What Is xG? Expected Goals Explained for Football Fans

Every weekend the same argument plays out on social media: one team had twenty shots, the other had four, and the side with four won. Possession stats and shot counts used to be the whole debate. Then expected goals — xG — gave the argument a sharper language, and it has quietly become the single most cited number in football analysis.

Expected goals measure the quality of a chance rather than its existence. Instead of counting shots, xG assigns each attempt a probability of becoming a goal, based on what historically happens from that situation. A penalty, which is scored most of the time, carries a high value; a hopeful lob from forty metres carries almost nothing. Add up a team's attempts and you get a number that describes how many goals its chances were worth.

Where the number comes from

An xG model is trained on hundreds of thousands of historical shots with known outcomes. For each new attempt, the model looks at similar past shots and asks how often they went in. The ingredients are mostly geometric: the distance to goal, the angle from which the shot is taken, whether it was a header or a foot shot, and what kind of pass or situation created it — a through-ball, a cross, a rebound, a set piece.

That is why the tap-in after a cutback is worth far more than the spectacular volley from the corner of the box, even though the volley looks better on television. The model does not care about aesthetics. It cares that cutbacks from close range have been converted at a high rate for as long as anyone has recorded them.

What feeds a typical model

• Distance from the ball to the centre of the goal.
• The angle the shooter has on the goalmouth.
• Body part used — foot, head or other.
• The type of assist, such as a through-ball or a cross.
• Whether the chance came in open play or from a set piece.
• How many defenders stood between the shooter and goal.

What xG tells you that the score cannot

The scoreboard records what happened; xG records what the chances were worth. Over a single match the two can disagree wildly — a striker can miss two sitters and his team loses a game it dominated. Over a season, though, the gap between goals and expected goals is one of the most useful signals in the sport. A team consistently scoring far more than its xG is usually riding finishing form or a hot goalkeeper at the other end, and the smart money says it cools off. A side conceding far less than its xG against is living on its keeper's streak.

This is how analysts spotted that several surprise league leaders were mirages months before the table caught up, and how recruitment departments find undervalued strikers: a forward whose xG is high but whose goals are low is often one good run of finishing away from looking world-class.

Where xG misleads

The metric has honest limits, and quoting it without them is how arguments start. Different providers build different models, so the same match produces different xG totals depending on who counted. The models also struggle with context no camera angle captures easily: a bobbling pitch, an injured defender lunging late, a shooter off balance because he was clipped a stride earlier.

Most importantly, xG describes chances, not performance in the round. A team can lose the xG battle while executing a perfect plan — Simeone's Atletico have made a career of conceding low-value shots and winning anyway. The number is a lens, not a verdict.

xG at the individual level

The metric works just as well on players as on teams, and this is where fans meet it most often. A striker's goals minus his xG over a season is a rough finishing index: consistently positive seasons suggest a genuinely elite finisher — the Son Heung-min profile, year after year of outscoring his chances — while a single spectacular season far above xG more often regresses. Scouts use the same logic in reverse, buying forwards whose xG stays strong even when their goal tally dips.

Post-shot xG adds a second layer by evaluating the attempt after it is struck, judging placement and power rather than just the chance. It is the metric that finally gives goalkeepers their due: a keeper conceding far fewer goals than post-shot xG predicts is measurably saving shots others would not, rather than merely standing behind a good defence.

The fairest way to treat expected goals is as the beginning of a question rather than the end of one. When the scoreboard and the xG disagree, something interesting is happening — and finding out what is the entire fun of watching football with a second screen.