Goals vs xG: overperformance and regression
A player or team is overperforming their xG when they score more actual goals than the quality of their shots suggests, and underperforming when they score fewer. A big gap over a handful of matches usually shrinks over a longer run, but it doesn't always vanish completely.
Team FootyMetrics
Updated Jul 2026 ยท 6 min read
- Overperforming xG means scoring more goals than your shots' combined quality would predict. Underperforming means scoring fewer.
- A large gap over a small number of matches is more likely to be variance, chance events that even out over time, than a sustained pattern.
- Some of the gap can be real, sustained finishing skill rather than luck, especially for elite players over a full season or career. It isn't purely random.
- Read a big early-season gap as a reasonable signal that results might cool off, or pick up, not a certainty. How much of any gap is skill versus luck is genuinely debated among analysts.
Our full expected goals (xG) explainer covers how the number itself gets built. This page is about what happens once you compare it to what actually went in the net, and how much of that gap you should trust.
What overperforming or underperforming xG means
Every shot carries an xG value between 0 and 1. Add up the xG of every shot a player or team has taken and you get a total. Compare that total to the goals they actually scored, and the gap between the two is what people mean by overperformance or underperformance.
A striker who scores 15 goals from shots worth 10 xG combined is overperforming, scoring more than their shot quality suggests. A team that scores 6 goals from 12 xG worth of chances is underperforming, scoring less than their shots suggest they should. Both are a comparison between two numbers, not a verdict on the team or player.
Why a big gap over a small sample is more likely to be variance
Football has a small number of shots and goals in any single match, and a small number of matches in any short run of form. A gap between goals and xG over five or ten games can come almost entirely from chance rather than anything sustained. Our variance and sample size explainer covers the general reasoning: short samples are noisy, and a real signal takes longer to separate from randomness than most people expect.
Applied to goals versus xG specifically, this has been studied directly. A 2025 paper by Felix Holzmeister and Magnus Johannesson, published in the Journal of Sports Economics, decomposed deviations from expected performance across Europe's top seven leagues over three seasons. Their estimate is that roughly 40% of a team's over or underperformance across a season is attributable to skill, with the rest coming down to luck. Over a run of matches shorter than a full season, the luck share is almost certainly higher still, because there's been less time for randomness to average out.
The more honest version: some of the gap can be real skill
It would be an overstatement to say every xG gap is just noise waiting to disappear. Some players sustain a real edge over their xG for years, and it shows up across more than one dataset.
Lionel Messi is the clearest example. Since the 2015-16 season, various xG models have him scoring around 33 more league goals than his own expected goals total suggests, a gap that has held up across multiple seasons rather than one hot run. In 2012-13 alone, his ratio of goals to xG reached a level analysts have described as happening roughly once in 1.6 million individual player seasons if it were down to chance alone. That isn't the pattern you'd expect from luck, it points to something about his finishing, shot selection or disguise that a generic xG model doesn't fully capture.
Robert Lewandowski makes a different point, because his numbers move in both directions. He underperformed his xG by more than 6 goals across the 2014-15 to 2019-20 seasons, including one season 11 goals below expectation, then overperformed by a similar margin over the following two seasons. Even a player capable of sustained overperformance can run cold for a stretch before it shows up again.
Not a two-bucket answer
How to read this in practice
Treat a big xG gap as a signal, not a rule.
- A team scoring well above their combined xG early in a season is a reasonable prompt to ask whether that continues, not a guarantee their results will fall away. The same logic runs in reverse: a team stuck below their xG is a reasonable prompt to expect an eventual upswing, not a promise of one.
- The size and duration of the gap both matter. A small gap over a handful of games tells you very little. A large, sustained gap held across a full season or several seasons, especially for one individual player with a track record of doing it before, carries more weight.
- Context sits underneath the number too. A run of tougher or easier opponents, a change in shot selection, or a goalkeeper having an exceptional run can all explain part of a gap without it being either pure skill or pure luck.
A genuinely unsettled debate
How much of any given xG gap is skill and how much is luck isn't settled science. Analysts disagree, and different studies land on different splits depending on the league, the sample and the xG model used. Treat any specific percentage, including the one cited above, as an estimate from one study rather than a fixed law. The safest read is the general shape of it: most short-sample overperformance shrinks, some players are a genuine exception, and you rarely know for certain which one you're watching in real time.
For the definitions this page builds on, see the full expected goals (xG) explainer and what non-penalty xG (npxG) is. FootyMetrics doesn't calculate or display xG itself. The closest read on our own site is shots on target and big chances created or missed, alongside actual goals, for every player and team across 115+ leagues.
Goals vs xG overperformance FAQs
Is a team overperforming xG guaranteed to get worse results?
No. It's a signal that their results might not hold at the same level, not a guarantee. Some of the gap between goals and xG can be genuine finishing quality rather than luck, and even a big early-season gap can persist for a while.
Does xG overperformance always regress to the mean?
Not always, though a lot of it does. One academic estimate puts roughly 40% of a team's over or underperformance across a season down to skill and the rest down to luck, so most of a gap tends to shrink over time, but not all of it, especially for a small number of individually excellent finishers.
Can a player sustain scoring more goals than their xG for years?
Yes, for some players. Lionel Messi has outscored his own xG total by a wide margin across multiple seasons, a pattern that has held up long enough that analysts treat it as genuine finishing quality rather than a lucky run.
How big does a sample need to be before an xG gap means something?
There's no fixed number of matches. A gap over a handful of games is much more likely to be variance than a gap held across a full season, and a gap an individual player has repeated across several seasons carries more weight again.