Expected Goals Explained — and Why This Site Doesn't Publish xG
Expected goals (xG) puts a probability on every shot. A tap-in from six yards might be worth 0.7 goals; a speculative effort from thirty yards, 0.02. Add up a team's shot probabilities and you get the number of goals an average team would have scored from those chances.
Before going further, the disclosure that belongs at the top rather than buried: this site does not compute or publish xG. We do not hold shot coordinates, and xG cannot be derived without them. What follows is an explanation of the metric and of the closest thing our data actually supports.
Why xG exists
Goals are the point of football and a terrible measurement of it. A team can play well and lose 1–0; it can play badly and win 2–0 on two deflections. Over one match the scoreline is mostly noise. Over a season it is mostly signal. The problem is that decisions — about a manager, a player, a bet — usually have to be made somewhere in between.
xG attacks this by measuring the quality of chances created rather than the outcome of those chances. Chance quality stabilises faster than goals do, because it throws away the finishing luck and the goalkeeping heroics that dominate small samples.
What goes into a shot's value is roughly:
- distance from goal
- angle to goal
- body part (foot, head, other)
- assist type (through ball, cross, rebound)
- defensive pressure and the number of bodies in the way
Model this over hundreds of thousands of historical shots and you get a probability for each new one.
What we can compute instead
Without coordinates, the honest substitute is shot conversion, and it is a much blunter instrument. Across the 147,719 matches in this database that carry shot data:
| Metric | Value |
|---|---|
| Shots on target | 1,340,961 |
| Goals | 392,538 |
| Conversion rate | 29.3% |
| Shots on target per goal | 3.42 |
| Goals per total shot attempted | 11.3% |
Roughly one shot on target in three becomes a goal, and one shot in nine overall.
Why that is not xG, and where it misleads
Shot conversion answers "how many shots did it take?" — not "how good were the chances?" Those come apart in exactly the situations you most want to understand:
- A team taking 20 long-range shots and a team taking 20 shots from the six-yard box look identical on this measure. They are not remotely the same performance.
- Shots on target already exclude the misses, which is precisely the information about chance quality you wanted. A player who blazes a great chance over the bar disappears from the numerator and the denominator.
- Conversion rate is heavily distorted by penalties, which convert at roughly three times the open-play rate.
So use it for what it is: a rough check on whether a team is scoring more or less than its shot volume implies. Do not use it as a substitute for chance quality, because it does not measure chance quality at all.
How we calculated this
Shot data exists for 147,719 of the 336,765 finished matches — mostly the more recent seasons and larger competitions. Older matches often carry only a scoreline.
The rule applied throughout this site holds here too: a missing value excludes the match from the calculation rather than counting as zero. A game with no recorded shots is not a game with zero shots, and treating it as one would be a serious error at this scale.
That exclusion means these figures describe modern, well-covered football rather than the full historical record.
Explore it yourself
The Game Statistics Explorer has shot filters — total, on target, off target, blocked, inside and outside the box — that you can combine with a date range or competition to see conversion for any slice you like. Any team profile carries the same underlying record.