The Silent Crack in Mid-Season

The Silent Crack in Mid-Season

In the last three matches, the PPDA of a team sitting in the European qualification places has risen from 8.4 to 11.9. To an outsider, that is a meaningless string of numbers buried in a spreadsheet. To me, it is the first crack. PPDA — the number of passes a team allows its opponent before making a first defensive action — is lower when pressing is more aggressive. When it climbs by 3.5 units in just three matches, that team's pressing structure is quietly rotting, even though the league table does not yet show it. They are still fourth. They are still two points clear of the chasing pack. The stands still applaud every back-pass.

I have sat long enough in front of data tables to know that real collapse never begins with a defeat. It begins with a number ticking upward in silence, several rounds before the headline appears.

Context: a method for reading what the eye cannot see

Forty-four years of watching this industry have taught me one simple thing: football is told in two languages, and most of the public only hears one. The first language is the scoreboard — results, goals, cards, things that flash for a second and vanish from memory within a week. The second language is the data stream behind it — volume of chances, quality of chances, pressing intensity, squad structure, and even the psychological variables no camera can capture.

I began my work as a sports betting analyst in 2026, at the age of 51, in Kuala Lumpur. In my first article, I introduced two concepts the analysts of the day called the tricks of a number-obsessed man: xG, or expected goals, and PPDA, the pressing intensity metric. I did not argue with anyone. I simply built a model from 387 matches across five major European leagues and let the numbers speak.

The result surprised even me. Underdog teams that took the lead tended to retreat too deep, causing the opponent's xG to spike between the 60th and 75th minutes. I called it the Retreat Effect. An exclusive contract from a betting platform arrived three weeks later — not because I had convinced anyone, but because the data had convinced them on my behalf.

Since that day, I have held one unbreakable rule: never write a judgement without a specific figure. Hypothesis first, statistical verification second, conclusion last. No sentiment, no bias, no flattering the crowd.

The annual season is the harshest environment for this method, because it has no clear ending. There is no single final for everyone to hold their breath over. There is only a long, slow current where form drifts, fitness erodes, and refereeing controversies accumulate into pressure beneath the surface of the table. Readers follow every round, and my job is to show them the signal before it becomes a headline.

This article is one sitting-down with that current. No club is named as a defendant. Only the numbers, and the story they whisper.

PPDA and the trap of aggression

Pressing is one of the most misunderstood concepts in modern football. Fans see a player charge forward, a midfield line swarm, and they call it spirit. But pressing is not an emotion. It is a system, and every system has a tolerance threshold.

PPDA measures how many passes an opponent is allowed before the defending team intervenes. An elite high-pressing side typically keeps PPDA between 7 and 9. This number does not say whether a team is strong or weak; it says which risk they have chosen to endure. The fiercer the press, the larger the space behind. It is a calculated gamble, not a moral quality.

What caught my attention this season is a rising PPDA trend in a group of teams that share one trait: a congested fixture list. Across their last three matches, their average PPDA climbed from 9.1 to 11.4. They no longer dare to swarm. They begin to wait. And when a team waits, it hands control to the opponent.

There is a paradox here I want to stress: the decline of pressing usually comes not from losing the ability to run, but from losing certainty about teammates' positions. When the schedule is dense, tactical training sessions are cut short, rotation breaks the links in the chain, and players hesitate to push up for fear of leaving space behind. That is why PPDA rises before goals are conceded, not after.

I have verified this across many seasons. In the sample I track, teams whose PPDA rises continuously for three rounds while still getting good results tend to collapse in the following four to six weeks, when the fixture density leaves no room to recover. Good results deceive the coaching staff. They think they are in control, when in fact they are slowly spending capital that has already run dry.

When xG rises up, I see the people in front of the screen split into two worlds: those who can read, and those who can only look.

Goals lie, xG does not

Nothing is more counterintuitive than saying that goals can lie. But it is a truth any data analyst must face.

Imagine a team firing 28 shots, holding 74 percent possession, and losing 0-2. On the scoreboard, that is a disaster. On the xG board, if those 28 shots produced only 1.15 expected goals, it is not a disaster. It is a reasonable outcome dressed in the clothes of randomness.

xG assigns each shot a probability of becoming a goal, based on position, angle, shot type, and the situation leading to it. A shot from close range with a wide angle might be worth 0.4 expected goals. A 30-metre shot through a wall of defenders might be worth only 0.02. When you add them all up, you get a number that expresses the true quality of the chances, separated from luck.

