Fun With Expected Goals

Would you look at that? American Soccer Analysis updated their expected goal data! Let's have some fun. Well, I think it's fun.

For those who don't know, here's ASA's explanation of expected goals. But that's really long and complicated. Essentially, expected goals looks at where people take shots and says how often you can expect that shot to go in based on every shot taken from that spot. ASA does it by distance and angle, some places do it by zone, but the idea is the same. WHAT EXPECTED GOALS DOESN'T DO is look at defender placement. This is important (thus the all caps) because some teams have extraordinarily good defensive positioning and some have very bad positioning, corners are hard to score on because there are defenders at the far and near post (in theory), breakaways are easy-ish even if taken from the top of the 18, etc... Because of this, you need to look at at least half a season of data before you should take any real conclusions.

Cool, now that we've done that, let's look at some data! The only team thing I'll do is say that the Colorado Rapids are good at defense as they only have 22.9 expected goals against, the lowest in the league by a solid three goals. Also, our defensive positioning is not as good as I would have thought. Looking at goals against minus expected goals against, which should show how good your defenders and goalie are compared to average, we're only at -4.1 (meaning we've let in 4.1 less goals than expected). That's still pretty well above average, but the LA Galaxy are at -7.72!

On to the players! Guess who's leading the league in expected goals. I'll give you two guesses. Giovinco you say? No, that's not it. David Villa you say? Well, yeah, that's the one. That wasn't very shocking was it? But guess who's leading in Goals minus expected goals (showing how well you finish the shots you take)? Ignacio Patti! With 4.68 more goals than expected. That's really good. Okay, that's all I really want to focus on for the whole league, because the whole league minus one team is the worst and we hate them. Let's talk Rapids.

Here's a table!

First Last Min Shots G xG G-xG xG/shot A xA xG+xA
Kevin Doyle 1620 48 4 5 -1 0.10 2 1.93 6.93
Marlon Hairston 720 12 3 1.94 1.06 0.16 2 1.68 3.62
Jermaine Jones 726 19 3 1.76 1.24 0.09 1 0.84 2.61
Shkelzen Gashi 1354 61 3 3.97 -0.97 0.07 3 3.4 7.37
Sebastien Le Toux 1295 15 2 3.89 -1.89 0.26 5 5.51 9.39
Axel Sjoberg 1869 10 2 1.15 0.85 0.12 1 0.7 1.85
Dominique Badji 779 21 2 2.02 -0.02 0.10 1 1.25 3.26
Marco Pappa 732 23 2 1.11 0.89 0.05 3 1.97 3.09
Bobby Burling 1012 9 1 1.25 -0.25 0.14 0 0.06 1.31
Dillon Powers 1225 22 1 1.99 -0.99 0.09 2 1.9 3.89
Sam Cronin 2070 19 1 0.88 0.12 0.05 1 1.27 2.15
Dillon Serna 318 5 1 0.22 0.78 0.04 0 0.3 0.52
Jared Watts 982 3 0 0.36 -0.36 0.12 0 0 0.36
Conor Doyle 40 1 0 0.08 -0.08 0.08 0 0 0.08
Marc Burch 1532 3 0 0.13 -0.13 0.04 2 1.39 1.52
Dennis Castillo 128 1 0 0.04 -0.04 0.04 1 0.35 0.39
Eric Miller 1092 4 0 0.13 -0.13 0.03 0 0.41 0.54
Micheal Azira 1918 10 0 0.32 -0.32 0.03 0 0.53 0.85
Mekeil Williams 1447 2 0 0.03 -0.03 0.02 2 0.93 0.96

Let's take a look at some stuff. We've been over what expected goal (xG) and goals minus expected goals (G-xG) means, but most importantly (because I'm narcissistic) is that xG/shot stat that I think I made up. Expected goals per shot should measure what kind of shots a player takes. Shot selection is really, really important (*cough cough* Gashi *cough cough*) and can really help or really hurt a team.

We all know Doyle leads the team in goals with 4 (which is tied for the worst team leader in the league), so it's no surprise to see he leads the team in expected goals with 5. However, that -1 G-xG value is really bad. Like second worst on the team bad. Like 343rd in the league bad. No, I didn't accidentally hit two keys when I meant to hit one key. He's three hundred forty third in the league in the best stat there is for finishing ability. Oof. (worth noting I didn't limit by minutes or shots or anything, so there's probably some weird stuff in front of him, but that's still bad). But how about Jermaine Jones, huh? 1.24 G-xG is really good! He's 29th in the league! But wait, he doesn't play. So how about Marlon Hairston! 1.06 is also pretty darn good!

Let's talk about my stat a little. xG/shot should show shot selection ability (this one gets a little funky when it comes to defenders whose only shots are headers inside the 6 on set pieces, so I'm gonna limit us to a minimum of 10 shots on the year). Le Toux leads our team with 0.26. That's really, really good. Only Robbie Keane is better at 0.27. Marlon Hairston is up there in this stat, too. Second on our team and 40th in the league.

Okay, now how about the inspiration for the creating of my xG/shot stat? The man, the myth, the legend, the divisive Shkelzen Gashi! Gashi (obviously) leads the team by a large margin in shots with 61. He is second on the team with 3 goals. He's got 3.97 expected goals, which means he is not a good finisher. 341st in the league in G-xG. But I want to look at his shot selection. Gashi's xG/shot is 0.07. That's really, really bad! Even Conor Doyle picks better shots and he's only been on the field 40 minutes. Gashi finally made a long bomb. It made for a great highlight and we all screamed and jumped up and down, but one goal doesn't justify all those other terrible shots. The Rapids don't score often, so any chance we get, we have to take advantage of it. To have one of your assumed offensive leaders pick shots that badly really hurts the team. Gashi is fun to watch, but his poor finishing on top of his poor shot selection is hurting an already terrible offense.

But, hey, the Broncos won the Super Bowl, so maybe offense doesn't matter at all anyway.

Looking at this xG stuff (and watching *almost* every game with my stupid, biased eyes) my ideal attacking midfielders are Hairston, Jones, and Le Toux. Until Jones get back, I do like Gashi in the middle, but he's fourth best in a field of three in my opinion. However, when Jones does get back, I'd like to see Gashi up top. It's not like Doyle is doing any better right now.

This was my first fan post, and I do like playing with expected goals even though I think they're pretty flawed, too. Got any other ideas? Like this? Hate this? Whatever it is, tell me and I'll read it! I promise I'll only cry a little.

Another genius post from a Burgundy Wave reader! Remember that this was written by a fan, not an official Burgundy Wave writer, and the content of this post might not necessarily be an official opinion of this blog's management.