American Soccer Analysis introduced the metric g+ (goals added) back in May of 2020. I don’t think it’s an overstatement to say that it has changed the very way people think about and discuss soccer. Ok, that’s a bit of an overstatement, but it’s NOT an overstatement to say that in the two ensuing years, people have definitely looked at the g+ tables and gone “Maynor Figueroa? REALLY?”

The thing about metrics like this is that they can to be a bit startling when you look at them. They can challenge your eye test and of course aren’t desperately concerned with aligning closely to “conventional wisdom”. This can be a problem when you’re evangelizing about it. After all, why trust a metric that says Sebastien Druissi isn’t very good at soccer? I certainly don’t, and I work here. But, what if somewhere in this algorithm, the ability to show us things that we don’t inherently think are true is actually the biggest strength? After all, if a metric only shows what we already know and expect to see, why bother making it at all? Plenty of metrics out there tell us that a guy with 18 goals is a good goalscorer, and most of those don’t require a PhD in statistics or computational science to understand. In fact, I’d say nearly none of them have this whole bit about probability trees:

This “partial subsetting” described in option #3 is basically what a forest of decision trees does. At each “decision,” individual tree models look for places to split a predictor variable into two groups such that the average xG value earned in one split is significantly greater than in the other one. Then the tree looks for another split that makes a big difference. Then another one, etc. So maybe the first split is made at the 18-yard box (e.g. x > 97 for a 115-yard field length), and then second split is made to include locations within 20 yards of the center of the pitch (e.g. |y – 40| < 20 for an 80-yard field width). So far in this example, the model has now defined four subsets of the data, combining vertical proximity to goal and horizontal proximity to the center of the pitch. We’ll keep growing trees with up to five or 10 splits each, allowing the algorithm to effectively make and combine thousands of segments, in order to calibrate the value of any number of situations using league average possession xG values earned across a few splits at a time.

I had to read that several times, so now all of you do too. I almost stopped caring about soccer two sentences into that. I almost became a monk.

The good news is that computers are magic and all of the math happens off screen, so the only thing we have to worry about is soccer. Let’s take a look at a selection of the best g+ plays of the last week and see if our human sense of aesthetic and the cold dispassionate gaze of (I guess math?) can find some common ground on the beautiful game.

Depending on how well all of this goes, we’d like to try and do this every week. This is an interesting week to begin, because there are really only two matches featured here, but that should help us keep our focus narrowed a bit and follow the narrative threads of the match wherever they lead.

+0.25 g+ | Cristian Espinoza | Fouling

This is a good start I think, because we can all agree that penalty kicks are in fact, quite good, if you’re the team taking them. An ability to win fouls is frequently overlooked by analysts. In fact, it’s somewhat frowned upon, because when we describe someone as “good at winning fouls” we mentally translate that in our heads to “good at diving”. This is certainly true on some occasions, but in this instance we see no such deception.

Espinoza makes a gorgeous run, no doubt. But to my untrained eye, most of the credit here should go to 17 year old Niko Tsakaris who not only spots the run, but plays a perfectly weighted ball from the outside of his left boot that splits four Galaxy defenders and leaves GK Jonathan Bond with a lot of decisions to make in a short amount of time. Whether fouling Espinoza was a conscious decision or the unfortunate byproduct of having to be very aggressive to close him down only Bond knows, but a yellow card and conceding a penalty is the cost of doing business here.

For his efforts, Tsakaris was awarded with 0.10 g+ for receiving the pass, and then another 0.11 g+ for the pass itself. Not a bad g+ haul on the play for the youngster, but the computer (probably correctly) sees the highest practical value being the winning of a penalty.

2. +0.22 g+ | Raymon Gaddis | Receiving

Receiving is probably one of the more confusing aspects of Goals Added. As soccer fans, we’re trained our entire lives to appreciate the pass. We have many adjectives to describe passes, but not quite as many for being in the right spot to receive it. You can’t pass to nobody, no matter how hard Lewis Morgan tries (Sorry RBNY fans, check the stats and then take it up with him). It does take two to tango, and Gaddis’ run and reception here puts the ball in an extremely dangerous position.

Now, obviously, in retrospect we know that Sergio Santos is about to announce his candidacy for whatever the opposite of a Puskas award is, but that doesn’t matter to goals added. As hilarious as Santos’ miss was (and, indeed it was incredibly so), it’s irrelevant to what comes before for the purposes of this study.

It seems a minor thing, and may be hard to understand why the computer picks that moment out of this counter attack as the most valuable, but I think it’s probably right. Here’s the thing. That play is dead if Gaddis doesn’t make that lung-busting run. Acosta isn’t even trying to find him, he hits a decent but ultimately mistimed ball towards Santos who can’t quite cover the ground fast enough to get on the end of it. Gaddis, a right back, has done brilliantly here to keep the play not only alive, but incredibly dangerous.

On most nights, you’d have expected him to be awarded with an assist, but this is why assists are a very stupid stat, and he’ll have to make do with some nice words from some soccer internet dork instead. Good job Gaddis. The computer sees you, and so do I.

3. +0.20 g+ | Luciano Acosta | Passing

Luciano has been immense since joining FC Cincinnati. I would imagine he’s in the conversation for the MVP shortlist this year, and while I don’t think he has quite the narrative gas to beat out Driussi and Mukhtar this time around, he’s the league leader in xA and assists, he trails only Gil in Key Passes, and is third behind Mukktar and Gazdag in xG+xA.

All of that being said, he’s a tad bit fortunate here. The pass is obviously the right idea, and he does a fantastic job working the quick exchange with Vazquez on the left flank to send Yeimar on a walk-about, but he gets the angle of the pass a little bit wrong.

