Goals minus Expected Goals
Goals minus expected goals is a player's actual goals less their expected goals, so a positive figure means they are finishing above the quality of their chances.
Why it matters. Sustained overperformance tends to regress, while a large negative figure often flags a player about to score more rather than less.
Actual minus expected, and nothing else
Take a player's goals and subtract their expected goals. What is left over is not covered by chance quality. Not the run into the box, not the pass that found them, just what happened once the ball left their foot or head.
- Positive. Scoring more than the shots deserved. Clean striking, favourable deflections, or a run of good luck against goalkeepers.
- Negative. Scoring less than the shots deserved. Poor placement, woodwork, or a run of bad luck against goalkeepers.
- Near zero. Output has matched chance quality, which is where most players sit across a full season.
Because xG already prices in distance, angle, pressure and shot type, the delta is not measuring whether a player gets good chances. It is measuring what they did with them.
This is the opposite convention to the deltas on a league table, where xPts and xGD run expected minus actual. There, positive means a team is underperforming and due an improvement. Here, positive means a player is overperforming. Keep the two the right way round, they answer opposite questions.
Three deltas, three verdicts
A forward on +6 after ten matches. Five goals from chances worth 2.1 xG is a hot month. It is not proof of a finishing gift. Verdict: expect the next ten matches to look far more ordinary, because sustaining that gap over hundreds of shots almost never happens.
A striker on -5 across a full season. Reaching good positions repeatedly and converting fewer of them than the model expects. Verdict: treat this as bad luck rather than a weakness, unless the shot placement data backs up a genuine technical issue. Compare against post-shot xG delta to check whether the shots themselves were well struck.
A poacher sitting close to zero for two straight seasons. Consistency across a large sample is closer to a real signal than any single-season spike. Verdict: this is closer to their true finishing level, whatever that level turns out to be.
Why a small sample lies to you
Five shots is not a sample, it is an anecdote. A single well-struck free kick from outside the box can swing a low-minutes player's delta by a full goal on its own.
- Under 10 matches: treat the number as descriptive only. It says what happened, not what to expect next.
- 20 to 30 matches: starting to mean something, though still capable of large swings from one or two events.
- A full season of regular starts: the delta is doing its real job, separating a settled level of finishing from the noise around it.
The more shots in the sample, the harder it becomes to keep beating or missing the model by a wide margin. That convergence is the whole mechanism behind treating a big gap as temporary.
Two readings that miss the point
"A positive delta proves this player is a clinical finisher." It proves they finished well across this particular sample of shots. Genuine finishing skill exists but it is small relative to shot quality, and it takes a career's worth of shots to isolate from variance. Treat one good season as evidence, not proof.
"The goals column already tells the story, why check the delta." Goals confirm what happened. The delta explains whether it is likely to keep happening. A player scoring at their expected rate is a stable source of goals. A player scoring well above it is due a quieter run, whatever the finishing highlight reel suggests.
Common questions
- Which way round is the subtraction?
- Actual minus expected, so positive means running hot. Note that the standings deltas run the other way, expected minus actual, so a positive number there tells the opposite kind of story.
- Does a negative figure mean a player finishes badly?
- Rarely, over a normal season. Finishing skill takes hundreds of shots to establish, so a negative figure across one campaign is far more often variance than a settled weakness.
- How many shots does it take before the number means something?
- Treat anything under about 40 to 50 shots as noisy. A season's worth of shots for a regular starter is usually enough to separate a real edge from a hot run, though even then the gap should be shrinking towards zero, not sitting flat.
- Should a big positive figure change how a player is rated?
- Not on its own. It says the shots landed better than the model expected across this sample, not that the underlying skill has changed. Rate the player on the chances they get. Rate the delta as a temporary bonus that is likely to fade.
- Is a penalty specialist's delta biased upwards?
- Slightly, because a penalty converts far more often than the open play average, and the taker gets full credit for it. Non-penalty xG strips penalties out entirely, so compare the two figures if a player takes most of the team's spot kicks.
- Can the delta be negative for a whole squad?
- Yes. A team collectively missing more good chances than the model expects shows up as a negative delta summed across the squad, and it is one of the clearer early signs that results are about to improve without any change in performance.
Related terms
See xG Δ across the Premier League
Every player, team and match measured on the chances created rather than the scoreline.
Last updated