Expected Goals (xG) Explained: What It Means and How to Use It

Expected goals (xG) is a number between 0 and 1 that estimates how likely a shot is to be scored, based on thousands of similar shots from the past. A shot with an xG of 0.3 is the kind that goes in about three times in ten. Add up a team’s shots and you get their expected goals for the match, a measure of the quality of chances they created rather than just the goals they scored.
That’s why xG has gone from a niche analytics idea to something you’ll see on broadcast graphics, club websites and post-match arguments every weekend. Used well, it tells you which team really played better and which results might not last. Used badly, it can be just another stat to wave around. This guide explains how it works, what it can and can’t tell you, and how to read it like an analyst.
Quick answer
- Expected goals (xG) measures the probability of a shot becoming a goal, from 0 (no chance) to 1 (certain).
- Models use factors such as distance, angle, body part, type of assist and game situation.
- A team’s match xG is the sum of all its shots’ xG values.
- Penalties are usually valued at roughly three-quarters of a goal, which is why many analysts quote non-penalty xG (npxG).
- xG is most useful over many matches. One game’s xG can be noisy.
- Different data providers use different models, so figures from two sources won’t always match.
What is expected goals (xG)?
Every shot in football happens in a slightly different situation. Expected goals tries to answer a simple question about each one: if an average player took this shot, from here, in these circumstances, how often would it go in? The answer is expressed as a probability.
A tap-in from two yards with an open goal might be worth 0.9 xG or more. A speculative effort from 35 yards might be worth 0.02. Most shots sit somewhere in between, and most are worth a lot less than fans instinctively think. A typical shot from the edge of the box goes in only a small fraction of the time.
Where the numbers come from
xG models are built from large databases of historical shots, collected by data companies that log every event in a match. Analysts look at which factors made shots more or less likely to go in, then use statistical models to assign a probability to new shots based on those factors. The more shots in the database, the more reliable the model.
The term was being discussed by football analysts in the early 2010s, and by the end of that decade it had moved into mainstream TV coverage and newspaper match reports. Today almost every major data provider publishes its own version.
What goes into an xG model
Different providers use different inputs, which is one reason their figures vary. These are the factors most models include.
Distance and angle
The two most important factors. Shots closer to goal and from more central positions are far more likely to go in. A shot from six yards out on the goal line has much less of the goal to aim at than one from the penalty spot, even though it’s closer.
Body part
Headers are converted less often than shots with the feet from the same position, so they get lower xG values. Some models also distinguish between a player’s stronger and weaker foot, though many don’t.
Type of assist and situation
A shot after a through ball that puts the attacker one-on-one is more dangerous than a shot after a cross. A shot from a fast break is usually better than one against a set defence. Models also separate open play, corners, direct free kicks and penalties, because each converts at a different rate.
Defenders and the goalkeeper
More advanced models use positional data from the moment of the shot: how many defenders were between the ball and the goal, whether anyone was closing the shooter down, and where the goalkeeper was standing. This is one of the biggest differences between basic and premium models. Two shots from the same spot can have very different xG values if one is taken with the keeper stranded.
Reading xG after a match
When you see “xG: 2.4 vs 0.7” on a post-match graphic, it means the first team created chances that would, on average, produce about 2.4 goals, and the second team created chances worth about 0.7 goals. It says nothing about who won. It says who created better chances.
Why the scoreline and the xG differ
Football is low-scoring and random. A team can create 2.5 xG and score none because the keeper had a great day, the shots hit the post or the finishing was poor. A team can create 0.4 xG and score twice from long range. Over one match, that kind of gap is normal. Over a season, the gap usually narrows.
Many small chances or one big one
xG totals can hide how chances were created. Ten shots from 25 yards worth 0.05 each add up to 0.5 xG, the same as one decent chance from the penalty spot area. But the team with ten long shots is very unlikely to score more than once, while the team with one good chance either scores or doesn’t. Looking at a shot map, which shows each shot’s position and size, tells you much more than the total.
The xG family of stats
Once the idea of expected goals caught on, analysts built a whole set of related metrics.
| Metric | What it measures | Why it’s useful |
|---|---|---|
| xG (expected goals) | Quality of shots taken | Shows chance creation for a team or player |
| npxG (non-penalty xG) | xG with penalties removed | Stops penalty-takers’ numbers being inflated |
| xGA (expected goals against) | Quality of shots conceded | Measures defensive performance |
| xGD (expected goal difference) | xG minus xGA | A strong guide to underlying team strength |
| xA (expected assists) | xG value of shots a player set up | Credits creators even when team-mates miss |
| xGOT (expected goals on target) | Likelihood a shot on target goes in, based on where it was placed | Separates finishing quality and goalkeeping from chance quality |
| xT (expected threat) | How much passes and carries increase the chance of scoring | Values build-up play, not just the final shot |
xGOT and goalkeepers
Post-shot models such as xGOT are particularly useful for judging goalkeepers. Once a shot is on target, its placement and power can be measured. If a keeper concedes fewer goals than the xGOT of the shots they faced, they’re stopping shots that an average keeper would let in. That’s a much better measure than save percentage, which doesn’t account for how hard the saves were.
Expected assists
Traditional assists depend on a team-mate finishing. A midfielder can play five perfect through balls and get no assists because the striker misses all five. Expected assists (xA) credits the quality of the chance created, regardless of the outcome. It’s a fairer way to judge creators over time.