Germany's 2026 defeat to South Korea is the classic example I remember vividly. Germany held 74 percent possession, fired 28 shots, and generated only 1.15 xG. That figure tells a very different story from the one 28 shots tell: Germany were not dangerous. They held the ball a lot, but their shots came from positions where the probability of scoring was near zero. They were banging their heads against a wall and calling it attack.

Germany collapsed before the World Cup kicked off; I only heard the crack from the silent numbers in the data table.

The most important thing about xG is not its absolute value, but the gap between xG and actual goals across many matches. A team that persistently scores more than its xG is living on individual excellence — and individual excellence is an asset that can run out. A team that persistently scores less than its xG has a finishing problem — and that problem usually self-corrects. This is the foundation of every forecasting analysis I have ever done.

Yet I must be careful here, because I myself have erred by trusting xG too much. Some teams build their game on creating many medium-quality chances rather than a few high-quality ones. For them, total xG may be low, but the relentless pressure they generate still has genuine tactical value. Football is not played on a spreadsheet.

The Retreat Effect: when fear becomes a variable

Of all my findings, the Retreat Effect is the one I am fondest of, and the one that made me understand that data measures not only actions but fear.

When an underdog takes the lead, the instinctive reaction is to drop deep to protect the result. It is a survival instinct, and in many cases it is rational. But my data shows that most teams retreat more than necessary. They do not merely fall back to defend; they fall back to the goal line. And that very act hands the opponent full control of the middle third.

Between the 60th and 75th minutes, the opponent's xG spikes systematically against teams that are leading. It is not because the opponent suddenly plays better, but because the space they are given has changed. A team that drops deep creates a paradox: it protects the goal by letting the opponent comfortably fire shots from the edge of the box, where xG is only about 0.05 to 0.08 per shot — but when the volume of shots is large enough, the accumulated xG crosses the danger threshold.

The mathematics of retreat is cruel: 20 shots at 0.06 xG equals 1.2 expected goals, more than a single 1.0 xG shot the opponent might never create if the leading team held its structure.

This explains why many teams are pegged back late despite defending in numbers. They think they are protecting, but in truth they are inviting. A crowd of players in the box creates a sense of psychological safety, not mathematical safety.

I presented this finding in an article a few seasons ago, and it met fierce backlash from those who believe in traditional defensive play. I did not argue. I simply offered the data from 387 matches and let it defend itself. Three weeks later, a betting platform offered me an exclusive partnership. Data does not need anyone to speak for it.

Five substitutions and the war of attrition

The five-substitution rule, introduced as a temporary pandemic measure and then made permanent, is one of the most important changes in modern football. It has two sides, and both deserve serious analysis.

First, it rewards teams with squad depth. A big club with two quality line-ups can rotate its entire attack in the second half without losing power. This deepens the gap between rich and poor clubs, because squad depth is the most expensive thing in football.

Second — and this is rarely discussed — it turns the final 20 minutes into a war of attrition. When both teams have five substitutions, the match is no longer a contest between two starting elevens, but between two benches. The tempo in the final 20 minutes rises sharply, because substitutes have fresh legs while those who stay on have run dry.

The data I track shows the number of goals in the final 15 minutes has risen markedly compared with the era before five substitutions. This means teams without squad depth face a battle they cannot win — not because they play badly, but because the structure of the rules works against them.

This raises a question about the fairness of the season. When a small club leads a big club at the 60th minute, it faces a double paradox: retreat deep and it triggers the Retreat Effect; keep pressing high and it runs out of energy before an opponent with a better bench. Neither option is good. It is a structural trap, not a tactical mistake.

I believe analysts are underestimating the impact of this variable. We tend to analyse matches as pure football events, when in reality they are increasingly like a calculated fitness contest, where the substitution rule is part of the tactics.

Set pieces: where data is undervalued

For years I have called set pieces the dark zone of football analysis. People talk about attack and defence as the two main states of the match, but most goals in modern football come from situations where the ball is not rolling — throw-ins, corners, free kicks, penalties.

