The Assist Stat That Rewards the Pass, Not the Finish

An assist is a strange stat to build an evaluation around, because half of it depends on someone else. A midfielder can play the perfect ball into a striker's path, low and hard across the six-yard box, and get nothing for it if the striker scuffs the finish. Play the same pass to a striker having a good week and it counts. The pass didn't change. The outcome did, because finishing is unpredictable in exactly the way that makes raw assist totals misleading over anything shorter than a full season.

Expected assists tries to fix that by scoring the pass itself rather than what happened after it. Every time a player creates a chance, that shot gets an xG value based on its location, angle, and type, the same way it would for any other shot. The passer is then credited with that xG value as their xA, whether or not the shot goes in. Set up a one-on-one with the keeper and the striker blazes it over the bar, the passer still gets full credit for creating a high-quality chance. That's the whole idea: separate the value of the creation from the luck of the conversion.

This matters most for players who keep setting up good chances for strikers who are missing them. A winger can lead a league in chances created and finish mid-table in assists, and xA is the number that explains the gap. It's also the number that flags a false alarm the other way: a player who racks up assists from low-quality service, a deflected cross that falls kindly, a cutback that a striker somehow buries from an impossible angle, will have an assist total that looks better than their actual chance creation.

Kevin De Bruyne is the standard reference point here because he has spent years living near the top of both charts at once, which is rarer than it sounds. Most players who create a lot of chances create some good ones and a lot of hopeful ones. Sustaining a high xA per ninety over a full season means doing it from your first touch to your ninetieth, week after week, not just producing one or two moments of magic that highlight reels remember.

xA has the same blind spot as xG, which is that it reads geometry and assist type but not the run itself. A pass that a striker had to reach with a perfectly timed sprint is scored the same as a pass hit to a striker standing still, provided the eventual shot comes from the same spot. Some more advanced models try to account for defensive pressure at the moment of the pass, but the public versions most fans see usually don't go that deep. It's a description of the chance created, not a full account of the buildup that made the chance possible.

The other thing worth knowing is that xA and assists converge over a large enough sample. A single season can have wide gaps between the two because finishing varies so much game to game. Look at a player's numbers over three or four seasons and the gap usually narrows, because luck stops being able to hide. If a player's actual assists sit well below their xA for one bad month, that's probably the strikers around him missing chances. If it's been three years, it might be worth asking whether his chances are as good as the underlying numbers claim.

It sits alongside expected goals as the other half of the same idea, and it's only part of the picture: for the buildup two passes before the shot rather than just the final one, see shot-creating actions.