Is this measuring the passer or the finisher?

The finisher, mostly. The passer controls the quality of the chance created, which is what expected assists capture. Whether it becomes an assist is decided by whoever takes the shot.

Does the gap correct over a season?

Usually. Finishing at the end of a creator's passes tends to drift back towards normal rates, which is why a large negative gap in autumn has often closed by spring.

How is this different from goals minus expected goals?

That delta is about the creator's own shooting. This one is about everybody else's. A creator has no control over the finish, so read xA delta as a comment on the team around them, not on the creator's individual quality.

How large a sample should this be trusted over?

A full season of a creator's chances is the minimum, and even then the number reflects the finishing of several different teammates, not one player's technique. A short run of matches with the same forwards on the end of it tells you very little.

Should a creator be judged worse for a negative delta?

No. A large negative figure over a settled sample usually means teammates are missing good chances, which is on them, not the passer. Judge the creator on expected assists and treat the delta as a note on finishing around them.

Can a striker have a meaningful xA delta too?

Yes. Any player who registers assists carries one, and a false nine or a forward who lays chances off for a strike partner is exactly the profile where this number is worth checking alongside their own goals minus xG.

Assists minus Expected Assists

xA Δ

Assists minus expected assists is a player's actual assists less their expected assists, showing whether teammates are converting the chances they create.

Why it matters. A negative figure means a player is creating more than the assist column credits them with, which is often the most undervalued profile on the board.

Figures from xG Stat’s own models, built on Wyscout event data. Updated 6 August 2026.

A number that depends on somebody else's finishing

Subtract expected assists from actual assists and the result is not about the passer's technique. It is about whether the chances that passer created went in.

Positive. Teammates are converting more of this player's created chances than the model expects. The creator is riding good finishing around them.
Negative. Teammates are converting fewer. The creator is doing the hard part, opening up the chance, and the final touch is letting them down.
Near zero. Finishing at the end of this player's passes has matched the chance quality on offer, which is the normal state across a full season.

Expected assists already accounts for how good the chance was: the pass type, the location it arrived at, the pressure the shooter faced. The delta strips that out and leaves only what happened once the ball reached the finisher's foot.

This runs opposite to the league table deltas, where xPts and x# are expected minus actual, so a positive number there flags underperformance. Here, actual minus expected means positive is overperformance, in this case by whoever received the pass.

Three scenes, three verdicts

A playmaker on -4 in September. Four clear chances created, none converted. The passes were good, the finishing around them was not. Verdict: expect the assist column to catch up over the following months, because a creator's chance quality is a much steadier signal than a handful of finishes.

A wide player on +5 by November. Every low cross has found the far post. Verdict: enjoy it, but do not expect the strike partner to keep converting at that rate. The number says more about their finishing streak than about this player's delivery improving.

A number ten holding close to zero across two seasons. The chances they create are converting at roughly the rate the model expects, season after season. Verdict: that consistency in expected assists itself, not the delta, is the real read on their creative quality.

Sample size matters twice as much here

Goals minus expected goals depends on one player's finishing across many shots. This delta depends on several different players' finishing across the chances one creator makes.

Under half a season: the finishers involved change too often for the number to settle. A different striker on the end of the same service can flip the sign.
A full season with a stable strike partnership: worth reading, though still with an eye on which player finished which chance.
Multiple seasons: the strongest signal, because it averages out changes in personnel as well as ordinary finishing variance.

Do not read this over five matches. The sample is small twice over, both in chances created and in who happened to be on the end of them.

Two misreadings worth correcting

"A big negative delta means this player cannot create good chances." The opposite. Expected assists says the chances were good. The delta says somebody else missed them. Judge the creator on the expected assists number itself, not on this one.

"This measures one player's skill, the way goals minus xG does." It does not. Goals minus xG is entirely down to the shooter. This delta depends on the shooter being someone else entirely, so it is really a two-player number wearing one player's name. Read it as a comment on the partnership, not the individual.

Both halves of this delta can differ slightly between sites, because each provider logs its own events and "assist" carries more than one definition. Why our numbers differ explains what will and will not match.

Assists furthest above expected

2025/26 final

1E. HaalandManchester City5.8
2Bruno FernandesManchester United4.3
3G. XhakaSunderland3.6
4L. DigneAston Villa3.4
5H. WilsonFulham3.2
6Gabriel MagalhaesArsenal3.0
7P. SarrTottenham Hotspur3.0
8Q. HartmanBurnley2.8
9Matheus NunesManchester City2.8
10M. SenesiBournemouth2.7

See every player's numbersData to Mon 25 May, 15:00

FAQ

Related terms

See xA Δ across the Premier League

Every player, team and match measured on the chances created rather than the scoreline.