How to Read Football Statistics Without Being Fooled

Football statistics went from niche to unavoidable in about a decade. They are now quoted in broadcasts, arguments and team selections, frequently by people who have not thought carefully about what they measure.

Used well, they correct genuine biases in how humans watch football. Used badly, they add false precision to a bad argument. Here is how to tell the difference.

Expected goals, properly understood

Expected goals assigns each shot a probability of scoring based on characteristics like distance, angle and situation. Summed across a match, it estimates how many goals a team would typically score from those chances.

What it is good for: judging performance over many matches. Over a season, a team scoring far below its expected goals is probably finishing poorly or unluckily, and will likely regress toward the model. That is genuinely useful for evaluating whether a run of results is sustainable.

What it is bad for: single matches. In one game the sample is a handful of shots, and a low-probability shot going in is not a failure of the model — it is exactly what a probability means. Quoting expected goals to argue that a team deserved to win a specific match is the most common misuse of the statistic.

It also cannot see everything. Defensive pressure at the moment of the shot, the goalkeeper's position, and whether a chance was created against an organised defence or a broken one are only partly captured.

Possession tells you almost nothing alone

Possession percentage is the most quoted and least informative common statistic.

A team can dominate possession by passing sideways in its own half against an opponent perfectly content to let it. Another can have thirty percent and create more, because its possession occurs in dangerous areas after winning the ball high.

Possession is only meaningful in combination with where it happened and what it produced. On its own it describes style, not quality.

Pass completion punishes ambition

A player with ninety-five percent pass completion may be excellent, or may be avoiding every difficult pass.

The passes that break defensive lines are the ones most likely to fail. A midfielder attempting them will have a worse completion rate than one recycling possession sideways, and will usually be more valuable.

Look instead at where passes are attempted and what they achieve. Completion rate alone rewards caution.

The sample size problem

This is the error underneath most bad statistical arguments in football.

A striker with a run of goals in six matches has told you very little. A forward converting an unusually high share of chances over ten games is probably going to regress. Season-long numbers mean something; six-match numbers rarely do.

This applies to teams too. Early-season tables are dominated by fixture difficulty and randomness, and the confident conclusions drawn in September are frequently embarrassing by December.

What statistics are genuinely best at

Two things, both underused.

Correcting memory. Human recall of a match is dominated by goals and by the last ten minutes. Data does not have that bias, and it regularly shows that the team you remember as dominant was not.

Finding players nobody watched. Filtering thousands of players across dozens of leagues for a specific profile is something no scouting department could do manually. This is where data has changed football most, and it happens entirely out of public view.

The reasonable position

Statistics are a corrective, not a replacement. They are best used to check an impression formed by watching, and to ask why a number disagrees with your eyes.

When the numbers and the watching disagree, the interesting work is explaining the gap — not declaring one of them correct.

KickD Sports Desk

Our editorial team covers Arab football and the FIFA World Cup 2026 live on kickd.net — real-time scores, group standings, and match analysis updated around the clock.

⚡ Follow the action live

Real-time scores for every World Cup 2026 match on KickD