Football Analytics in [2026]: What the Numbers Actually Tell Us

Football analytics has moved from a niche interest to a mainstream part of how the game is discussed, with expected goals (xG), progressive passing, and pressing metrics appearing in mainstream broadcast and journalism. Here is the honest guide to what these metrics actually mean, how to use them, and where they hit their limits.

Expected Goals (xG): What It Actually Measures

Expected goals (xG) is the metric that has most changed football analysis. An xG model assigns a probability (between 0 and 1) to each shot attempt based on factors that historically determine whether similar shots are scored: shot location, shot type (foot, header), assist type (cross, through ball, set piece), and the positioning of defenders and goalkeeper. A shot from the center of the penalty area at close range might have an xG of 0.6 (a type of chance that's been scored 60% of the time historically); a shot from 30 yards on the right flank might have an xG of 0.02.

What xG is useful for: evaluating teams and players over large sample sizes, identifying whether a team is creating high-quality chances or relying on long-shots, and comparing team performance across seasons while controlling for finishing variance. A team that significantly outperforms its xG over a short period is likely experiencing positive finishing variance that regresses toward its underlying chance quality over time; a team that significantly underperforms its xG is experiencing negative variance. Over 38 matches, xG becomes a fairly reliable predictor of where teams should finish in the table.

What Analytics Doesn't Capture

The honest limitations of football analytics: the data captures what happens (shot location, pass completion rate, distance covered) but imperfectly captures why, and the "why" is often where the interesting football is. Off-ball movement that creates space for teammates doesn't appear in basic statistics. The quality of pressing shape (coordinated defensive pressure) is partially captured by pressing intensity metrics but the coordination element is harder to quantify. Player interactions — the combinations that make two specific players more effective together than apart — aren't captured in individual player statistics.

The specific error that football analytics sometimes produces: treating high xG performance as reliably predictive of future success without accounting for the tactical adjustments opponents make when they've identified how a team creates its chances. A team known for wide crossing creates xG from those crosses; as opponents specifically defend that pattern, the same quality of crossing produces lower xG because defenders are better positioned.

My honest take: xG is most useful over large samples for team and player evaluation — not for explaining individual games. What analytics doesn't capture (off-ball movement, coordination, player combinations) is often where the most interesting football insight lives. Use the data to identify patterns, not to replace watching and understanding the game.