What is xG? What expected goals tell you about a match, and what they don't
2.4 xG against 0.7 and the score is 0-1. How expected goals are calculated, why one match can mislead, and how to read the number on a simulation screen.
Full time, 0-1. Below the score there is another line: xG 2.4 – 0.7. A supporter of the losing side points at it and says ‘we should have won’. The winning side has an answer ready: ‘look at the scoreboard’. Both are right up to a point. Separating what xG says from what it does not makes that argument more useful.
The chance that one shot becomes a goal
xG stands for expected goals. Every shot is given a value between 0 and 1. A shot worth 0.10 means that roughly one in ten shots from similar situations is scored. Add up the values of all of a team's shots and you have its xG for the match.
The value comes from the situation, not from the name of the player shooting. Models look at a very large number of past shots and ask: from this spot, at this angle, after this kind of pass, how many went in?
What goes into the number
Details differ between models, but the main inputs are similar:
- Distance to goal. The strongest factor; a shot from the six-yard box and one from outside the area are worlds apart.
- Angle. Facing the goal square on is not the same as shooting from near the byline.
- Body part. Headers get a lower value than shots with the foot from the same spot.
- Type of chance. A through ball that leaves a forward one against one is rated differently from a shot after a cross into a crowded box.
- Set piece or open play. A penalty is its own category, worth about 0.76 in most models.
Why xG misleads over one match
xG is a probability, and probabilities drift easily over a small number of attempts. A team producing 2.4 xG and scoring nothing is surprising but not impossible; play the same match a hundred times and it happens in some of them. A goalkeeper's good day, a shot off the post or one poor finish is enough to decide a ninety-minute sample.
That is why xG is better for spotting a trend than for passing judgement on a single game. A team that clearly out-creates its opponents over ten matches is probably doing the right things, even if the table says otherwise. The reverse holds too: a side winning repeatedly from very few chances will find that run hard to sustain.
What xG does not see
The number measures the quality of the shot, not everything before it. A dangerous attack that ends without a shot adds nothing. A forward who is through on goal and takes a heavy touch scores zero in the model, even though the defence had been opened completely.
Game state is easy to miss as well. A team leading 1-0 often drops off and allows shots; most are low-value efforts from in front of a packed defence, but the total grows. Before reading the final xG as ‘we were battered in the second half’, check where those shots came from.
Finally, simple models do not know who is shooting. A good finisher scores slightly more than his xG in the long run, though the difference is usually smaller than people assume.
Reading xG on the simulation screen
In the DreamerBuild match simulator, xG builds up shot by shot as the game runs. It is most useful read alongside the score:
- 1Score and xG agree: the match went as expected; the stronger side created chances and took them.
- 2High xG, low score: the team created but did not finish. Simulate the same fixture again and the result is likely to change.
- 3Low xG, high score: many goals from few chances. That shows things went well in this match, not that the squad is superior.
- 4Both sides low: a closed game. Change the formation or the starting XI and see whether it makes a difference in attack.
The momentum tab shows when the xG was accumulated. If a team found all its chances in the last twenty minutes, that is usually a consequence of the score and the substitutions.
| Situation | Score | xG | How to read it |
|---|---|---|---|
| Deserved win | 2-0 | 2.1 – 0.6 | Chances created and taken |
| Unlucky defeat | 0-1 | 2.4 – 0.7 | Creation without finishing, a replay may end differently |
| Efficient but risky | 2-1 | 0.8 – 1.9 | Many goals from little, hard to sustain |
| Closed game | 0-0 | 0.5 – 0.4 | Neither side found a way through |
What it means for building a lineup
You raise xG by shooting from better places, not by shooting more. When you build a lineup, ask who actually arrives in the penalty area. A midfielder running alongside a lone striker, or a winger arriving at the far post, produces more expected goals than three players shooting from distance.
Simulating the same squad in two formations several times and comparing the average xG tells you more than one final score. One match is luck; the average of five starts to be an idea.
Starting shapes to try on the pitch
Open this setup, then change the players and make it your own.
Open this setup, then change the players and make it your own.
Look at the xG of your own match
Pick two teams, play the match minute by minute and see how closely the score follows the expected goals.
More articles
We simulated Beşiktaş v Galatasaray 1,000 times: who has the edge at Tüpraş?
Ahead of the 30 November derby we played Beşiktaş v Galatasaray a thousand times at each of three venues: win rates, common scores, top scorers and home advantage.
Real Madrid–Barcelona: a 3,000-match experiment with open records
Two fixed lineups, three venues, 3,000 simulations. Explore the results, exact squads, uncertainty and downloadable match records from our Clásico experiment.
How much is home advantage worth? We played seven fixtures 21,000 times
The same two squads at three venues. Seven big fixtures simulated a thousand times each at home, on neutral ground and away: what home advantage is worth, and when it does nothing.