Why Analysts Strip the Penalties Out of a Striker's Numbers

A penalty is worth roughly 0.76 expected goals almost regardless of who's taking it or which goalkeeper is facing it, because the model is built from thousands of penalties taken from the same spot, at the same distance, against a keeper who has to guess. That consistency is exactly why penalties distort a player's xG total if you don't separate them out. Two strikers can finish a season with identical expected-goals totals, one of them built almost entirely from open-play chances and the other inflated by five converted penalties, and a plain xG comparison would call them equally threatening from open play when they clearly aren't.

Non-penalty expected goals, npxG, is the fix: take the same shot-quality model used for xG and simply exclude penalty kicks from the calculation. What's left is a cleaner picture of how much danger a player or team creates through actual buildup play, movement, and finishing from open play and set pieces other than the spot kick.

The gap between a player's npxG and their actual non-penalty goals tells the same overperformance story that xG does, just without a penalty taker's free conversions muddying the comparison. It's especially useful for comparing strikers who play for teams with different penalty-taking arrangements. A striker who doesn't take penalties for his club will always look artificially less prolific on raw goal counts next to a teammate who does, even if the non-taker is creating more dangerous chances from open play. npxG per ninety minutes is the number that corrects for that and lets you compare the two fairly.

It's also useful at team level for judging whether a side's attacking output is sustainable. A team that's scored a lot of goals partly because they've won and converted an unusually high number of penalties in a short stretch, more than their underlying play would suggest, is a candidate for regression once the penalty count normalizes, and npxG is one of the clearer ways to see that coming before it shows up in the results.

Recruitment departments lean on npxG per ninety specifically because it travels better across leagues than raw goal totals do. A striker scoring seventeen goals in a mid-table league and a striker scoring seventeen goals in a title-chasing side aren't necessarily creating the same quality of chance; league quality, teammate quality, and yes, penalty allocation all shape the raw number. npxG per ninety strips out at least one of those variables, which is part of why it shows up so often in recruitment reports even though it's one of the simpler adjustments in the analytics toolkit.

The stat isn't a verdict on penalty takers, and it shouldn't be read as suggesting penalties don't matter or don't take skill; converting them consistently under pressure over a full season is its own real skill, and missing them costs points just like missing any other high-value chance. The point of npxG isn't to dismiss penalties. It's to keep them from being counted twice: once as a converted goal, and again as evidence of open-play attacking quality they didn't actually demonstrate. Kept separate, both numbers, penalties and non-penalty output, tell you something a combined total blurs together.

It's the natural companion to expected goals itself, and if you're judging the goalkeeper rather than the striker, the equivalent adjustment on the other end of the shot is post-shot xG.