npxGD
flagship metric · data to Mon 12 Oct 2026How many more (or fewer) non-penalty xG a side creates than it allows, per match (each match counts as 90 minutes).
For every league match of a side: the non-penalty xG of its shots minus the non-penalty xG of its opponent's shots. A season's npxGD is the mean over the side's matches, which is the same as per 90.
- npxG for: per match, penalties left out
- npxG against: per match, penalties left out
What goes in
our base data only- The shots of both sides of every league match since 2014-15, with 11footy xG.
- The league and the season of the match; nothing from the table or the result.
How it is computed
the formula in words- npxGD per match = (non-penalty xG for - non-penalty xG against) / matches.
- A side at +0.50 creates half a goal more than it allows in an average match.
When to trust it
sample and threshold policy- A side's number is ranked from 15 league matches; under that it is marked provisional.
- The check below uses full seasons only (30+ league matches).
- Compared within a league and season; the numbers of different leagues are not adjusted to each other.
How well it works
worked out from the data, not typed in- Over 1,150 full team-seasons, the even matches and the odd ones agree at r = 0.76 for npxGD against 0.68 for goal difference.
- The first half of a season's npxGD correlates 0.78 with its second half's goal difference against 0.75 for the first half's goal difference.
| Number, per match | Even v odd matches, r | Whole season, reliability | First half v second half's goal difference, r | Season v next season's goal difference, r |
|---|---|---|---|---|
| npxGD (11footy) | 0.76 | 0.86 | 0.78 | 0.78 |
| xGD, penalties in | 0.75 | 0.86 | 0.77 | 0.78 |
| Goal difference | 0.68 | 0.81 | 0.75 | 0.76 |
Sample: 1,150 team-seasons with 30+ league matches, 2014-15 to 2025-26; 886 sides played the next season in the same league. "Even v odd": the correlation of a side's mean over its even matches with its mean over its odd matches. The goal difference of the second half and of the next season is what actually happened, so a number that predicts it better is a better measure of how good a side is.
What it does not tell you
limits- The score changes how sides play: a side that leads sits back and creates less. The number is not adjusted for that, nor for the strength of the opponents.
- Numbers of different leagues and seasons are not adjusted to each other.
- Only the chances that are shots: no possession, pressing or defensive shape enters.
- It rests on 11footy xG, with the limits of the xG metric.
Where you see it
each one has a ? beside itExpected goals source: 11footy's own expected goals. The same description as data: /lab/api/metrics/npxgd.
Data to (the latest league match played), built . Unless a season is chosen, every view shows 2025-26 (2026-27 is still young: 0 players have 900 league minutes so far, so the numbers are of 2025-26 until half as many as last season do.). Percentiles and per-90 ranks from 900 league minutes, within the player's role, five leagues together; under 450 minutes a small sample.