Backpass: Things that make you go hmmm
Here's four statistics about the Rapids that explain who they are - and why they've been so successful to date. Hat tip to American Soccer Analysis for some of this good stuff.
With 21 games down and 13 left to go, the Colorado Rapids are officially ‘over the hump’, with three-fifths of the season squarely behind them. Three-fifths is the point in a movie at which you pretty much know what you’re going to get; if it was bad, you’ve already turned it off by now, or you’ve said to yourself “This is pretty bad… but I need to know how it turns out.” If the movie is a romantic-comedy, then you’ve hit the ‘boy loses girl’ point of the film, probably. You’re steaming towards the finally bright and cheery upswing. If it’s a Marvel movie, this is right where a secondary character that you really liked dies.1
Colorado is having an epic season. Not to jinx it, but 2021 is almost certainly going to be remembered in the same breath as 2010 and 2016. The team is very likely to make the playoffs, and well positioned for a decent run in those playoffs when they get there. And they’re certainly still in the mix for Supporters Shield - as of today, New England sits atop the heap on an impressive 52 points, but Colorado are third with 41. So ‘in striking distance.’
It’s good to zoom out once in a while and - much like noticing the plot pacing of all the Marvel movies - look at some broader team trends. This is especially useful to do using numbers that either point out something we haven’t noticed, or reinforce something that we might have seen with our eyes repeatedly up till now. Here’s what I’ve picked up.
1) Home. Away. Doesn’t Matter to Robin Fraser and Crew.
There are teams in MLS that follow the normal pattern of being better at home than on the road. In the Eastern Conference, 12 of the 14 teams are better at home than away. Only the worst two, Cincinnati2 and Toronto, are better away, and not by much. In the Western Conference, 9 of the 13 teams are stronger at home. The exceptions are Seattle, Kansas City, San Jose … and Colorado.
The Rapids have a nearly identical 6-1-3 (WLT) record at home and 6-3-2 record away. Curious.
What possible reason could explain this? The most plausible is, of course, altitude. The Rapids gain a significant advantage by training at 5280 feet, while their opponents experience the disadvantage of having that much less oxygen emanating from their red blood cells.And when the ‘Pids hit the road, they retain that higher O2 count while down the hill or on the coast. It means that, as long as the fitness and training crew finds the right balance, and the tactics force opponents to run, Colorado will always have a slight advantage over their opponents, as long as the talent is equal.
But altitude can’t be the only reason. Because of course, the Rapids have had years where their home-away records varied widely. In every year except two3 of the past decade, they were much better at home than on the road.4 And it’s not like we just discovered we were really high up this year.
So a better answer than altitude is probably Robin Fraser. The tactics he brings on the road are better suited for road games. Or maybe - there aren’t significant road/home tactical adjustments. Maybe the reason the Rapids are so consistent this year is that the approach to games specifically does *not* take the home/away factor into account. Many MLS coaches employ a ‘win at home, draw on the road’ mentality in the way they set up. Clearly either Robin Fraser does not follow this dictum, or if he does, he’s a miserable failure at it, since he seems to win on the road an awful lot.
2) The Rapids are over-performing their Expected Goals Against. By a lot.
Expected Goals and Expected Goals Against are stats that, in regards to small differences, seem to indicate the provenance of luck. Sure Cole Bassett seems to be ‘unlucky’ and hit the post a lot. We say that because Bassett has recorded 44 shots (11 On Target, 3 Goals) this season. His 25.0 percent Shots On Target rate is low for this team and it might lead us to conclude he’s a bad shooter. Except that if just 3 of those post-toinkers rattled in instead of out, suddenly he has a 32.2 percent SOT rate, and his Goals minus Expected Goals (G-xG)5 would be a lot better than it is. As it stands, he has the worst G-xG rate for a midfielder in all of MLS, at -2.85. If three goals bink in off the post, that number becomes +0.15 and we aren’t having this conversation. So on a small scale, we can’t tell if Cole’s struggling, or just snake bit.
But on a team-wide scale, the sample size tells us that luck has less of a factor.
That’s why this other weird numerical phenomenon is so interesting: the Colorado Rapids are wildly over-performing their Expected Goals Against. In laypersons terms, that means they are conceding far fewer goals than would be expected from any other team that faced the shots that they have faced. And it’s winning them games.
The chart above shows that, in red, they have conceded 19 goals to opponents this year. But the Expected Goals Against numbers indicate that they ought to have conceded 25.91 goals, a difference of +6.91 goals.
They’re not the only team that have conceded fewer goals that xGA predicts. DC United has outperformed their xGA by +5.65, and Philadelphia Union have outperformed their xGA by +9.52. Meanwhile LAFC has underperformed xGA by -11.07, and Toronto has underperformed by -7.92.
