From World Cup 2026 to Euro 2026: Six Years of Reading Data to Understand Why Every Goal Carries a Hidden Story

On the night of 15 July 2026, at Luzhniki Stadium in Moscow, France lifted the World Cup after a 4–2 win over Croatia. In a small apartment in Los Angeles, more than ten thousand kilometres away, a fourteen-year-old schoolboy still had not turned off his computer. His Excel file had stayed open through all six weeks of the tournament, growing past twelve hundred rows, each row a shot recorded by hand: distance, angle, number of defenders in front, the shooter's preferred foot, the phase of play that produced the attempt. The broadcasters that night praised the beautiful attacking football of the team in blue. The spreadsheet in the boy's machine told a very different story: France won because they restricted opponents to an average of 0.7 xG per match, the lowest figure among the four semi-finalists.

That boy was me. What I learned that night shaped how I have worked for the six years since — from hand-built spreadsheets in middle school, through the pandemic season played in empty stadiums, to Morocco's run to the World Cup semi-finals, and then to an internship at a sports data company in California, where I handled set-piece data for a national team at Euro 2026. Every time I open a new workbook, I remind myself of what the first xG sheet taught me: the first xG spreadsheet taught me that every goal carries a hidden story.

When the ball meets the spreadsheet

Over the past two decades the way we read a match has changed fundamentally. A match used to be summarised by the scoreline, the shot count and possession percentage. All three share the same weakness: they count events without measuring their quality. A shot from 35 metres counted the same as a tap-in from five metres. A sideways pass in your own half counted the same as a through ball that broke two defensive lines. The result was that viewers were led by numbers that looked impressive but were hollow.

Expected goals, or xG, was created to fill exactly that gap. Every shot is assigned a probability of becoming a goal, based on distance, angle, defensive pressure, the body part used and the type of situation. PPDA measures how many passes an opponent is allowed before each defensive action, which makes it a measure of active pressure rather than passive control. Metrics such as progressive passes, xT and field tilt continue to break the idea of "controlling a match" into layers that can actually be measured.

Esports travelled the same road far faster. A single patch can reshape an entire tactical landscape within weeks, which makes its data far more time-sensitive than football's. Champion win rates, pick-ban rates, gold differential at minute fifteen, objective control rates — all are living variables that live and die with each version. Football and esports differ on the surface, but the same data layer sits underneath: both are exercises in reading a match before it is over.

In Vietnam this wave arrived later but is accelerating. Community analytics groups are multiplying, commentary channels increasingly cite xG instead of simply narrating events, and a new generation of viewers now demands evidence for every claim. That is a cultural shift, not a technological one.

The first spreadsheet: France won with defence

Back to the summer of 2026. I had no access to official xG data. Major data sites had not opened free APIs, and I would not have known how to use them anyway. So I did the most manual thing possible: I rewatched all 64 matches, logged every shot into Excel, and scored chance quality on a scale I invented myself. A shot inside the box, unchallenged, on the favoured foot, after a through ball — highest score. A shot from distance, pressed by two defenders, on the weaker foot — lowest score.

More than twelve hundred rows later, three conclusions emerged. First, France allowed opponents an average of just 0.7 xG per match, the lowest of the four semi-finalists. Second, their actual goals exceeded total xG by roughly three, largely through set pieces and long-range strikes at decisive moments. Third, and most importantly: every opponent France beat in the knockout rounds created fewer chances in the second half than in the first. Didier Deschamps' team did not control the ball to attack. They controlled the ball to suffocate their opponent's rhythm.

That conclusion ran against the story global media told. People remember Kylian Mbappé exploding against Argentina, Antoine Griezmann's free kicks and penalties, Paul Pogba's goal in the final. Those moments are real, but they are the visible part. The submerged part was a defensive system organised to the point where opponents almost never got a quality chance — and that only becomes visible when you count.

When home is no longer home

In March 2026 global football stopped. In that gap I decided to do something nobody had given me data for: measure home advantage in numbers. I gathered data from more than three thousand matches across Europe's five major leagues before 2026, normalised by season, removed matches with first-half red cards to reduce noise, and isolated the contribution of the crowd.

The result was stable enough to be suspicious: home teams enjoyed an average benefit of 0.38 goals per match. Part of that came from differences in team quality, but most of the remainder could not be explained by football alone. It belonged to the crowd, to noise, to referees being unconsciously influenced by the pressure of the stands, and to away players passing less boldly when screamed at.

When the Bundesliga resumed on 16 May 2026, matches were played in empty stadiums. I published a prediction before the round: home win rates would fall noticeably, by roughly eight to twelve percentage points against the pre-pandemic baseline. The first three rounds confirmed the model. Home win rates dropped almost exactly within the forecast range, and away goals rose correspondingly.

It was the first time a prediction I built from raw data came true in front of me. But the bigger lesson lay elsewhere: when home is no longer home, you are forced to rewrite every assumption you hold. For years I had treated home advantage as a fixed variable in every model. From the summer of 2026 it became a conditional variable, dependent on the crowd, the schedule, travel distance, and whether the away side had played three matches in seven days.

