MLS 2020 According to g+: The Overperforming, the Underperforming, and the Ugly, Part 1
/by Mark Asher Goodman (@soccer_rabbi)
Ever since the first coach had the first microphone stuck in his or her face, or the first pundit expounded upon their local sports team, folks have opined that “we were better than the results” or “we should have won that game.” And for a long time, you would pretty much have to take their word on that.
But the new g+ metric, and the even-newer aggregate measure of g+ that ASA’s Mattias Kullowatz (twitter: @MattyAnselmo) and John Muller (@johnspacemuller) rolled out last week has given us a tool that lets us actually say with real certainty that a team is better, worse, or exactly what the win-loss results show. In other words, g+ is a giant neon sign that blinks “Regression Ahead” or “We wuz robbed, again.”
To test the metric, I put all the fine ASA math into a table and produced four different grouping of MLS teams: ‘Overperforming’, ‘Underperforming’, ‘Performing as expected - Good’, and ‘Performing as expected - Bad’. I’ll dig into the numbers a little bit and hypothesize why some teams are the way they are. Note that all the data is current to Tuesday, October 6 - none of the Wednesday matches after this week are part of it.
Part 1 will start with teams whose performance generally lines up with what the team g+ model would expect of them. There are 15 of these teams, broken up into two extremely scientific subgroups: ‘good’ and ‘bad’. Part 2 on overperformers will drop next week (darn it, I’m going to have to update all the numbers). Part 3 will include the underperformers, a fitting end since it’ll go out on a note of hope for the teams that have about a quarter of the season to play and have yet to actually maximize the results based on performance.
This is a relatively Captain Obvious statement, but before reading this, you probably ought to understand the basics of what g+ is and what it measures. The quintessential primer is here - it includes fantastic infographics. The explanation of how team g+ is different than player g+ is here. And to see my first swipe at thinking about team g+. before John Muller actually did the legwork to make the math come together, read this, and skip to section 3.
I’ll keep you guys in suspense for the really important stuff, though, and start with unpacking the team g+ numbers that track neatly with actual team win-loss performance. Because all of MLS is outta whack in terms of games played, we will use Points Per Game as a proxy (PPG).
Performing as Expected - Good
Four teams in MLS are finding success while also performing pretty much at the level their underlying g+ would anticipate
Four teams in MLS are finding success while also performing pretty much at the level their underlying g+ would anticipate (“but rabbi” - you say - “there are five teams on this chart.” Yes yes, I’ll get to that...)
Toronto FC are second in MLS in team g+ with an expected 8.68 g+ differential. A lot of that is on the strength of their defense, as they have allowed a third-lowest in MLS 12.20 g+ against. And although Alejandro Pozuelo is the sexiest beast on the pitch with his fantastic close-quarters dribbling and silky passes, the numbers tell us that it is TFC’s strikers, young Ayo Akinola and veteran Jozy Altidore (1.77 and 0.60 g+ added above average, respectively), and wide players Justin Morrow and Jonathan Osorio (0.46 and 0.60 g+ added), who are actually the primary contributors to this team’s success.
Anyone that’s watched Seattle knows that they are, once again, an elite team in MLS this year. Not only that, but their strikers are taking an already league-best offense in terms of chance creation (according to g+), and adding to that above-average finishing. They have 5.21 goals more than the ASA Expected Goals model would predict. Even if they regressed back to the mean in terms of finishing, they’d still produce more and concede fewer goals than the rest of the Western Conference.
Sporting Kansas City are flying under the radar and quietly putting together another good year. Despite a few hiccups here and there, they have one of the top g+ differentials in the league. You might argue that having the 8th-overall PPG would mean they’re even underperforming a little. One reason the model likes them so much is that they’ve only been blown out once this year, a 5-2 drubbing against Houston. They’ve played in 9 one-goal matches, and drawn 2 matches. The individual player g+ numbers also point to a lot of depth - they have 17 different players contributing 300 or more minutes, and most of them are around average contributors or better. There aren’t any glaring weaknesses, so Peter Vermes can rotate the squad or put on subs and know that there’ll be little dropoff.
NYCFC are performing well-ish, although that negative G-xG number should give us some pause. Additionally, the loan of Alexandru Mitrita to Saudi club Al-Ahli might sting a little; the Romanian winger produced 0.58 g+ Above Average for NYC, 5th best on the team. The obvious replacements, Ismael Tajouri-Shradi and Jesus Medina, have been good for a -0.15 g+aa and -0.77 g+aa - only four wingers in MLS have a g+aa worse than him.
Finally - why is Colorado on this list? Well, they’re performing as expected, so they didn’t belong on the forthcoming ‘over’ or ‘under’ performing lists. Of course, on the other hand, 1.46 PPG, good for just 8th in the Western Conference, isn’t exactly setting the world on fire, now is it? Their Team g+ of 0.61 is closer to average than any team in the league except FC Dallas. Their PPG lines up with that as well. And because I don’t think it’s smart to create a table with only one team in it called ‘Results are meh’, I stuck ‘em here. Colorado’s biggest problem has nothing to do with math, though: it’s health and wellness. They’ve had 15 positive Covid results in the past week for players and staff. They’ve postponed three games, and it looks as though their latest delay will require them to play a regular season game the first week of the MLS playoffs. In addition to that, they’re looking at playing a Wednesday-Saturday, Wednesday-Saturday set of games every week from now till mid-November. And that’s if they can even field a healthy team this Saturday. If they can’t, we’re moving into uncharted ‘Covid-forfeit’ territory.
