clock menu more-arrow no yes mobile

Filed under:

The AfterMath: Atlanta United 1-0 LD Alajuelense

There are lies, damned lies, and Opta data

SOCCER: APR 13 CONCACAF Champions League - LD Alajuelense at Atlanta United Photo by Rich von Biberstein/Icon Sportswire via Getty Images

Sometimes getting to useful information can be very difficult. Consider this from MLS’ stats page for the game:

Keep that shots stat in mind too. Then there’s this from the very same page:

And then there’s this from the CONCACAF page on the game:

And also this:

From this you might conclude that Opta, the source of most soccer data worldwide these days, is about as reliable as a polling firm.

The fact is that sports statistics are often pretty mysterious (anyone care to explain xG?). The soccer possession stat being a prime example. American sports fans are well-acquainted with the concept of relative possession from watching football. An in that sport it is extremely easy to calculate: you know exactly when each team has the ball and for how long. In soccer it’s a bit more amorphous.

The above images gave rise to a conversation in the DSS Slack. Most of us assumed – and I will hazard a guess that most of you do too – that the number was arrived at because some intern sits there with a chess clock kind of gadget clicking madly away every time the ball changes hands. One upon a time, and in some leagues even now, this was indeed the case. But Opta in all its wisdom thought it much more useful to calculate possession simply by the relative number of passes each team makes.

So that poor intern now has to click even more madly every time a player touches the ball. And is that stat truly meaningful anyway? Who knows.

Differences in shots on the other hand I can understand. What one observer considers a shot might look like a pass to another.

OK, complaints over with. As far as possession is concerned for this game, I’m inclined to view CONCACAF’s take as more reliable, if only because that’s how the general drift of play went. Atlanta certainly seemed to have the edge, but there is no way they had 2/3rds of it. Moreover, they were nowhere near as dominant as they had been in the first leg up until Guzan’s red card.

And considering that Alajuelense were hamstrung by poor decisions about international availability (a problem Atlanta certainly did not have), that thin a possession advantage is not great. Despite starting a very young and somewhat untested side, Alajuelense were not about to give up easily. Indeed, that they survived 90+ minutes without conceding, given that Atlanta were peppering their goal all game, is a very creditable performance.

In fact, if not for Rios Novo’s left foot they would have taken the lead, and that would have been a big game-changer.

Still, the game was largely compressed into the middle of the pitch. Here are the heatmaps (Atlanta on the left):

The passing network for both teams is similarly compressed.

The key change in the game was Heinze’s half-time adjustment. He had started the game with a 3-4-3, with George Bello and Brooks Lennon as wingbacks. At half-time he yanked Franco Ibarra and inserted Jürgen Damm, shifting to a 4-5-1, pulling Bello and Lennon back to fullbacks, and leaving Josef Martinez (and later Lisandro Lopez) as a solo forward. That seemed like a mistake; for all his speed Damm has not impressed as a winger in Atlanta, and Bello in both legs of this match-up has shown himself weaker on the defensive side than going forward.

But the Five Stripes had managed only 5 shots in the first half; they doubled that in the second, and penetrated deeper more frequently. Damm’s crossing was erratic at best, although he of course ended up ensuring the victory, albeit not by passing.

The other key was Heinze’s use of Ezequiel Barco. Barco started out the game looking shaky, even coming close to another disastrous play in the back like the one that made Guzan a spectator for this leg. But overall he was the playmaker. The passing network is key to understanding how he was utilized:

Barco is #8. Josef, of course, is #7. So Barco started the game as the leftmost of the three forwards, and ended it as the wide man in the midfield 5. Not that the map above would lead you to think that. Gabriel Heinze has stated that he believes Barco can play both wide and centrally. Since he came to Atlanta, we have regularly seen him pushing play into the middle. Frankly, he is probably better in the middle than out wide. He also looks energized, and perhaps more importantly, comfortable in the new scheme. This has been flagged as Barco’s breakout season. So far, it looks like he knows that, and is playing that way too.