Morocco 2026: when defensive data speaks first

In November 2026, at eighteen, I launched my own analytics newsletter, inheriting the entire methodology from the 2026 home advantage model. Before the World Cup began I extracted PPDA and average defensive line distance for all 32 teams. The goal was specific: find the most proactive defensive side in the tournament, regardless of how much or how little they controlled the ball.

The result made me re-check the data three times. Morocco ranked near the top on PPDA — meaning they allowed opponents very few passes before launching a defensive action — while their possession share sat near the bottom. In other words, Morocco did not defend by sitting deep and absorbing. They defended by actively squeezing space, keeping the distance between their lines extremely tight, and forcing opponents to make decisions under disadvantage.

I published a call before the tournament: Morocco would not be easy to eliminate, and their likelihood of going deep was being systematically underrated.

When Morocco eliminated Spain on penalties in the round of sixteen and then Portugal in the quarter-final, a tactics account with more than two hundred thousand followers shared my newsletter. In the semi-final on 14 December 2026, Morocco lost 2–0 to France, but the manner of the defeat deserved recording: they held their defensive structure until injuries broke their back line, and still created clear chances at the other end. The line I wrote in November 2026 still holds: Morocco 2026 is the case where defensive data spoke first and the world listened later.

That success brought dozens of connection requests, including one from a senior European analyst, who two years later became my mentor when I landed an internship in California.

Euro 2026 and the problem of valuing players

In the summer of 2026 I took an internship at a sports data company in California, handling two jobs at once: set-piece data for a national team at the Euros, and transfer target evaluation for a mid-table European club.

The second job taught me the most about how models misprice. The club wanted to sign a striker whose expected goals for the previous season exceeded his actual goals by 4.5. The board was worried: a striker who scores few goals is not worth buying. I argued the opposite: a 4.5-goal gap between xG and actual goals, given the volume of chances generated, is not a sign of decline but a sign of bad luck. A striker who keeps arriving in the right positions and shooting from hot zones will regress to the mean, and the market price at that moment was discounted by exactly that bad luck.

The club agreed. The striker scored in the opening match, and my model was confirmed by results on the pitch rather than by a pretty report.

But that was also the period when I almost derailed my own career through perfectionism. I missed a deadline on the set-piece report because I wanted the model to be absolutely accurate, even though it was already at 80 percent and the remainder depended on variables outside my control. A colleague told me something I still carry: a model that is 80 percent right and delivered on time is worth more than a perfect model delivered after the match has ended.

That changed how I write. I still keep a systematic analytical frame, but I compress every report into four salient points and one clear recommended action. Complexity goes into an appendix; decision-makers need clarity, not completeness.

The contrarian angle: models cannot see the dressing room

Here I must argue against myself, because that discipline is mandatory for anyone who works with data.

First, correlation is not causation. When I found that home teams score 0.38 more goals per match, that was a strong and verifiable correlation. But the conclusion "the crowd produces 0.38 goals" is a leap my data cannot support. Much of that gap could come from scheduling, from home teams more often facing weaker opponents at home, or from variables I never measured. Data shows that an effect exists; it does not automatically show why.

Second, transfer valuation models systematically overrate young potential and underrate a variable that never appears in a spreadsheet: dressing-room chemistry. A twenty-year-old with outstanding progression metrics, whose projected transfer value rises linearly in the model, can collapse simply because he cannot speak the language of his new dressing room. I have watched a club pay heavily for a young talent with every metric in order, then take eighteen months to discover that those metrics do not measure the ability to withstand pressure. When models have no variable for the human factor, they are not wrong in their data; they are wrong in assuming that everything that matters can be measured.

Third, referees and VAR. People expected technology to end controversy. What actually happened is that controversy moved from the pitch into the review room, and in many cases became harder to resolve. A semi-automated offside decision still depends on which frame is chosen as the reference moment. A penalty-area foul still depends on a threshold of "enough influence" that the laws cannot quantify. When I analyse referee data in top leagues, I see that metrics such as yellow cards per match have not fallen since VAR was introduced — attention has simply shifted to grey-area incidents. Technology does not erase ambiguity; it relocates ambiguity somewhere more closed.

These three self-critiques have not made me abandon data. They have made me write more slowly, and to include what I do not know. Every dataset is a scripture, and I am a slow reader — slow because I know most of my error lies in the rows I never entered into the table.

Signals for the next cycle

Based on my experience watching matches over the past six years, three signals deserve tracking in the coming major-tournament cycle.

The first is the value of proactive defensive teams. After Morocco in 2026, many smaller sides began building around low PPDA and tight line spacing instead of trying to control possession they lack the personnel to control. If the trend continues, we will see more and more matches where the side with 65 percent possession is the side with the lower xG.

The second is the return of the crowd variable after a period in which it was treated as negligible. After 2026, analysts learned that a crowd can be worth more than three-tenths of a goal per match — not a small figure in a sport where most matches are decided by one goal.

The third is the gap between data and decisions. Clubs have more data than ever but do not always make better decisions. I do not predict the future by intuition; I only read the traces the numbers leave behind, and the clearest trace today is the gap between the analytics department's spreadsheet and the head coach's meeting room.

The question I leave for the next round of fixtures does not point at a particular name. It points at a method: when a team is underrated by the media but rated highly by defensive data, which side will you believe — and do you have the patience to wait a full season for the answer?