Diving Deep: Birmingham Legion 3-1 New York Red Bulls II

Examining the xG

To be brutally honest, there isn’t a lot to be said about this game tactically. In fact, this graphic pretty much sums it up:

On the left are shots, on the right is the heatmap. The Legion is attacking the right-hand goal in both cases. While musing the shots I recommend playing Nils Lofgren’s “No Mercy” and for the heatmap Led Zeppelin’s “No Quarter”. And if you don’t know who Nils Lofgren is, shame on you.

This was, to say the least, a very high press. For meaningful stretches of the game Matt van Oekel was the only player on either team inside the Legion half. Only 3 Legion field players averaged inside their own half (shockingly, they were Jake, Phanny and Alex) and they weren’t exactly backed up anyway. No bus parking in this one. Even the endgame was to keep the ball in the New York end. Ryan James played the final 8 official minutes on clean-up duty and he was offside twice.

So to fill out this post (because content is what we do here), I thought we could spend some time talking about Expected Goals. The infamous xG. For this game, the Legion’s xG was 4.841All xG data is from American Soccer Analysis.. That means the team underperformed its xG by almost 2 goals. Frankly, it should have been the pther way around; another 6-goal outing was hardly off the table.

In fact, the Legion has only exceeded that xG once this season, against Atlanta United 2, when it racked up a whopping 5.32. But on that occasion, the team outperformed its xG, scoring 6. The team’s other 6-goal game against Loudoun United had a lowly xG of 4.35. That game was the team’s best performance relative to its xG by quite a margin, the 1.65 difference being almost twice the next best result. For that you have to go all the way back to the season opening 1-1 draw with the Tampa Bay Rowdies. Yes, the Legion got a good xG differential in a game they only 1 goal. The xG on that day was just 0.15 though.

If you track the Legion’s goals and xG over the season, they both show steady improvement, which in fact is a repetition of last season’s trend. Here’s 2022 in graphic form:

The dotted lines are trendlines. Note that until very recently, the xG trend was better than the actual goals scored trend. They cross at about game 23, and the goals scored over the past 4 games are now trending better, even with the big negative difference against New York.

Now, we are all well aware that the Legion takes a ton of shots. Currently they are second in the league with 426 shots, behind Phoenix Rising’s 443. El Paso Locomotive in third trail badly on just 383 shots. For that reason, you would expect that the Three Sparks would have a relatively high aggregate xG for the season. And you would be right. The Legion’s 52.03 xG to date is in fact the best in the league on both an aggregate basis and on a per game basis (1.93). But here’s the kicker: the -8.03 difference between xG and actual goals is the worst in the league. Clearly, the above chart indicates that is changing, so it’s not all bad news. In fact, what it suggests is what we really already know: the Legion took a while to figure out its best attacking options and is just lately coming into its own.

Additionally, having a high goals to xG difference is not necessarily an indicator of success. Top in that category are the Colorado Springs Switchbacks on a gaudy 14.18. The Switchbacks started out really strong but have stumbled over the summer, remaining a contender nonetheless. But second at 12.29 is Orange County, the worst team in the Western Conference (a bit more on that in a moment).

Since both teams in a game get an xG, we can dive a bit deeper. The Legion has allowed 25 goals this season, 5th best in the league. But its xGA (i.e., expected goals allowed) is 32.58 (10th best). That’s a positive difference of 7.58, which pretty much cancels out the negative difference in goals scored. That is also 3rd best in the league behind Detroit City and Memphis 901. So on a net basis, the Legion is tracking pretty close to where it should be. Back to Orange County for a second: they have the worst xGA difference in the league at a rather humbling 10.32, which explains a lot.

A bit deeper again: with both teams getting xG, that means there is also an expected result. Which means expected points. At 50 points, the Legion currently has the 6th best total in the USL (but only the 4th best in the East). However, the team’s expected points to date are 47.54. That’s 4th best in both the entire league and in the East. The Rowdies come out on top in that category at 52.62. That puts the Legion 5.08 points off the lead. Where are they in the real world? 5 points off.  But it also means the team has earned more points than expected. Which is good, obviously.  Not as good as several other teams: San Antonio is a big overachiever, beating its xPts by a massive 14.08. That rather reflects their style, which is to absorb pressure and get goals where they can. Memphis also do really well at 11.95 extra earned points.

Players also get tracked on their xG. The Legion’s leader here is, unsurprisingly, Enzo Martinez. Enzo’s 11 goals (tied for 10th in the league) are very close to his xG of 11.23, which is 7th best in the league. The best performer here is Orange County’s Milan Iloski, whose 19 goals for the Golden Boot lead blow away his xG of 8.25, accounting by himself for almost all of OC’s team xG difference.

Finally, here’s the thing to keep in mind about xG. It’s a cool stat, granted, but one thing it is not is a leading indicator. Rather, it is a trailing indicator. It give you an idea of how well the team is performing, both in attack and in defense, relative to how well it should be performing. In the Legion’s case, we can see that the answer is that performance has gotten better over the course of the season. And as we get down to the final few games, that is reason for optimism.

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.