My data shows the share of goals from set pieces has risen steadily across seasons. This is not random. As teams organise defence ever more tightly in open play, set pieces become one of the shortest routes to breaking the balance. They turn a seemingly dead situation into a structured attack.

Leading clubs recognised this long ago. They hire dedicated set-piece specialists who focus solely on designing corners and free kicks. This is a sign of how specialised football has become. No single coach can cover every facet of the game.

From a data perspective, set pieces are a paradox. They have a low average xG but high variance. A typical corner has very small xG, but executed correctly it can produce a header from a position with far higher xG. This is why judging a team by total xG while ignoring set pieces is a serious error.

I once saw this in a match I analysed a few seasons back. A team rated as defensively sound based on a low goals-conceded rate turned out, when I separated the set-piece data, to be extremely weak at defending corners. Their goals conceded had not come yet out of luck, not structure. And when luck ran out, the structure showed. That is how I learned that aggregate data can hide detailed truth.

Fitness, fixtures, and the forgotten variable

One of the biggest differences between a data analyst and an ordinary viewer is how they see the fixture list. To a viewer, it is a sequence of matches. To an analyst, it is a sequence of measurable energy expenditures.

The annual season is the harshest test of fitness. There is no long winter break to recover, no period slow enough to regenerate. Teams must play at a dense cadence, and the difference between a title winner and a fading side often lies in energy management, not raw talent.

I measure fitness indirectly through several indicators: kilometres covered per match, number of sprints, average pressing distance, and most importantly the change in these indicators across rounds. A fit team does not merely run a lot; it runs a lot consistently. When pressing distance falls while the number of pressing actions rises, that is a sign of functional fatigue — players still try but are no longer fast enough to create real pressure.

Interestingly, fatigue often does not show in the first half. It shows from the 70th minute onward, when decisions slow and defensive positions loosen. This is why I always separate data by match time window. A team can dominate the first half and collapse in the second, and if you look only at full-match totals, you will miss the real story.

There is a personal anecdote I often tell about the importance of splitting data. One season, I received a request to analyse a team in dreadful form. At first, aggregate data showed they were completely overrun. But when I split the first 30 minutes from the last 30, a different picture emerged: they played evenly for most of the match and collapsed only in the closing phase. Their problem was not talent but fitness and squad depth. Those are two fixable problems, and I advised my client based on that truth.

Referees, VAR, and what cannot be measured

An honest data analyst must admit his limits. Some variables affecting match results cannot be fully measured by data, and referees are one of them.

VAR arrived with a promise of fairness. To some degree it delivered: clear offside and penalty errors are corrected more often. But VAR also created a new kind of injustice — the injustice of digitised subjectivity.

When a referee decides whether to review a situation, he makes a subjective judgement. When he reviews and decides whether to overturn, he is still making a subjective judgement. VAR does not eliminate subjectivity; it only changes where subjectivity is exercised. And because these decisions directly affect results, they become variables my model cannot predict precisely.

This does not mean I abandon analysis. It means I must acknowledge a margin of error I did not have before. In recent models, I add an adjustment factor for the referee, based on the history of each referee's decisions. It is an imperfect effort, but better than pretending the variable does not exist.

The truth is that football always contains an unmeasurable portion. One whistle can change an entire season. One wrong decision can relegate a team or crown one. And no model, however sophisticated, can capture that randomness. It is what keeps me humble, and it is also what keeps me working.

The transfer market: a shattered mirror

The transfer market is like a shattered mirror: each fragment reflects a different fear of the board.

I have spent years analysing transfers, and what I learned is that a player's value is decided not by his quality but by the buyer's fear. A team fearing relegation will overpay for a mid-level player. A team fearing the loss of a European place will pay through the nose for a goalscorer. This is what I call the panic premium.

The panic premium is not an abstract concept. It is measurable. When I compare a transfer fee with a player's estimated market value, I often find a significant gap at the end of the winter window — the moment when teams are in a relegation fight or a title race. That gap is the price of fear.

There is an irony in how the market operates. Clubs often buy players to solve a short-term problem but sign them to long-term contracts. A striker bought to score in the final six months of a season can become a wage burden for the next four years. This is the mismatch between the timeframe of fear and the timeframe of the contract, and it is the cause of many club financial crises.

