If you ask any Alabama pro soccer fans attending a live game Tuesday evening – well, those not committing apostasy in Auburn at any rate – how many goals should have been scored they would likely respond with a number of at least 5. And all of those by the home team.
The Legion ended up defeating Corpus Christi FC 3-0. A strong result on any ordinary day but probably not strong enough to advance to the knockout rounds of the USL Cup should the Three Sparks not win Group 3 next month.
The stats nerds will however tell you that they overperformed. A caveat: I generally consider myself something of a stats nerd, so I am going rather against my own inclinations here. They will point to the Expected Goals number as proof. xG (not to be confused with the K-Pop girl group) for the Legion was 1.80 according to Opta, the official statistics provider of the USL. For Corpus Christi it was 0.37.
As to the latter, considering that the visitors managed just 4 shots in the game and none on target, that seems generous. For the Legion it implies they were close to double the goals they should have produced.
It gets even more stretched when you factor in that a penalty kick has a very high xG, for obvious reasons. Bofo Saucedo got a personal xG of 0.79 and the PK was his only shot. The other 8 Legion players who had a shot combined for 1.01 xG, none better than 0.28 (Romario Williams, who had a goal, an off-target shot and a blocked shot).
The history of xG is a bit cloudy but appears to have its origins in an academic paper back in the early 1990s but did not come into 2010 or thereabouts. The concept also exists in hockey, by the way, and given how chaotic that sport can be coming up with a meaningful number must be nearly impossible (unless you are talking about Canadians, in which case the number is always 0).
And it’s not as if there is an agreed-upon methodology. xG for USL games are also produced by American Soccer Analysis (which does not cover the USL Cup) and are usually close but not identical to Opta’s version. Algorithms vary but generally factor in distance from goal, angle to the goal, number of defenders, and, for some models, the build-up leading to the shot. Under these models a shot is considered to be what is known as a Bernoulli trial, that is, an experiment with only two possible outcomes. In this case, it would be a goal or no goal. The xG for a game is simply the sum of the xGs of every shot taken. I would note that political polls are also Bernoulli trials (yes or no for any candidate) so take that as you will.
Let’s go back to Corpus Christi’s xG. Of their 4 shots, 3 were taken from inside the Legion 18, one of which was even inside the 6. The other shot was from a fair distance out. One was on the penalty spot. The shot inside the 6 was their first of the game 33 minutes into the game and accounted for 0.23 xG, 62% if the total for the game. It was a header off a corner by defender Patrick Langlois (#6):
Yeah, that’s not going to get you much of a result. Sandwiched between 2 defenders and outjumped by Keegan Hughes, Sam McIllhatton covering the post and Jassem Koleilat right there. If anything, 0.23 is generous. Their second best of the game was in the 93rd minute by Blake Bowen from about 16 yards out for 0.10 xG and was similarly under serious defensive pressure.
But any attempt on goal has a non-zero chance of finding the net regardless of how weak it is. They all get credit. Not an A for effort, but not an F either. Thus Corpus Christi gets an xG score. Realistically, none of their shots had a chance.
Conversely, take a look at Romario’s goal:
Granted, it is kind of a backheel, but I am not aware of any xG methodology that takes the kind of shot into account. He is facing away from goal, though, which makes the shot trickier. Or possibly he thought it would make it easier, since he couldn’t know his view of Kaylor Hodges was being covered by keeper James Talbot. That aside, he’s barely 11 yards out, directly in front of goal and completely free of the defense. Remember, his cumulative xG for the game was 0.28 for 3 shots. So this goal probably got him less xG than Patrick Langlois’ header above. Here’s where the build-up possibly has an effect. Sam McIllhatton was credited with the assist but his delivery looked more like a shot and only got to Romario thanks to a deflection by midfielder Enock Kwakwa. Which turned out to be a total gift.
Sebastian Tregarthen, who got the final Legion goal, was awarded 0.12 xG. He had 2 shots total as well, one inside the 18 was blocked, one outside the 18 went wide. The goal was also outside the 18 but was an absolute banger that was never going to be stopped.
In summary, xG can be very misleading. At the very least it can disagree markedly from subjective opinion as to how good a scoring chance was. And let’s face it, soccer is not a sport that lends itself very easily to statistical analysis given how open and unpredictable play can be. As Americans we are very used to stats in our sports and argue about them incessantly. But football and baseball in particular are much more amenable to quantification. Of course, that doesn’t mean we won’t try to shoehorn soccer into our stats-obsessed mindset.
And there is one more issue to take into account, and it’s one that affects the Legion quite heavily. xG does not give you any credit for shots you don’t take. The Legion had 14 shots in the game. 4 were on target and 9 were inside the box. Those are good numbers by the team’s standards; the Three Sparks are averaging 11.2 shots per game overall. However, the team had 25 touches inside the box, a very good number by the same light.
However, that also means 16 of those touches did not result in an attempt on goal. Sometimes that isn’t going to be possible but you have to convert touches into chances as much as you can. Ronaldo Damus had at least one, possibly two, opportunities that he hesitated on and Peter-Lee Vassell had the ball on his chest at one point and couldn’t do anything with it. A chance is a chance. A player really should not think too much about improving it, which seems to be what the Legion attack has a habit of doing. Looking for that better chance often means losing the chance entirely. You improve your scoring chances by improving the number of chances. Quantity, after all, has a quality all of its own.
The big positive from this game will be the confidence-building. This was only the second 3-goal game this season (the first was the 3-1 win over Rhode Island FC) and the first at home. With a key stretch of 5 league games in 23 days coming up (3 at home, 2 away) the Three Sparks have a huge opportunity to climb the standings. Proving to themselves that they can score and create chances to score. That last is another wierd statistiv but much easier to understand; it’s simply any pass that leads to a shot on goal. The Legion put up 9 of those Tuesday; again, it doesn’t include passes where the shot was not taken. After the game Bofo put it this way: “In reality we create so much, so we just need to find the back of the net and that’ll change the dynamic for the rest of the season for sure.”
