You have probably seen xG pop up on Match of the Day graphics, on football Twitter, or in post-match analysis. It gets talked about constantly, but rarely explained in a way that is actually useful. Most explanations are either too academic or too vague.

If you play a prediction game, xG is probably the single most valuable stat you can learn to read. Not because it tells you the score, but because it tells you something better - whether a team is performing above or below their actual level. And teams that are performing above their level tend to come back down to earth. That is where the smart predictions live.

What xG actually measures

Expected goals is a measure of chance quality. Every shot in a football match gets assigned a value between 0 and 1 based on how likely it is to result in a goal. A penalty has an xG of about 0.76. A shot from 30 yards out has an xG of about 0.03. A one-on-one with the keeper from six yards might be 0.45.

The model considers things like:

Distance from goal
Angle to the goal
Whether it was a header or a foot shot
The type of assist (through ball, cross, cutback)
Whether it was a fast break or a set piece
How many defenders were between the shooter and the goal

Add up all the xG values for a team's shots in a match and you get their total xG for that game. If a team had chances worth a combined 2.3 xG, they 'should' have scored about 2.3 goals based on the quality of their chances.

The key word there is 'should'. xG is a measure of what would typically happen given those chances. It does not account for individual finishing skill, goalkeeper performance on the day, or just plain luck. That gap between xG and actual goals is exactly what makes it useful.

Where to find xG data

You do not need a subscription or a data science degree. Several free sources publish xG data after every Premier League match:

Understat.com - free, detailed xG for every match and every player
FBref.com - comprehensive stats including xG for all major leagues
Infogol - good match-by-match xG breakdowns
The Athletic and BBC Sport both show xG in their match reports
Sky Sports and Premier League broadcasts now show xG on screen

Understat is probably the best starting point. You can look at any team's season-long xG, match-by-match xG, and even individual player xG. It is free and does not require an account.

xG vs actual goals: where the gold is

Here is where xG goes from being an interesting stat to a genuinely useful prediction tool.

Every team has two numbers that matter: their actual goals scored and their xG. When those two numbers are close, the team is performing roughly as expected. When there is a big gap, something interesting is happening.

Teams outperforming their xG

If a team has scored 25 goals but their xG is only 18, they are massively outperforming their expected output. This usually means one of two things. Either they have an exceptional finisher who converts low-quality chances at a high rate, or they have been lucky.

Most of the time, it is a mix of both, but the luck element tends to fade. Over a large sample of matches, finishing rates tend to regress towards the average. A team scoring well above their xG in the first half of the season will often see their goal output drop in the second half.

For prediction purposes: if a team has been winning matches 1-0 and 2-1 while creating very little, those wins might dry up. Consider predicting lower scores for them, or even draws and losses in tougher fixtures.

Teams underperforming their xG

The opposite is just as useful. A team creating 2.0 xG per match but only scoring 1.0 goal per match is due for a correction. They are making the chances but not finishing them. Eventually, the finishing catches up.

These teams are often sitting lower in the table than their quality suggests. They look like struggling sides based on results, but the underlying data says they are playing well. When you spot this pattern, it is worth being more optimistic about their upcoming results than the table would suggest.

Practical xG for prediction games

Step 1: Check the season averages

Before making your predictions for a gameweek, look up both teams' season xG averages. You can find these on Understat or FBref. You want two numbers for each team: xG per match (how many goals they 'should' be scoring) and xGA per match (how many goals they 'should' be conceding).

These numbers give you a starting point for the likely scoreline. If Team A averages 1.8 xG per match and Team B concedes 1.5 xGA per match, a prediction of 1 or 2 goals for Team A is reasonable.

Step 2: Look for the gaps

This is the step most people skip. Compare each team's actual goals with their xG. Are they overperforming or underperforming? If a team has scored 30 actual goals from 22 xG, they have been running hot. If another team has scored 15 goals from 22 xG, they have been unlucky.

When an overperformer meets an underperformer, the actual quality gap is probably smaller than the table suggests. These fixtures are prime candidates for upsets or closer matches than expected. That connects directly to what we covered about why underdogs win more than you think.

Step 3: Use xG to pick scorelines

Once you have a sense of each team's expected output, use it to narrow down your scoreline pick. Here is a rough guide:

Team averaging 0.8-1.2 xG per match: predict 1 goal for them in most matches
Team averaging 1.3-1.7 xG per match: predict 1 or 2 goals depending on the opponent
Team averaging 1.8+ xG per match: predict 2 goals, possibly 3 against weak defences
If both teams have low xG averages (under 1.2), a 0-0 or 1-0 is very plausible

This approach works well alongside what we know about the most common Premier League scores. The data tells us that 1-0, 2-1, and 1-1 are the three most frequent results. xG helps you decide which of those three is most likely for a given fixture.

Step 4: Spot regression candidates

This is the real power move. Every few weeks, scan the xG tables for teams with the biggest gaps between actual and expected performance. These are your regression candidates.

A team that has won their last four matches but has an xG of only 0.7 per game during that run is living on borrowed time. Their results are likely to turn. A team that has lost three in a row but has been creating 2.0 xG per match is due a win.

When you spot these situations, you can make contrarian predictions that have a genuine statistical basis. You are not just guessing. You are identifying teams whose results are about to shift.

Common xG mistakes to avoid

Do not treat xG as a prediction by itself. An xG of 1.8 does not mean the team will score 1.8 goals. It means their chances were worth that much on average.
Small samples are unreliable. One match's xG can be skewed by a single penalty or a missed sitter. Look at rolling averages over 5-10 matches instead.
xG does not capture everything. Set pieces, individual brilliance, defensive organisation, and game management are only partially reflected in xG models.
Do not ignore the eye test completely. If you watch a match and a team looks dominant despite a low xG, there might be context the numbers miss.

Putting it together

You do not need to become a data analyst to use xG effectively. Spending five minutes on Understat before your weekly predictions is enough. Check each team's xG averages, spot any big gaps between expected and actual performance, and use that information to nudge your scoreline predictions in a smarter direction.

The predictors who do this consistently are the ones who avoid the most common mistakes new predictors make. They are not guessing based on gut feel or blindly following the league table. They are using data to find the matches where the obvious result is not the most likely one.

Combined with an understanding of home advantage and form table analysis, xG gives you a genuinely well-rounded approach to picking scores. It will not make you right every week, but over a full season, it makes a measurable difference.

Try it this weekend on the football score predictor. Pick one or two matches where the xG data disagrees with the table, make a contrarian pick, and see what happens.

Keep reading

xG works best alongside form analysis. Our guide to reading a form table shows you how to combine the two.

Curious about which scorelines come up most? Our breakdown of the most common Premier League scores will help you pick realistic results.

New to prediction games? Start with our first gameweek walkthrough and build from there.

Expected Goals Explained: What xG Means for Your Predictions