From a data perspective, I assess a transfer through three main indicators: the player's age and development curve, his fit with the tactical system, and the contract structure. A 28-year-old may be at his peak, but his transfer value will fall fast over the next two years. A 23-year-old may be incomplete, but his development curve lets the club recover capital or profit. These are the calculations boards often ignore when they are panicking.

The limits of the model: when data also trembles

I must tell a story I rarely tell, because it contradicts the very image I have built over many years.

When football paused in March 2026, I thought I had a long holiday. I had spent years building a model from thousands of matches, and I believed it had captured the core laws of the game. I was wrong.

When football returned to empty stadiums, my model began to deviate systematically. The draw rate rose 23 percent above the historical average. Home teams won markedly less. Teams I predicted to win dropped points instead. I did not understand what was happening.

Empty stadiums broke my faith in data in silence — because when the noise vanished, I realised that data also trembles.

It took me a while to grasp the truth. For years I had overpriced home advantage. I had treated it as a constant, a fixed law of football. But home advantage is not a constant. It is a variable made of cheering, psychological pressure on referees, the confidence of home players, and the hesitation of away players. When the crowd vanished, all those factors vanished with it.

I withdrew for three months. I rewatched 212 post-lockdown Bundesliga matches, one by one, and built an adjustment factor I called neutral-adjusted xG. I delayed a newspaper delivery by two weeks just to perfect it. It is the habit of a perfectionist, and sometimes it slows me down, but it is also what keeps my work credible.

The biggest lesson I drew was not about home advantage. It was about the nature of data. Data is not truth. It is a photograph of a changing world. When the world changes fast enough, the photograph becomes outdated, and the analyst must be humble enough to take a new one.

Since then, I never write about data as absolute truth. I always pose the counter-question: if the environmental conditions change, does this model still hold? If the crowd returns, which variable shifts? If the rules change, which assumption collapses? These are the questions an honest analyst must ask himself every day.

Each signal from the data is not an answer; it is a door opening onto another corridor that must be illuminated.

Correlation is not causation: the blind spot of the number-reader

This is the hardest part of the job, and the part I want to attach my most serious warning to.

In recent years, data analysis has become a trend. TV programmes talk about xG as if it were a form of truth. Pundits quote PPDA as if it were praise or blame. And most dangerously, many people have begun to believe data can predict everything.

It cannot. And the difference between correlation and causation is where every mistake begins.

A high-pressing team may win many matches. That does not mean high pressing causes the wins. Perhaps that team simply has better players, and those better players both press well and score well. When you mistake correlation for causation, you build a model on an illusion, and that model will collapse when circumstances change.

I have watched many colleagues make this mistake. They find an indicator correlated with success and turn it into a formula. But football does not run on formulas. It runs on complex systems where every variable interacts in ways no model can fully capture.

This is why I always stress that data is a lens, not a truth. It helps you see more clearly, but it does not replace your eye. And it never replaces the humility of the one using it.

There is a gap I call the gap between the reader and the looker. The looker sees the score and believes he understands the match. The reader sees the data stream and knows he understands only a part. The true reader is not the one who knows many numbers; he is the one who knows the limits of those numbers. And within that gap lies an opportunity to educate — not to belittle the looker, but to lift him up.

I never write to prove I am smarter than my readers. I write to share a way of seeing, and to let readers draw their own conclusions. If there is one thing I want to pass on after forty-four years, it is this: doubt every number, even the numbers you like.

Signal for the next round

As the season continues, I will track three signals I consider most important: the PPDA trend of teams with congested fixtures, the gap between xG and actual goals for teams in good form, and the share of goals from set pieces in the final 15 minutes.

These three signals do not tell me which team will win the title. They tell me which team is living on borrowed capital, and which is building on real foundations. The difference between the two tends to reveal itself only when the season enters its decisive phase, when everything is tense and there is no room left for luck.

Football will always be a game of uncertainty, and that is what makes it beautiful. But within that uncertainty, there are repeating patterns. The analyst's job is not to predict the future but to recognise those patterns before they become the past.

Viewers believe in drama; I believe in repetition; and drama repeats too if we wait patiently. I will keep sitting down, opening the data table, and listening to the silent crack of numbers — because there, before anyone else sees it, the truth of the match has already begun to speak.