Expected Goals and Atlanta United

By Harrison Crow (@harrison_crow)

Despite winning 6-1 on Sunday, according to our model Atlanta (1.79) had fewer expected goals than Minnesota (2.04). I spent a bit of time on twitter Monday evening trying to explain why expected goals don't have a special distain for Atlanta, and why all those goals over the weekend by Josef Martinez and company weren't as easy to score as you might think. But, I'm also not saying they were poor attempts, either.

Here is a quick look at each of Martinez's goals on Sunday.:
2:02, 1st goal: 0.0936 xGoals value

26:17, 2nd goal: 0.2633 xGoals value

74:13, 3rd goal: 0.4166 xGoals value

Yes, these are all very good chances. But they’re not sure things, and ASA now has seven years of examples that can provide us with enough demonstrations of similar attempts that were both successful and went awry. I say that neither to demean those players that did not finish the shot nor to lift Martinez above his peers as “better” just because he did finish it.

Sometimes with goal scoring it’s… well, I don’t want to call it dumb luck but, maybe, just circumstances of the moment: weather, keeper positioning, defenders and just the general feeling of that moment. There are tons of variables that we, and others, don’t and in some cases can’t take into account that factor into the equation of scoring a goal. And then sometimes it is just luck—see: Lefty Gomez if that’s the case.

That said, don’t look at these values as negatives. Martinez’s goal in the 75th minute ranked as the 10th highest xG moment for the weekend and third highest xG from the run of play — as opposed to a penalty kick or a fast break which rank higher.

The only two moments from a run of regular play which ranked higher this weekend were Jacob Peterson’s opportunity (to the right) in the 94th minute of the same game and Ola Kamara’s shot on Tyler Deric in the 93rd minute of Houston hosting Columbus the day prior, both of which were converted.

The reason that Atlanta had a generally low xG value is that they have not created a lot of chances, but they've created really good ones. Minnesota likewise created a lot opportunities , which being behind and pushing as many figures forward as they could, isn't really surprising. What is surprising is they didn't score at least one goal from the run of play and that is at least in some small part a credit to the Atlanta defense.

Expected goals isn't sentient and it doesn't know everything about a game. Sometimes it will match the eye test and other times it’s going to see different things than you or I. The later presents opportunities to show either what we’re not seeing or not valuing, that the model does.

Expected goals doesn't hate Atlanta. Likewise, it doesn't expect (and neither do I) their very talented attack to continue to finish 5 of every 8 chances in the box, even with better than average chances. This is life and despite the early claims, Josef Martinez and Atlanta are only human.