Every club says it wants players who fit. Almost none can tell you whether fit is real or just the thing you say after a signing works out. So here is a blunt question with, for once, a number attached. At the 2022 World Cup, did fit beat talent?
A tracking-data analysis of the 2022 World Cup built a measure of team chemistry from the off-ball movement of all twenty-two players and checked it against where each team actually finished. It predicted the final standings better than a talent rating did. Chemistry came in at a rank correlation of 0.704. The talent baseline, the players’ FIFA Overall ratings, managed 0.548.
We did not run this study, and we are going to spend most of this piece poking at it rather than selling it. But the headline survives the poking, and the honest version is more useful than the slogan.
The ratings are better than they sound
Before we credit chemistry with anything, the talent baseline has to be worth beating. It is a video-game number, so the fair question is whether it measures anything real.
It measures plenty. EA compiles those ratings from a network of roughly nine thousand reviewers, a mix of season-ticket holders, scouts and coaches who watch the players, blended with performance data to catch tendencies the eye misses. Thirty-five attributes feed each player’s Overall, weighted by position, so a defender is judged on tackling and heading rather than finishing. It exists because no single data vendor covers every league and every player on earth, which is exactly what makes it a usable global talent index.
And it works where it counts. Take the last five World Cups and look only at the champions. Every one of them came from the FIFA top five. Not the top fifty, the top five. Before a ball is kicked, the ratings reliably tell you the trophy will go to one of about five teams, and for five tournaments running they have been right. That is a serious thing for a number built partly from season-ticket holders to keep getting right.
It is not flawless. The known knock is that individual ratings get inflated by the fame and size of a player’s club. But this runs on squad ratings, twenty-three players pulled from clubs and leagues all over the world, so that club-size noise largely averages out across a national team. As a read on raw talent, it is sound, which is the point: it is a baseline worth taking seriously.
Where the rating runs out
Here is the catch, and it is the whole story. The rating tells you the winner will be one of five teams. It cannot tell you which one. Line the champions up against the most talented squad in each of those tournaments and the ceiling is obvious.
In 2006 Brazil were the best squad in the game by a distance and went out in the quarter-finals; Italy won. In 2014 Spain were top-rated and did not survive the group stage; Germany won. In 2018 Brazil, Spain and Germany led the ratings; France won, and Germany went home early. In 2022 France were rated highest, reached the final, and lost to Argentina. Only in 2010, when Spain were both the best squad and the champions, did the most talented team actually lift the trophy.
So the rating brackets the contenders and then stops. The single most talented squad has won once in five attempts. Everyone left standing in the second week is separated by a rating point or two, and something the rating cannot see decides which of them wins. It is the same uncomfortable shape we found counting Belgian transfers: past a certain point, paying for more talent barely moves the odds.
That something, the 2022 tracking data suggests, is how well the eleven actually play together.
What “chemistry” actually means here
Not vibes, and not goal celebrations. The study reads chemistry from tracking data as off-ball coordination: how the model relies on pairs of players moving in concert across the pitch, whether or not the ball is anywhere near them. A striker’s run that drags two defenders away and opens a lane he never receives in is the kind of thing it counts and a box score never will.
The number that matters is what they did to it before reporting it. Every figure is computed on residuals, after stripping out talent, experience, games played and the strength of the opponent. It is not measuring “are good teams good.” It is measuring “did this team play better together than its talent said it should.” That residualisation is the whole reason the result is worth anything, and it is the discipline we hold our own analysis to.
The honest reading
Now the caveats, because they are the point rather than the fine print. This is one tournament, correlational, with forty-four of sixty-four matches tracked. It is suggestive, not settled.
And there is a trap inside it the authors are careful about. If you measure chemistry from on-the-ball events alone, it looks beautifully predictive, a correlation around 0.67. Then you control for how many minutes the players had actually spent on the pitch together, and most of that signal evaporates, down to 0.27. Players who play more together combine more, which is obvious and not chemistry.
What does not evaporate is the off-ball part, the coordinated movement away from the ball, which is precisely the part event data cannot see. The lesson cuts in a direction we have to be honest about: the durable signal lives in tracking data, not in the event feeds most analysis is built on. A read on fit from ordinary data captures some of this. It does not capture all of it, and pretending otherwise is how you end up selling a number that does not hold up.
There is no single fit score
The other thing the study kills is the dream of one tidy chemistry number you can rank the world by. The four teams that ran deepest in 2022 did it in four different shapes. Argentina won with chemistry concentrated almost entirely through Messi, a nucleus, and by the squad-wide density measure they ranked only seventh. Morocco were a wall, a compact defensive block that conceded less than far more talented sides. France were a network with several attacking hubs. Croatia were an engine built around one midfield triangle.
And two of them had no business being there on the ratings at all. Croatia reached the 2018 final with a squad rated 80, and Morocco the 2022 semi-finals rated 77, both a clear tier below the 84-and-85-rated favourites they outlasted. These were not elite squads overperforming at the margin. They were sides the talent index had written off, carried within a goal of the final by how they played rather than by who they had. It is the winners’ table read from the other end: talent sets the baseline, and how well a team plays together decides how far above or below that baseline it finishes. Morocco’s wall and Croatia’s engine are simply the two most extreme readings of the same effect that nudged Argentina past France.
Four winners’ brackets, four structures, no common formula. Which is the part most relevant to anyone trying to recruit for fit: there is no universal fit score, because the fit a team needs depends on the team. A side built around one irreplaceable creator wants something completely different from a side built on a back line that never breaks. Sell a single number and you are selling Argentina’s answer to Morocco’s question.
Will we see it again this year?
The 2026 World Cup is being played as this goes out, which makes the thesis testable in public over the next few weeks, so we will put a marker down rather than wait for hindsight. If chemistry really does the work talent cannot, the teams still standing in the final week should not simply be the most expensive squads on the board. And going by five straight tournaments, the single highest-rated squad will probably, once again, not be the one lifting the trophy.
We might be wrong. That is rather the point of saying it out loud beforehand. But it is also the question we spend our days on away from the tournament, because a club choosing a signing is asking a quieter version of the same thing the bracket is about to answer: not who is the most talented player available, but who makes the team you already have play better than the sum of its ratings.
Source: the 2022 World Cup chemistry figures are from the tracking-transformer analysis, which sets out its full method and limitations. Squad ratings are FIFA per-edition overall ratings via FIFA Index.