Abdoulaye Cissoko is not a first choice player in any of Seattle’s formations, but he’s here tonight due to a number of extenuating circumstances. This is one of those situations I personally find a bit difficult to comment on. On the one hand, Cissoko does well enough here to read the play and get any touch on it at all, but that’s about all that goes right for him. Instead of stopping the ball or deflecting it in a more advantageous direction, his touch only manages to delay the inevitable. Brenner is alert and quick enough to stay with the play, and he finishes with aplomb.

The computer will show Cissoko no mercy on this play, rating it the worst touch of the week from a g+ perspective.

Admittedly, It’s not the most aesthetically pleasing pass from Acosta, but this doesn’t really matter as far as the computer is concerned.

4. +0.19 g+ | Sergio Santos| Dribbling

Ah Sergio. You didn’t wind up looking too good the last time we talked about you, but the computer is pleased with you now. This series is just an absolute mess from Seattle, and it’s really hard for me to pinpoint why. It’s not really that there was any ONE moment where it just fell apart, but a series of cascading failures. Let’s explore. Dobbelaire’s throw-in is fine, perhaps a bit slow, but Yeimar actually backs away from it, and just simply doesn’t recognize the press threat quickly enough. This forces him to rush his pass to Ragen. The result is a pretty sloppy ball. That wouldn’t ordinarily be a huge problem on its own either. However, Ragen, (who has gone, in my estimation, over the course of this season from one of the most promising young center backs in the league to a young center back in the league) makes a four course meal out of it. It’s just not good enough from the young defender, and Sergio Santos pounces.

Santos just weight-rooms a recently recovered from injury Roldan here, and then manages to dance his way into enough space to create a decent look at goal. Again, the final product was lacking. But the computer appreciates the effort in creating it. Interestingly, the computer can’t see that Santos probably could have gotten the ball to Acosta, and that’s a much higher percentage opportunity if he had.

Santos has actually been excellent at finding teammates in his limited minutes this season. He leads all strikers in xA per 96 with a minimum of 800 minutes. This was a big let-off for Seattle, but a solid effort from Santos all the same.

5. +0.17 g+ |Chicharito | Dribbling

Oof. Rodrigues, my friend. Listen. It happens. San Jose steps back and lets LA build out the cross. Brugman, Delgado, and Araujo are only too happy to oblige. This is a textbook Chicharito goal. Rodrigues just isn’t expecting Chicharito to suddenly run laterally towards the center of the box. It’s such a good, sneaky, little run that throws San Jose’s defensive assignments into disarray.

Rodrigues is simply too slow to recognize that the run is about to become A PROBLEM. He’s quick to cover the lost ground, but then everything goes wrong as he loses his footing, swipes desperately at the ball in vein, and then gets megged by a delighted Chicharito, grabbing the dividends of that chaos he so wisely invested in 3 seconds ago. This was not a great moment from Rodrigues, but let’s also celebrate the ingenuity of that run from Chicharito.

6. +0.16 g+ |Chicharito | Passing

At this point, it probably goes without saying that Chicharito was having a pretty good night against the Earthquakes. Not much to this that isn’t self-evident. Obviously, the pass is what we’re here to talk about, but I want to look at this turn that creates the passing lane. I’ve gone back and looked at this a number of times trying to find some subtle shift or feint from Chicharito to send Cardoso in the incorrect direction. I’m not saying it isn’t there, but I can’t see it from this camera angle. We can either go with subtle Chicharito magic or Cardoso just completely misreading the play. Honestly, I wouldn’t argue with either. The pass is obviously exquisite.

An interesting note here, is that the computer actually viewed Grandsir’s dribble as one of the most negative g+ plays of the week. I had to ask around a bit to figure out why it was so upset with him, and the answer is interesting. One thing the model is really good at is reading the speed of a possession and attacking move. It likes things that move the ball quickly and Grandsir takes the ball with a lot of space in front of him, uses none of it, and also slows the play down by moving centrally into traffic and even AWAY from the goal at one point. I don’t think I would have been too upset were I a Galaxy fan watching it live, but watching it over and over again, I see why the computer is so annoyed. A potentially great chance turned into a very pedestrian effort.

7. +0.15 g+ |SERGIO Santos | Passing

If we could somehow erase the missed-sitter from our collective consciousness, we’d all have very little choice but to agree that Sergio Santos had a really nice match against Seattle. The throw-in here is taken quickly, and Seattle is slow to react. Acosta steps on the ball to bait Yeimar off of the overlap. Yeimar bites and the second he commits, Acosta plays a simple ball to Barreal. The cross is a bit overhit, but Santos is a very athletic gentleman, and he does brilliantly to head the ball perfectly back across the middle of goal. Vazquez, who has proven himself to be good at scoring goals, gets two bites at the apple here. There’s very little Frei can do in this situation other than exactly what you’re supposed to do: Cut the angle and make yourself as big as possible. It’s great work to deny the first effort, and Vazquez just can’t connect solidly on the second. Another big let-off for Seattle here to keep their microscopically slim postseason hopes on life support.

Bonus points to Santos’ visible dismay that Vazquez wasn’t able to finish such a great chance. That other shoe is gonna drop soon my friend.

I hope you enjoyed this little peek behind the g+ curtain to see what kinds of things the weird soccer computer that lives in our breakroom likes from soccer players. Going forward, if people enjoy this, hopefully having more than two matches to draw from will make a more interesting and diverse showcase.

Until then, the soccer computer has informed me that it has solved soccer, but will only tell me the secrets if I beat it at minesweeper. Wish me luck.

American Soccer Analysis

+0.25 g+ | Cristian Espinoza | Fouling