, a hot take, and a few tens of thousands of engagements. Storytelling speed is outrunning data accumulation, and that is the moment when sports analysis digs its own grave.

Context: When Every Stand Has a Stat Sheet

Football viewers have never had access to so much data. A single league match is now logged as thousands of events: every pass, every sprint, every metre covered, every touch inside the box. Analytics platforms sell data packages to broadcasters, to clubs, and to content people like me. Statistics have become a shared language, and the broadcast graphic has become a stage.

But there is a paradox few state out loud: more data does not mean better conclusions. When a new season begins, every metric is computed on a tiny sample — three, four, five matches. One own goal, one red card, one lucky penalty can bend an entire trend line. And because the table looks so “scientific,” readers assume it is trustworthy.

For years I wrote about sport under a rule of my own: every argument needs a statistic as evidence. That rule helped me stand firm against doubt, but it also taught me the reverse lesson — a statistic does not produce a conclusion; the sample size is what decides whether that conclusion is valid. When the denominator approaches zero, every percentage becomes a magic trick.

Here is a simple way to picture it. Imagine you flip a coin ten times and get seven heads. You could declare the coin biased, but anyone who does statistics will tell you ten flips are not enough to conclude. Early-season football is that ten-flip run. The problem is that we do not flip the coin in silence — we flip it in front of millions, with a graphic and a headline already waiting.

The Core: Re-reading the Three Most Misread Metrics

PPDA — a measure of pressure, not of quality

PPDA is the number of passes an opponent is allowed to complete before the defending team makes a defensive action — a tackle, an interception or a foul. The lower the number, the more intensely a team presses. It was born in the analytics community, where independent analysts tried to quantify pressure, and broadcasters quickly adopted it because it is intuitive and easy to chart.

The problem is that PPDA measures how a team defends, not how effectively. A team that deliberately concedes possession and sits deep will show a high PPDA entirely on purpose. Reading that number as a sign of decline is misreading the nature of the data. In esports, people clearly separate “pressure” from “control”: a team may choose not to apply pressure in order to hold its shape, and that is a tactical choice, not a mistake. A team that controls the map by standing in the right places does not need to dive into duels, so its PPDA will be high. Looking at the number, you think it is passive. Looking at the map, you see it holding position.

Distance covered — an effort metric packaged as a value metric

This is the metric that bothers me most. Distance covered and sprint counts get packaged as praise: the team that runs most is the team that cares most. But ineffective running also produces pretty numbers. A back line that is constantly stretched and forced to chase the ball will record far more distance than a back line that holds its shape and never has to chase. A high number here is a sign of being reactive, not of effort.

In gaming terms, this is a “damage dealt” figure with no “damage taken” beside it. You deal heavy damage but let the enemy walk past you. Reading the damage number, everyone thinks you are playing well. Reading the map, you are losing. Football is the same: the team that runs the most in a match is not necessarily the best team — it may simply be the one being dragged around.

xG — expected goals and the limits of a small sample xG, or expected goals, is the probability that a shot becomes a goal, based on location, angle, shot type and context. It is one of the great contributions of modern analytics, separating outcome from process. But xG is also the metric most sensitive to small samples. After three matches, a team may have high xG without scoring, and people call them “unlucky.” After three other matches, a team may score more than its xG, and people call them “clinical.” Both conclusions are drawn from a sample whose error exceeds the signal.

Regression to the mean always wins in the long run. Teams that outscore their xG usually come back to earth. Teams that undershoot it usually explode. But in the short early-season window, both can keep drifting far from the mean — and that is when the media gets it most wrong.

A real denominator: Morocco and the 2026 World Cup

Recall Morocco at the 2026 World Cup. Before the knockout rounds, coach Walid Regragui’s side was widely labelled “anti-football” for conceding possession. In the round of 16 against Spain, Morocco held far less of the ball while Spain dominated it. The result: Morocco won on penalties and became the first African nation to reach a World Cup semi-final.

Read only the possession table and you conclude Morocco were weak. Read the actual structure of the match — a low block, disciplined distances, clearances inside the box — and you see a team that had studied the game state closely. Morocco were not a dark horse; they were a team that had read the meta before entering the tournament.

What is striking is how the pundits reacted. Many called that style “anti-football,” whereas in esports a team that defends to protect an advantage is considered legitimate and often wins titles. Defence was never cowardice; the majority simply had not learned to read the survival meta.

Do not compare stats, compare team comps

Another common error is comparing metrics between two teams while ignoring squad structure. A team playing two holding midfielders will record a different number of ball recoveries than a team playing two attacking midfielders. Comparing those numbers directly is comparing two things produced by two different systems. Do not compare stats, compare team comps — modern football is a game of meta.