Performing as Expected - Bad
None of these teams are what any sane person would call ‘good’. However, we must note briefly the insane chaos that MLS 2020 has been - what with the TV graphic overlaid tarps which long balls occasionally “disappear under” and fake crowd noise and folks on the sidelines wearing masks and every freaking team and their mother making the playoffs this year plus kneeling for racial justice and all of us now accepting all of this as normal when MY GOD NONE OF THIS IS NORMAL. NONE OF IT. 2020 is the scene in “Willy Wonka and the Chocolate Factory” with the boat ride which used to terrify me as a child - my God, in the middle of it Veruca Salt’s father blurts out ‘You can’t possibly see where you’re going, Wonka!’ And he calmly responds ‘You’re right. I can’t.’ That’s this year for everything, including soccer.
So, part of this fateful boat voyage includes the reality that some really bad soccer teams will be in the playoffs. If the playoffs started tomorrow, five of these teams (NYRB, Nashville, Montreal, Atlanta, and San Jose (!!)) would be in. So if I put your team on the bad list, and they also make the playoffs or suddenly tear through the playoffs and hoist the cup, kindly miss me with your ‘I told you so’ hate mail.
NYRB are weirdly inconsistent - they’ve lost to bottom-feeders Cincinnati twice and DC United once. I don’t have a really good explanation for what’s going on there, although half of their regular starters on the backline, Tim Parker and Jason Pendant, are well below-average in their g+ contributions, at -0.51 and -0.47, respectively.
Gary Smith should be lauded for what he’s accomplishing in Nashville. Although it isn’t particularly uptempo or exciting soccer - eight of Nashville’s 15 matches have ended in a 0-0 or 1-0 scoreline - it gets the job done, as Nashville are in a playoff position in their first year in the league. Their 4-2-3-1 setup kills everything in the midfield - and as a result, defensive midfielders Dax McCarty and Anibal Godoy are among the top 12 amongst MLS d-mids in g+. That finishing though: a -5.48 G-xG. [Shudder.]
Remember when Real Salt Lake’s Corey Baird was really, really good? It was 2018, and the exciting, fresh-faced winger picked up the Rookie of the Year trophy on the strength of an 8 goal, 5 assist season. That led to non-stop hype from the MLS punditocracy, and earned the kid his first USMNT appearance the following January. Yeah, about that. Baird’s been awful this year; his g+ of -1.58 is 2nd-to-worst amongst MLS wingers, and 6th-worst amongst all players in the league with 600 or more minutes. The specifically putrid aspects of his game are a -0.36 dribbling g+, -0.37 passing g+, and a comically bad -0.71 in receiving g+. There are other things going on at RSL, like the chaos around their racist owner being forced to step down, and the fact that their only win in the past 6 outings was against the woeful LA Galaxy. Baird’s just one of several problems at RSL right now. But you can’t really be called a good team if one of your key players is having a very bad year.
The Josef Martinez-less, post-Frank DeBoer era in ATL is essentially a holding pattern for Atlanta United. But even without Martinez, even with DeBoer playing joyless, dire, low-risk defensive football, did you ever think the electric five stripes would go from winning MLS Cup in 2018 with the attack of Miguel Almiron, Julian Gressel and Ezequiel Barco to sitting third-to-last in all of MLS in both g+ Added For and Expected Goals? Making the playoffs is not the goal for a team with this level of expectation - a team that holds all ten of the top ten best-attended MLS regular season matches in history. We are all awaiting the first steps in the rebuild, and they haven’t happened yet.
Montreal very nearly got thrown into the ‘overperforming expectations’ grouping, which is pretty wild considering their 5W-8L-2T record. But look at their g+ differential and you’ll see that it’s one of the worst in MLS. Maybe the reason is that Impact are overperforming expectations in G-xG by more than any other team in this group - with 22 goals on just 18.37 xG. Chalk a lot of that up to Romell Quito’s 6 goals on an xG of just 3.49.
Exhibit A:
Exhibit B:
San Jose were a train wreck inside a dumpster fire, and all of a sudden they’ve won three straight. What gives? I’m not sure. But g+ seems to think their defense is better - expected to concede only 22.02 in g+ - than one that has actually conceded a league-leading 40 goals against. If you’ve watched them get eviscerated by teams like Colorado, Portland, and Seattle, you might think g+ doesn’t fully appreciate the exploitability of a man-marking system.
G+ thinks Houston is better than just a paltry 4 wins in 15 tries, but not by much. Now that Alberth Elis and his team-leading 1.93 g+ is gone, team g+ might sing a different tune over Houston’s final 7 games.
Vancouver. Bad in 2018, but with Alphonso Davies! Bad in 2019, nothing of note to speak of. And, once again, bad in 2020. The Whitecaps defense hasn’t kept a clean sheet in three months. I can’t tell you what the identity of this club is under Marc Dos Santos, or in general, other than to say Vancouver ‘are rebuilding,’ once again.
All you need to know about Cincinnati is that g+ thinks they should have produced a league-low 9.31 goals, and yet they can’t even achieve that miserly sum - they’ve found the net just 8 times this season. From August 21 till now they have just one goal, and it’s an game-winning olympico against NYRB. Soccer is weird sometimes, man.
Ben Olsen is out at DC United after 11 years as a player and another 11 years as their head coach. And that’s because they’ve been really, really bad this year. Case in point, Benny’s last game:
That graphic from our friend Cheuk Hei Ho shows that DCU had maybe 5 or 10 minutes where they were making things happen, and another 80 or 85 minutes where NYCFC were running the show. There’s been entirely too much of that this year, and it spelled curtains for Olsen.
So ends part 1. Stay tuned for part 2, coming sometime later next week.