Stats explainer

Expected Goals (xG) Explained: The Visual Guide

Expected Goals (xG) is the single most important stat in modern football betting. It measures the quality of every chance created, not just whether it scored. Here is what xG really means, why it predicts the next match better than goals do, and how to read it visually.

xG = SHOT QUALITY × POSITION × CONTEXT Goal 0.42 xG 0.08 xG 0.05 xG
Get today's value picks at BetBot
Free stats-based football tips across 40+ leagues. Updated daily. No signup.

What xG actually measures

xG assigns a probability to every shot. A tap-in from two yards has an xG near 0.95: shoot it 100 times and you score 95. A speculative shot from 30 yards has an xG of 0.02: shoot it 100 times and you score twice. Add every shot's xG across a match and you get the total expected goals for each team. That number tells you who created the better chances, regardless of who finished them.

0.95
xG of a tap-in from 2 yards
0.42
xG of a close header from a cross
0.08
xG of a shot from 25 yards
0.02
xG of a long-range speculator

The number is built from ten million-plus shots in historical data. Every shot location, body part used, defensive pressure, and assist type gets compared to the historical conversion rate of similar shots. The output is a single probability per shot. Sum them up and you have the team's expected goals for the match.

Why xG predicts future results better than goals

Football has high finishing variance. A team can dominate possession, take 18 shots, accumulate 2.5 xG, and lose 1-0 because the one shot the opponent took was a fluke deflection that found the corner. That match looks like a defeat on the scoreboard but it was a winning performance under the surface. The same team will produce different results in their next five matches if their underlying numbers are repeated, because conversion regression to the mean is the strongest force in football statistics.

Goals vs xG: prediction accuracy at 10 matches
Just goals
38%
Goals + xG
58%
xG only
51%
xG difference
67%

The chart shows how often each method correctly predicts the winner of the next match between two sides after observing the last 10 matches. xG difference (a team's xG minus their xG conceded) beats simple goals-based predictions by 29 percentage points, which is the largest single-variable improvement in football prediction modelling.

How to use xG for betting

Look at xG difference over last 10 matches when projecting a team's future performance. The 10-match window smooths out single-match variance while staying current enough to reflect recent form.
Compare xG to actual goals scored to identify overperforming and underperforming sides. Teams scoring above their xG are likely to regress. Teams scoring below their xG are likely to revert positively.
Adjust xG for opposition strength. A team's xG against weak opposition tells you less than xG against strong opposition. Rebase by opponent quality before drawing conclusions.
Do not use single-match xG to predict the next match. One match is too noisy. Use the running average across 8-12 matches instead.
Do not assume the highest xG team always wins. xG measures chance quality. Finishing variance still produces upsets in the short term.

Where xG breaks down

xG is not perfect. It does not capture every nuance of a shot. A shot taken with the keeper out of position has the same xG as the same shot with the keeper square, even though the real conversion rate differs. A header from a corner has the same xG whether the centre back marking is the best in the world or the worst, even though the real outcome differs. The model uses average historical context for each shot location, which means the specific match context can deviate.

The other limitation: xG does not measure defensive contribution beyond shots conceded. A team that prevents the opponent from getting into shooting positions in the first place looks better in xG conceded than a team that allows many shots but blocks most of them. Both might end with the same xG conceded number but the defensive style is completely different. Watch the match to see which style produces the number.

Reading an xG match report

Most modern football stats sites publish post-match xG reports. The key numbers to look at: total xG per team (chance quality created), shots per team (chance volume), xG per shot (chance quality on average), big chances (shots with xG above 0.35). Together these tell you what kind of match it was: low-volume high-quality, high-volume low-quality, balanced both ways, or one-sided in either direction.

PatternxG TotalShotsxG/ShotReading
High-quality2.480.30Created elite chances, finished some
Volume scoring2.1220.10Lots of speculative shots, low conversion
Lucky win0.860.13Won despite weak attacking output
Unlucky loss2.6180.14Lost despite dominant attacking output
See today's full pick list
Updated by 06:00 CEST every morning across all major leagues.

Frequently asked questions

Expected Goals. The sum of all shot probabilities for a team in a match, where each shot's probability is based on its location, body part used, defensive pressure and other historical context.
xG is calibrated against millions of historical shots. The average error per match is about 0.3 xG. Across 10 matches the noise averages out and xG difference becomes the strongest single predictor of future results.
Higher xG means better chances created. Compare a team's xG to their actual goals to spot finishing variance, and compare xG to xG conceded for net performance.
fbref.com (free), Understat (free), Sofascore (live match xG), and StatsBomb (subscription). BetBot uses xG as a primary input in our daily picks.
Yes. xG difference over the last 10 matches predicts the next match's result more accurately than any single-variable method. It is the foundation of most professional betting models.