How to use xG properly
xG is a tool. Here’s how analysts, scouts and sharper fans use it.
Look at trends, not single games
A team that wins five matches in a row while being out-created in each of them is probably riding its luck. A team that loses while consistently creating more than its opponents is likely to improve. Rolling averages over 5 to 10 matches smooth out the noise and show the real direction of travel.
Use it to spot overperformance
When a striker scores far more goals than their xG over half a season, two things could be going on: they’re an exceptional finisher, or they’re on a hot streak that will cool. A small number of elite players, Lionel Messi being the obvious example, have outscored their xG for years at a time. For most players, though, big overperformance tends to fade. It’s one of the most reliable patterns in football data.
Compare per 90 minutes
Season totals favour players who play a lot. A striker with 12 xG from 3,000 minutes is getting into dangerous positions less often than one with 10 xG from 1,800 minutes. Analysts divide by minutes played and multiply by 90 to put everyone on the same footing. It’s a simple step, and it changes a lot of conclusions, especially when you’re comparing a regular starter with a player who often comes off the bench.
Sample size matters too. A substitute who has played 300 minutes and posted a huge xG per 90 might be a hidden gem, or might just have had two lucky cameos. Wait for a bigger sample before drawing strong conclusions.
Use it alongside your eyes
xG doesn’t see everything. It doesn’t measure a dangerous cross that nobody got on the end of, or a counter-attack that broke down before a shot. Watching the match, or at least the highlights, gives context the numbers can’t. Our highlights section is a good companion to any stats page.
Famous matches where xG tells a story
Some of the most famous results in football are reminders that the better-chances side doesn’t always win, and that finishing on the day matters enormously.
Take Brazil 1-7 Germany at the 2014 World Cup. Germany scored five goals in the first half-hour with clinical finishing from good positions, and the scoreline reflected a team taking almost every chance. Our Brazil vs Germany highlights show just how ruthless they were. At the other extreme, cup upsets where a side defends deep and scores from one of very few chances are exactly the matches where the losing team’s xG is much higher than its goal tally.
Title races offer longer lessons. Over a full season, the teams at the top of the league table and the teams at the top of an expected goal difference table tend to look similar, which is a big reason why clubs trust these numbers. When a side sits much higher in the real table than its xG suggests, analysts start predicting a fall. They’re not always right, but they’re right often enough to take seriously. You can follow the current standings in our Premier League section.
How clubs and scouts use xG
Professional clubs use far more detailed models than the ones shown on TV. Recruitment departments use xG and xA to find players who get into good positions regularly, even if their goal tallies at a smaller club don’t stand out. A striker who reaches 0.5 non-penalty xG per 90 minutes in a weaker league is getting into good positions, and that habit tends to travel.
Coaches use it to judge tactics. If a team is creating lots of low-quality shots from outside the box, the coaching staff may encourage players to work the ball into better areas. The general shift in recent years away from long-range shooting in elite football reflects exactly this kind of thinking.
Clubs also use xG against to assess defences. A defence conceding few goals but lots of high-quality chances might be relying on an excellent goalkeeper, which isn’t always sustainable.
Common myths and mistakes
“xG says who deserved to win”
It says who created better chances on average. It doesn’t account for the scoreline changing the game (a team 2-0 up often sits back and concedes shots), and it doesn’t measure everything that happened. “Deserved” is a judgement, not a number.
“All xG numbers are the same”
Different providers build different models. One might include defender positions and another not. The xG on one website and the xG on a broadcaster’s graphic for the same match can differ, sometimes noticeably. Compare like with like.
“A player with high xG and few goals is a bad finisher”
Maybe, but often they’re simply unlucky over a small sample. Finishing skill is real, but it takes a lot of shots, usually several seasons’ worth, before you can say with confidence that a player is a genuinely poor or genuinely brilliant finisher.
“xG ignores quality of players”
This one is partly true. Most public models assume an average finisher, so a chance falling to Erling Haaland gets the same xG as the same chance falling to a centre-back. That’s by design: xG measures the chance, not the shooter. Comparing a player’s goals to their xG is how you assess the shooter.
Frequently asked questions
What does xG mean in football?
xG stands for expected goals. It’s the probability, from 0 to 1, that a shot will be scored, based on historical data about similar shots.
What is a good xG for a team in a match?
It depends on the league and the opponent, but creating more than 2 xG in a match is usually a sign of a strong attacking performance, while conceding less than 1 xG suggests a solid defensive one.
What is the xG of a penalty?
Most models value a penalty at roughly 0.75 to 0.8 xG, reflecting how often penalties are scored. Exact values vary by provider.
Is xG a good predictor of future results?
Over a large number of matches, a team’s xG and xGA tend to predict future results better than its actual goals scored and conceded, because goals contain more luck.
What is the difference between xG and xGOT?
xG measures the quality of the chance before the shot is taken. xGOT measures the quality of a shot on target after it’s struck, taking its placement into account. xGOT is used to judge finishing and goalkeeping.
Why do different websites show different xG figures?
Each data provider uses its own model and its own inputs. Some include defender and goalkeeper positions, others don’t, so the numbers for the same match can differ.
If you want to put xG thinking into practice, head to our fan ratings page and see whether the crowd’s verdict on a match matches what the numbers say, or check the latest results and ask which ones might not repeat.
Comments