One game in, the World Cup makes no sense — and the data says wait

Sports 4 min 80 sources

Twenty-four matches into a 48-team World Cup, the results defy the form book: favourites stumbling, minnows holding draws, a coach fired off a single loss. The numbers underneath say most of it is noise that will fade.

Key takeaways

Twenty-four games into the World Cup, several favourites dominated possession but failed to win — because they took low-quality, long-range shots, not because they collapsed.
Tunisia fired its coach after a 5–1 loss to a Sweden side that wildly overperformed its underlying numbers — a single noisy result mistaken for a verdict.
A pattern only becomes real when it survives a big sample: the Brewers and Guardians beating forecasts for eleven straight years is signal; one hot or cold game is almost always noise.

The 2026 World Cup has played its first round of group games — all 48 teams, 24 matches — and the early table looks like someone shuffled the form book. Germany beat Curaçao 7–1, as expected. But Spain drew 0–0 with Cape Verde, Portugal drew with DR Congo, and Saudi Arabia held Uruguay. One federation has already fired its coach. The temptation is to read each result as a verdict. The data underneath says most of it is noise — and that gap is the most useful thing in sport this week. [12][16][17]

The favourites who dominated and still didn’t win

Five of the tournament’s strongest sides — Portugal, Spain, Switzerland, Turkey and Uruguay — rank top five for possession and final-third passing accuracy. They had the ball exactly where a good team wants it. Most of them generated 25-plus shots. And, Portugal aside, none of them won their opener. [12]

The reason is a single number sports analysts now live by: expected goals, or xG. It estimates how likely each shot was to score, based on where it was taken and how. A team’s xG is the goal tally it “deserved” from the chances it created. Spain averaged 0.08 xG per shot against Cape Verde — worse than Burnley and Wolves managed in the Premier League this season, two teams nobody wants to copy. [12] They took plenty of shots. They just took bad ones, from distance, low-percentage looks that flatter a possession count but rarely go in.

So the headline — “Spain held by Cape Verde” — and the mechanism — “Spain took 0.08-xG shots all night” — point in different directions. The first reads like a collapse. The second reads like an off night that the next two games will probably wash out.

The coach who got fired over a coin that landed badly

Sweden beat Tunisia 5–1. On paper, a statement. Underneath, Sweden out-scored their expected goals by 3.67 — the biggest overperformance of any team in the round. [12] That number doesn’t mean Sweden are back; it means their finishing ran hot for ninety minutes, the kind of streak that, by its nature, does not last.

Tunisia, on the losing end of that same coin-flip of a scoreline, fired manager Sabri Lamouchi the next day. [12] One match, one bad bounce of variance, one job gone. It is the cleanest example this week of a recurring mistake: treating a single, noisy sample as if it revealed a permanent truth. The performance and the result diverged, and the federation acted on the result.

The draws nobody predicted, and the format that invites them

Nine draws came in the first round — the highest proportion of stalemates this phase of a World Cup has ever seen. [12] Some were genuinely even: Haiti and Scotland finished level on expected goals to two decimal places, 1.05 apiece, before Scotland nicked it 1–0. [12] Côte d’Ivoire over Ecuador and Ghana over Panama both arrived on very late winners that the underlying numbers say neither side had earned. [12]

Part of this is the new 48-team format — bigger, with less jeopardy in the group stage, so more teams play for a point. But part of it is the trap of a small sample. With one game each, a single deflection or late goal flips a country’s story from “crashing out” to “through.” There simply isn’t enough football yet to separate the good teams from the lucky results.

Where the pattern actually holds — and why it’s rare

The flip side is real. In American baseball, the Milwaukee Brewers and Cleveland Guardians have beaten their forecasts for eleven straight years. [1] The Brewers were projected to win 81 games a season on average and won 89; the Guardians were pegged for 84 and won 90. [1] That is not one hot fortnight — it is a decade of the same teams quietly outperforming, which is the signature of something real rather than noise.

How? Their young players keep producing faster than the projection systems expect. Milwaukee’s Jacob Misiorowski was modelled for 1.53 wins above replacement this year — a fair number for a pitcher with 66 big-league innings — and is on pace for 9.2. [1] Cleveland’s Brayan Rocchio was projected at barely above replacement after a demotion last season and is now a top-four shortstop in his league. [1] The difference between Sweden’s hot night and Milwaukee’s eleven years is sample size. One is a coin landing heads; the other is a weighted coin.

