What expected goals (xG) actually tells you
xG is a measure of chance quality, not of who deserved to win. What goes into it, what it is good for, and the four ways it gets misused every weekend.
Expected goals is the one advanced statistic that has escaped into everyday football conversation — and it is used wrongly about as often as it is used well.
What it is
Every shot is given a probability of being scored, between 0 and 1, based on what has happened to many thousands of similar shots in the past. A penalty is worth a lot; a header from the edge of the area through a crowd is worth very little. Add up a team's shots and you have its expected goals for the match: the number of goals an average team would be expected to score from those chances.
The ingredients are the things you would guess: distance from goal, angle, whether it was a header or a kick, and the type of pass that created it. Better models also know where the defenders and the goalkeeper were. Different providers build different models, so two sources will give two different xG figures for the same match. That is not an error. It is a reminder that xG is an estimate, not a measurement.
Expected assists (xA) applies the same idea to the pass before the shot: how likely was that pass to become a goal, regardless of whether the striker finished it?
What it is good for
Separating performance from results over a run of matches. Goals are rare and lumpy; a team can win three in a row on a deflection, a penalty and a goalkeeping error. Chances are created far more often than goals are scored, so xG stabilises sooner. A team that keeps creating good chances and keeps losing is usually about to improve; a team winning on thin chances is usually about to stop.
Judging finishing — carefully. A striker who scores far more than his xG over one season has usually been fortunate as well as good. Over several seasons, the few who keep doing it really are exceptional finishers.
Judging chance creation and chance prevention, which is most of what a coach is trying to control.
The four common mistakes
1. Treating a single match's xG as the "real" score. One match is a handful of shots. A team can lose the xG count 0.6 to 2.1 and deserve its win completely, because it scored early and then chose to defend. Game state shapes xG: a team that is behind shoots more, and from worse positions.
2. Ignoring what is not a shot. The cross that flashes across the six-yard box untouched, the through ball a striker just fails to reach, the attack broken up by a last-ditch tackle — none of them register. Territory and danger that never produce a shot are invisible to xG.
3. Adding up tiny chances. Ten shots worth 0.05 each are not the same as one chance worth 0.5, even though both sum to half a goal.
4. Comparing figures from different models. Always compare like with like.
Where you will see it on 4Futbol
We only show xG where a source publishes it and we may use that source. At the moment that means the Premier League: player pages and leaders carry the expected goals and assists from the Fantasy Premier League feed, and match pages for the big European leagues show match xG where our results file includes it. For Liga MX, Argentina and Brazil, no free source publishes xG, so you will not find an invented one here.
xG will not tell you who was the better team. It will tell you who had the better chances — which, over a season, is usually the same thing.