Diving Deep: Tampa Bay Rowdies 0-2 Birmingham Legion

Messing with data in yet another way

Normally in this series we take a detailed look at how the Legion performed tactically in the most recent game played. Well, in this case the most recent game played was yesterday, of course, and thanks to US Soccer we have pretty much zero data to look at and since we weren’t able to go up to Chattanooga, did not even get to see the game. Just great.

We did watch the game down in Tampa, of course, and an excellent game it was. Tactically speaking, though it was only interesting in that Tommy Soehn changed things up this time by…not changing things up. He put the boys out there in a 4-2-3-1, and for once they stuck with it for virtually the entire game. They did take the same low press approach, and did so magnificently. They ripped a page out of the San Antonio playbook (the only page, truth be known) and took their opportunities where they arose. Following on from last week’s introduction of gameflow data, here’s this week’s chart:

As can be seen, Tampa had the lion’s share of possession. 65.6% to be exact. They just about doubled the Legion in keeping the ball. They managed just 12 shots with that possession compared to the Legion’s 11. Here’s the kicker: they had no shots on goal. When you cede control of the ball, it is very hard to keep the opponent from getting good scoring chances. Half of those 12 shots were outside the penalty area, 5 of which were blocked and the other just flat out missed. Of the 6 inside the box, 5 missed and 1 was blocked. All of which is to say that the Three Sparks defense was stellar.

A couple of other things are evident in this chart. First, the Legion had some decent early chances. Indeed, they could have put a couple of goals away in the first 15 minutes. But they had effectively two very good hances and they converted both of them. The second, and the one I really want to highlight here, is that with all that possession the Rowdies managed to accumulate an xG better than the Legion’s by a decent margin. It was even noted in the broadcast by the Tampa announcers, who seemed somewhat surprised as every attack got completely snuffed out.

Remember, xG stands for expected goals. That is, goals a team should score (or give up) given a probability analysis of each chance. So, what this means, then, is that the Rowdies got into good positions and failed to do anything with those positions. The Legion in contrast, exercised the chances they had (of those 11 shots 7 were on target, a very respectable percentage) and converted them, ultimately exceeding their xG. They’ve done that in 3 of 4 league games this season, the exception being last week against Hartford. In that game they were also on the wrong end of the xG battle. Further, the only game in which they had a higher xG than the opponent was against Tulsa, which ironically was the toughest game of the season to date.

Now, don’t get me wrong here: I am not trying to knock the xG concept. In fact, I find it very useful, especially as a tool to measure efficiency. Clearly, the Legion are being very efficient with their chances so far this season, and denying opponents theirs. Which leads to an interesting question (and what I really wanted to focus on in this post): how does Birmingham compare to the other teams in the league?

Well, take a look at this chart:

This will need some explanation. First off, all xG data here is drawn from American Soccer Analysis (by the way, the venerable John Morrissey of USL Tactics did a similar chart using his own xG data that differs quite a bit from this one). The horizontal axis shows xG scored on a per game basis (since the teams have all played varying numbers of games already). The vertical axis shows xG allowed on the same per game basis. The dotted yellow line represents the net zero point, i.e., where xG For and xG Against are equal. Teams below the yellow line have a net positive xG, team above it have a net negative xG.

Remarkably, only 9 of the 24 Championship teams have a net positive xG. And those include Oakland, Detroit and Orange County, none of whom have really impressed so far. Note that Sacramento is a massive outlier on the positive side and Louisville an outlier on the negative side, thanks entirely to that 5-0 shellacking which closely mirrored the xG (Louisville’s 0-3 loss to El Paso was surprisingly even on an xG basis).

Moreover, some of the better clubs in the league are very close to that yellow line. El Paso is the closest, though, with a net -0.02 xG, which given their poor start is a bit of  surprise. San Antonio is on the positive side (+0.18), but its xGF is barely above 1. New Mexico, who are expected to compete, are also very close to the line, but they have only 2 games under their belts here, so that’s not very representative.

Other than Lou City’s horrible numbers, perhaps the most surprising data point here is San Diego, who sit at a net -0.20, probably heavily influenced by the 0-1 loss to Sacramento. I would expect them to move closer to the yellow line in short order.

The Legion are a net -0.13, not bad, and if not for Memphis (!!!) would be pretty much the midpoint of this chart. Not sure what to make of that. Other than the fact that the Three Sparks like to make games close, which I guess we already knew.

As far as this weekend is concerned, Phoenix is allowing a league-high xGA of 2.01. They do create scoring chances at a respectable rate, but their net negative xG is third-worst in the league (after Louisville and Loudoun). Their actual goals allowed is exactly 2 per game, so they are right on the money in that regard, but their 1 goal scored per game is well under their xGF of 1.40.

That’s another area where xG is useful – comparing actual scoring to expected scoring. We won’t get into that here, but at a later date I imagine we will. Until then, let’s focus on actual goals, which always more fun than the goals you should have scored.


You may also like...

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.