It means that the Rapids defense has been very good. Expected Goals calculates the odds of whether a shot taken will go in, but it doesn’t account for where defenders are standing when those shots are taken. So a defender tightly marking his man, or a header that’s well contested or a shot that blocked outright will of course be a more difficult shot and will be more likely to skew off target. The Rapids do this more than pretty much every team in the Western Conference.
3) William Yarbrough is one of the best GKs in MLS this year.
Another factor that is helping the Rapids in defense here is the exceptional play of William Yarbrough. Yarbrough has conceded 19 goals, against an Expected Goals Against of 20.95, a difference of -1.95.6 That’s the sixth-best rate amongst all MLS goalkeepers with more than 500 minutes played.7
The defense is good, and Yarbrough is good, and that means the numbers I circled above in blue are good. If the Rapids defense and goaltending perform exactly as the model would expect, the team has 30.27 points to date - good enough for a three way tie with Portland and RSL for sixth in the West. Instead, they’ve been 10.73 points better than that and sit in second - one point behind Seattle.
Maybe they’re due for a little regression here. But the overall trend line here indicates a team that keeps its players in position to get back in transition and defend in the final third and on set pieces, and that’s what is pushing the Rapids to succeed this season.
4) Go around. Not through.
The last number I found that made me say ‘huh’ out loud is this one: the Rapids are 25th out of the 27 teams in Major League Soccer at Dribbles Attempted, with 308. The top team in MLS in Dribbles Attempted, LAFC, has 500.8
And it’s not because they’re afraid or that the Rapids aren’t likely to put it on the grass and turn their man - they dribble well. They’re 5th in MLS in Dribble Success Percentage, with a 61.7 percent rate.
The Rapids under Robin Fraser would simply much rather pass the ball around you and into the final third or the 18-yard box than dribble it in.
This follows an old soccer coaching axiom - a passed ball moves faster than you can run. But it’s more than that. The Rapids style of play is to move the ball around, shift the defense, move the ball around, shift the defense. The only player who really violates this rule is Michael Barrios, and that’s because he’s the guy they bomb it to that keeps everyone else honest - the outlet pass they have if the pass-and-move formula is getting interrupted. And yet Barrios is still frequent to cross it or cut it back after his big dribble. More than any other team in MLS except two (NYC and Nashville), the Rapids go around and not through.
MARVEL MOVIE SPOILERS OF PEOPLE WHO DIE ROUGHLY 3/5ths OF THE WAY THRU THE FILM BELOW:
Dr. Strange - The Ancient One
Ant Man - Janet Van Dyme (she doesn’t per se die, but at roughly the 3/5ths point we learn the details of her disappearance into the quantum realm. I’m counting it.)
The Avengers - Agent Coulson. Which still pisses me off. Clark Gregg is a damn national treasure.
Avengers: Age of Ultron - Pietro Maximoff (Quicksilver). It’s kind of nearer to 4/5ths, but also this movie was disappointing so whatevs.
Avengers: Infinity War - Gamora (although Heimdall dies at the beginning and a dozen characters die at the end, so this movie messes with my thesis a bit.)
Avengers: End Game - Black Widow. Not a secondary character, but since this is the ‘big finale’, they had to kill a main character. Had to.
Thor - Laufey. (I just realized I think this is the only Marvel movie I haven’t seen. Which - how?)
Thor: Ragnarok - I think nobody? Because Taika Waititi cares not for stupid conventional movie structures and such.
Obviously there are a bunch of other Marvel movies where this 3/5th death thing doesn’t happen. But it’s pretty common. I’ve cracked the code.
Cincinnati are winless at TQL Stadium, which would seem like a huge deal except they have just 3 wins total.
In 2015, they were roughly equally bad at home (5-7-5) and away (4-8-5). In 2020 … the numbers don’t matter because 3 of the games were played in Orlando and the season was truncated for Colorado by Covid, so we can deduce nothing.
In 2017 they were 8-5-4 at home, 1-14-2 on the road! That’s… that’s… what? How?
G-xG tells us what whether a player is scoring the goals at a rate above expectation (+), at expectation (0) or below expectation (-).
You might notice the xGA number for Yarbrough of 20.95 is different than it is for the Rapids team of 25.91. That’s because ‘Team xGA’ and ‘Player xGA’. ASA explains that here, if you want to really nerd out hardcore. It’s mostly down to differences in the values of penalty kicks and sequential shots.
Andre Blake leads the league with a -6.32 G-xGA. Matt Turner is second with a -3.70 G-xGA, so this math passes the observation test. Those two goalkeepers are good.