Arsenal left the Etihad with nothing. Manchester City 2, Arsenal 1. The scoreline reads like a deserved home win. The data reads like something else entirely.
The xG Verdict
Arsenal generated 1.52 xG to Manchester City's 1.41 xG. In a match where they held just 41% possession and managed only 9 shots to City's 15, Arsenal's chances were simply better. Eight of their nine shots came from inside the box, a conversion of shot selection into quality that the xG model rewards. City had volume. Arsenal had precision.
And then there were the posts.
Arsenal hit the woodwork twice. Two shots that needed millimeters to change the scoreline, and the entire complexion of the title race. The xG model already accounts for shot location and angle, but hitting the post is the cruelest form of finishing variance: you did everything right except bend physics by a fraction. Those two post hits are baked into Arsenal's 1.52 xG, which means the model already "knew" those were good chances. They just didn't go in.
City, meanwhile, converted 2 goals from 1.41 xG. That's a finishing rate of 142%. In any given match, that's fine. Over a season, that's the kind of number that regresses.
The Poisson Says Draw
Running both xG totals through a Poisson match simulation, Arsenal win approximately 38% of the time, the match ends in a draw 29% of the time, and City win 33% of the time. A City victory was the least likely of the three outcomes.
Arsenal didn't just deserve a point. They were, statistically, the slight favorite to take all three.
What Actually Happened
Riyad Cherki opened the scoring on 16 minutes for City, assisted by Matheus Nunes. Arsenal responded immediately: Kai Havertz equalized just two minutes later. The match was even, the xG was even, and the momentum was Arsenal's.
Then Erling Haaland scored on 65 minutes. The winner. From a City side that had generated less expected threat over 90 minutes than the team chasing the game.
The final 25 minutes saw Arsenal throw everything forward. Posts. Saves. Nothing.
The Title Race Implications
In a title race decided by margins, this result is the kind of data point that separates champions from nearly-men. Arsenal's CURSD Luck Score will tick further negative after this. They were already on the wrong side of finishing variance in several matches this season, and today added to the ledger.
The uncomfortable truth for Arsenal fans: the data says they deserved at least a draw. The standings say they didn't get one. That gap, between what should have happened and what actually happened, is the entire reason the CLS exists.
Two posts. More xG. Fewer points.
The data says what the scoreline hides.