What we still don’t know

It is genuinely early. Twenty-four games is roughly one-third of a single past World Cup’s fixtures crammed into a few days, and the gap between performance and result will narrow as more matches pile up. [16] Whether the record run of draws is the bloated format or just first-round caution will only be clear by the knockout rounds. The teams that look broken now may simply have taken bad shots on a bad night — and the ones who look reborn may have caught a coin mid-air.

02 · Lesson · why it matters

The standout result is mostly a coin you watched land

Extreme outcomes tend to drift back toward the ordinary — so the hot streak, the shock loss, the breakout night are usually noise we mistake for a new truth.

A coach lost his job to a bounce

Tunisia lost 5–1 to Sweden and fired its manager the next day. Underneath that scoreline, Sweden scored 3.67 more goals than the quality of their chances said they should have — the biggest finishing fluke of the whole round. The result was lopsided. The performance was not. A federation looked at one game, saw a verdict, and acted on it.

This is the most expensive mistake in sport, and it has a name. It is regression to the mean: when an outcome depends partly on skill and partly on luck, an extreme result is usually the luck talking, and the next result drifts back toward normal. The hot hand cools. The collapse recovers. The shock fades. Not always — but far more often than our instincts believe.

Why your eyes lie to you here

Every result you see is two things added together: how good the team actually is, and how the day happened to break. A deflection, a post struck instead of a goal, a referee’s call, a keeper having the night of his life. On any single afternoon, the luck can swamp the skill entirely.

The trouble is that a scoreline hides the split. “Spain held 0–0 by Cape Verde” looks like a true measurement of Spain. But Spain took shot after shot from bad angles and long range — 0.08 expected goals per attempt, worse than the Premier League’s weakest sides. They didn’t forget how to play. They had a night where the chances were poor and nothing dropped. The number reveals what the scoreline conceals: most of that result was the day, not the team.

The extremes are where the luck piles up

Here is the part that feels wrong but matters most. The teams at the top and bottom of any one-week table are not just the best and worst — they are disproportionately the luckiest and unluckiest. To finish at an extreme, you usually need your skill and your luck pointing the same way.

That is why the standouts revert. Sweden’s finishing ran hot; it will cool. A team buried by a freak scoreline overperformed its bad luck; it will rise. This is selection: the very thing that put a result at the edge of the distribution — a run of fortune — is the thing least likely to repeat. Pick the most extreme sample and you have, without meaning to, picked the one most contaminated by chance.

The rare case where it’s real

So how do you ever know a streak is the real thing? Sample size. One game is a coin flip. Eleven years is a weighted coin.

The Milwaukee Brewers and Cleveland Guardians have beaten their win forecasts every season for over a decade — projected for 81 and 84 wins, delivering 89 and 90. That is not a hot fortnight that regression will erase. It is the same teams, the same edge, surviving hundreds of games and thousands of bounces. When something holds across a large enough sample, the luck has had time to cancel out, and what is left is skill. The lesson is not “nothing is ever real.” It is “one result is almost never enough to tell.”

You are standing in the same spot

It is easy to feel clever watching a federation fire a coach over one game, or a club hand a record contract to a player off a career year that was half luck. But the reflex is human and it is yours too. You judge a new colleague by their first week, a restaurant by one bad meal, a decision by how it happened to turn out rather than whether it was sound. You read the loudest signal as the truest one.

The honest move is slower and humbler. When a result is extreme, ask how much of it was the bounce — and assume the next one will be quieter. The standout you are watching is mostly a coin you happened to catch in the air. Most of what looks like a new truth is just the world returning, as it always does, to ordinary.

03 · Lab · your turn

Judge the Sample

Rehearse waiting for enough games before reacting, and feel an early hot or cold streak regress back to a manager's true level.

04 · Hope · carry this

The same patience that waits out a bad streak is what lets the real thing show itself. Give the world a few more games, and most of what looks like ruin or genius turns out to be ordinary — which means our worst days rarely define us either.