Take the holding-midfielder role. A holding midfielder in a high-pressing system will record many recoveries in the opponent’s half but few blocks in front of his own box. A holding midfielder in a deep block will show the opposite picture. If you look at a single column of numbers, you will misjudge both. N’Golo Kanté was once described by me as a “map-opening support” because his contribution lay in clearing space — work that the basic stat sheet omits.

Fitness and fixture density

There is one more variable the early-season graphic tends to hide: fixture density. The modern regular season runs at a punishing pace, and that density is one of the biggest causes of injury. No medical staff can save a team playing two matches a week for months on end. When a team’s PPDA drops or its distance covered rises, the cause may not be tactical — it may be legs that have run out of charge.

This is where metrics and fitness intersect in a way viewers cannot see through a screen. A team pressing less may be saving energy for a midweek match. A team running more may be compensating for players who have not recovered. Reading the numbers while ignoring the calendar is reading half the story.

A lesson from esports: early ranking does not decide the season

I began my career in esports, and there is something there that football is relearning from scratch: how to read a long season. In ranked games, players go through placement matches to set an initial rank. Nobody judges a pro on ten placement games alone. People wait through at least one patch cycle — when the meta shifts, when the roster settles, when form finds its true horizontal line.

A football season is the same. The first three matchweeks are like placement matches: enough to seed you, not enough to conclude. The hasty will write stories they must delete in October. In esports, people talk about the “patch cycle” — a period long enough to reveal who is genuinely good. Football has an equivalent, only slower: a squad cycle, a fitness cycle, a transfer cycle.

My tracking experience

Based on my experience following matches across many seasons, I have noticed a repeating pattern. In August and September, the media builds up “phenomena”; by November, half of them have vanished. In 2026, when I was still writing a personal blog, I counted 14 explosive acceleration runs from Kylian Mbappe in the World Cup final and called him a true hypercarry at 19. That conclusion was right, but its rightness came from choosing the right match — a final, where the sample was large enough and the pressure high enough. Had I counted those 14 runs in a July friendly, the conclusion would have meant nothing.

In the summer of 2026, at the Euros, I wrote about Lamine Yamal after his goal against France in the semi-final, when he was just 16 and became the youngest scorer in the history of the European Championship. I called him an “early-game prodigy” with successful dribbles. A middle-aged female reader remarked that she wanted to understand this boy, not learn gaming slang. I rewrote the whole piece, kept only three comparisons, and the new version reached three times the audience. The lesson was not to drop gaming language, but to translate it for people who have never played. And it was also a lesson about sample size: a semi-final goal at the Euros is a big observation, but it is still one observation.

The Contrarian Angle: The “Idiot Coach” Trap

When responsible contrarianism becomes a brand, there is a strong temptation: to blame every bad result on one wrong coaching decision. After three matchweeks, a team loses twice, and immediately there is a wave demanding a sacking. But two very different things must be separated: a wrong decision and a bad outcome. A coach can make a tactically correct decision and still lose to a penalty, a red card, or simply because the opponent finished better on the day.

In esports, people call that “variance.” A correct play can still lose a single game, but repeated a hundred times it will win most of them. Judging a coach on three games is like judging a pro on three ranked games: you are measuring variance, not skill. A derby is not a match, it is a life-or-death ranked game — whoever tilts first loses first, and one derby result should not be used to measure a whole season.

This does not mean every decision is immune to criticism. Some choices are clearly wrong at the systemic level — a lopsided formation, late substitutions when the team has run out of charge, keeping an overloaded player on the pitch. But to point out that wrongness, you need more than three matches, and more than one stat sheet cut off from its context. A good coach is not the one who owns the most stars, but the one who builds a squad from undervalued pieces.

The Key Takeaways

When the denominator approaches zero, every ratio becomes an illusion. This is the first principle for anyone working with data, and the one most often violated at the start of a season. A stat table says nothing on its own; it only says something once you know how many observations it was built on.

Metrics measure behaviour, not value. Distance covered tells you how far a team ran, not whether they ran to the right places. PPDA tells you how a team pressed, not whether the pressing worked.

Defence is a legitimate tactical choice. Sitting deep, conceding possession, waiting for the counter is not cowardice. It is a composition built for the late game — accepting an early weakness to be strong late — and it has won major titles.

Closing

Over the last three matchweeks, the PPDA of the team I opened with has dropped, and that might mean they are declining. It might also mean they have just faced three long-ball opponents, or lost a holding midfielder, or are simply testing a new structure. The only thing I am sure of is that three matches are not enough for me to dare write a conclusion. And perhaps, in a season still long ahead, that patience is exactly what the data has been trying to teach us.