“Expected goals” — usually shortened to xG — has gone from an obscure data-nerd metric a decade ago to the dominant lens through which professional clubs, broadcasters, and the most informed bettors evaluate football matches. If you have spent any time watching post-match analysis on TV, you have seen the bar charts and heard the phrase. But what does xG actually measure, where does it work, and where does it mislead? This article is the practical, betting-focused explanation.
The Core Idea
Every shot in football has a probability of becoming a goal that depends on a small set of physical factors: how far from goal, the angle, the body part used, the type of pass that created it, whether the shooter was under pressure, and so on. xG models look at hundreds of thousands of historical shots and ask, “Out of all shots taken from this exact position and context, what fraction have ended up as goals?”
The output for any single shot is a number between 0 and 1. A penalty has an xG of about 0.78 (penalties convert at roughly that rate over time). A long-range volley from 35 yards out has an xG of maybe 0.02. A close-range header from a corner with a defender on the shooter has perhaps 0.10. Sum the xG of every shot a team takes in a match and you have their expected goals total — what an “average finisher” would have scored from those exact opportunities.
Why Bookmakers and Betting Models Love It
The traditional way to assess team strength was win-loss records. The trouble is that football is a low-scoring sport with significant variance, and a few weeks of unusual finishing — either side of the average — can completely warp a team’s record without their underlying performance changing at all.
Consider two teams. Team A has won 5 of their last 10 games but generated 17 xG and conceded 12. Team B has also won 5 of their last 10 but generated 11 xG and conceded 18. From the table, they look identical. From the underlying performance, Team A is much stronger and Team B has been propped up by an unsustainable run of clinical finishing and below-average goalkeeping luck. Over the next 10 games, Team A will probably outperform Team B regardless of what the table currently says.
This is the heart of why xG is so useful for betting. It cuts through finishing variance and tells you what was actually happening on the pitch.
How to Read xG Numbers
Per-Match xG
The first thing to look at is xG and xGA (expected goals against) per 90 minutes. These are usually averaged over a rolling window — last 10 matches, last 20 matches, or season to date.
- 1.8 xG per 90 / 1.0 xGA per 90 — an elite attacking, solidly defensive side. Champions League contender level in most leagues.
- 1.4 xG per 90 / 1.4 xGA per 90 — typical mid-table profile.
- 1.0 xG per 90 / 1.7 xGA per 90 — relegation profile, struggling at both ends.
Cumulative xG vs Actual Goals
One of the most useful diagnostics is the gap between a team’s cumulative xG over a season and their cumulative actual goals scored. A team that has scored 30 goals from 22 xG is enjoying an unsustainable hot finishing streak — they will almost certainly cool off. A team that has scored 18 goals from 26 xG has been unlucky in front of goal and will probably score more freely going forward.
xG Difference (xGD)
xGD is xG minus xGA. Over a 30-game window, it is a remarkably stable predictor of which teams are genuinely strong and which are getting fortunate results. Pre-season power rankings published by serious analytical outlets are heavily based on the previous season’s xGD per 90.
Where xG Works Best
xG is most reliable in domestic league matches where teams play 30+ games per season and the sample size of shots is meaningful. The metric was designed and validated against this kind of dataset and it works extremely well there.
It is also strongest for assessing the quality of open-play chances, which depend on geometry and pressure in fairly predictable ways.
Where xG Has Real Limitations
Set Pieces
Most public xG models treat all corners or free kicks similarly, but a team with an outstanding aerial threat (think Sean Dyche-era Burnley, or any Diego Simeone team) systematically outperforms its set-piece xG over long periods. If you ignore this, you will under-rate teams that win matches through patterns the model cannot fully capture.
Cup Competitions and Knockout Football
One-off cup matches, where one team rests starters or both teams play with a different game plan than they would in the league, are noisier. xG remains useful but should be combined with line-up analysis.
Goalkeeping Quality
xG measures the quality of the chance, not the quality of the goalkeeper facing it. A side with an elite goalkeeper systematically concedes fewer goals than their xGA suggests. Post-shot xG (psxG) — which factors in shot placement and goalkeeper positioning — is a more advanced version that helps here.
Stylistic Mismatches
Two teams with identical xG profiles can produce wildly different match outcomes depending on how their styles interact. A possession side that struggles against deep blocks may underperform xG specifically against teams that play that way, and a model that does not account for the match-up will mislead you on those nights.
How to Use xG as a Bettor
- Trust the long term over the short term. A team that has out-xG’d opponents 10 games in a row but lost half of them is mispriced. Bet against the form, not with it.
- Recalibrate when the table lies. If two teams sit on equal points but one has an xGD of +12 and the other -4, the table is misrepresenting the gap. The first team is far more likely to keep winning.
- Never bet purely on xG without context. Lineups, suspensions, fatigue, and game state all matter. xG is one input, not the whole story.
- Use it to find total goals value. Combined xG of the two teams in a fixture is among the best public predictors of total goals. Compare to the bookmaker’s Over/Under line.
Final Word
xG is not a magic number. It is a way of converting the noisy, low-scoring sport of football into a richer dataset that lets us evaluate process rather than just outcome. The teams that win the long game in betting are the ones who think in process, because the outcome is dominated by variance over short stretches and process over long ones. Once you start watching matches with xG in mind, you will find yourself unsurprised by the results that surprise the league table — and that is exactly the perspective from which value is found.