Reading football stats fairly: per-90, sample size and form

Reading football stats well is mostly about context: how many minutes a number comes from, how many matches, against whom and in what situation. Football is low-scoring and noisy, so a short run can make an ordinary player look brilliant or a good team look lost. This guide collects the checks that stop a number from misleading you, in the order worth applying them.

Why use per-90 numbers?

Per-90 statistics express a player’s output per ninety minutes played rather than per season or per appearance. That puts starters, substitutes and rotated players on one scale: a forward who plays half as much should not be judged on half the goals. Per-90 figures from very few minutes are unreliable, which leads straight to the next check.

How much data is enough?

Sample size is the amount of evidence behind a number. Because goals are rare, rates built from a handful of shots or matches swing wildly and mostly settle down as more games are played. Before trusting a figure, ask how many minutes, shots or matches it rests on, and be most sceptical of the most extreme numbers.

What is regression to the mean?

Regression to the mean is the tendency for extreme performances to drift back towards a player’s or team’s long-run level. A striker scoring from almost every shot, or a defence that keeps conceding from the few chances it allows, is usually riding luck that will fade. It is not a force pulling teams back; it is what happens when part of an extreme run was noise.

How much does recent form matter?

Form, usually a team’s last five or six results, carries real information, but less than instinct suggests. Results are noisy, and a run may have come against weak or strong opponents. Performances such as chances created and conceded are a steadier guide to form than the win-loss column.

What is game state?

Game state is the scoreline context a number is produced in: leading, level or trailing. Teams behave differently in each. A side chasing a game shoots more and concedes counter-attacks; a side protecting a lead sits deeper. A team that led early in many matches will show lower attacking numbers than its quality suggests.

Why does the fixture list change the picture?

Strength of schedule measures how hard a team’s fixtures have been, usually as the average rating of the opponents faced. Identical records built against very different opposition mean different things. Early in a season especially, a table partly reflects who has played whom.

How big is home advantage?

Home advantage is the edge teams gain at their own ground, from crowd, travel and familiarity. It is real but varies by league and season, and it shrank noticeably when matches were played behind closed doors. Models estimate it from the data rather than assuming a fixed amount, and so should anyone reading football stats about home and away splits.

Putting it together when reading football stats

Check the minutes, the sample, the opponents and the scoreline context before the headline number. The guide to expected goals metrics covers the chance-quality numbers these checks apply to, and the accuracy page shows how TACTIX reports its own record with the sample beside every figure. The methodology explains how the model handles form and home advantage.

18+. Information and analysis only, not betting advice. If gambling stops being fun, free and confidential help is at BeGambleAware.org.