We are but five games into the regular season for most teams, and while that is a very small sample size to analyze, I thought it would be interesting to see where D.C. United is shaping up statistically. As the season goes on, I’ll use this space to examine the numbers behind D.C. United, and hopefully the Washington Spirit. Yes, it’s a Members Only pun. No, I don’t have jackets.
Obviously the most important number for DCU right now is four, the points they’ve earned via an opening week comeback win over Toronto and a draw from behind against Orlando. Perhaps you’ve noticed a theme. Those four points have them 13th in the Eastern Conference in points per match (0.80), tied with Charlotte and just a shade better than last place Montreal (0.75). It goes without saying that the results thus far have been bad, but how are the underlying numbers?
All statistics found on fbref.com or whoscored.com
Attack
D.C. United ranks second in the league with 1.50 xA per match (Expected Assists, measuring the likelihood that any completed pass becomes an assist,) behind San Jose’s 1.60 xA in first, and ahead of Vancouver’s 1.34 in third. The stat can be complicated because it does not consider whether there was a shot taken, instead using the the type of pass and location of the receipt of the pass, among other things. This suggests that D.C. has been quite good at putting teammates in dangerous positions, even if they don’t shoot. Along those lines, United are third in the league in touches in the opposing penalty area with 26.0 per match, behind first-place LAFC (28.5) and San Jose (26.6.).
Contrast this with D.C.’s low-mediocre 1.12 xG per match (Expected Goals, used to measure the likelihood that any shot results in a goal,) tied for 17th out of 29 with the Red Bulls and Toronto, and I can’t help but wonder how a team that is so good at getting into the opposing penalty area is having trouble generating good scoring chances. I believe the answer is in the over-reliance on headed shots, which are less likely to result in goals than shots taken with the foot from the same location. D.C. United has taken the most headed shots in the league, with 4.0 per match, ahead of Charlotte’s 3.4. Christian Benteke is the team leader in this, with 1.8 per 90 minutes, which also happens to lead the whole league among players with more than 90 minutes of play. To put this in further perspective, Benteke is sixth in the league in total shots with 17, nine of which have been headers. The five players above him in total shots (Erik Thommy, Martin Ojeda, Thiago Almada, Daniel Salloi, Luiz Araujo) have been credited with one headed shot combined. No other team relies on headers for shots as much as D.C. United, and it stands to reason that it’s not a reliable way to win.
Defense
D.C. is tied for second worst defense in MLS with Portland, conceding 2.0 goals per game, 10 total so far this season, but the news isn’t all bad. One positive is that the team is first in the league in attempted tackles in the attacking third (3.8), and fifth in attempted tackles in the middle third (7.8). Willingness to contest higher up the pitch can be a recipe for success, as United is second in MLS in combined attacking and middle third tackles per match, between early Supporters’ Shield leaders St. Louis, and defending MLS Champion LAFC.
Where they fall off is attempted tackles in the defensive third, with 6.8 per game, which is 22nd overall, which is easy to link to also allowing the fifth most touches in the defensive third, with 218.4. While it’s good that they haven’t been too permissive of passes into their penalty area, allowing 6.8 per game, tied with St. Louis for 15th in the league, they’re second worst in the league in allowed carries into the penalty area, with 5.6 per game, behind Portland (remember them?) with 5.8. For perspective, the best teams in this stat are NYC and the Red Bulls (1.8).
My conclusion here is that I would like to see the team scale back slightly in challenging higher up the field in exchange for being more willing to challenge closer to goal. They have been reasonably good at keeping the opposing team from shooting, allowing 10.6 shots per game, tied for eighth, but they’re fourth worst in the league in allowing opposing shots on target, 41.5%. Perhaps contesting a just a little more often would lead to harder shots and fewer carries into the area.
Goalkeeping
I think the best statistical measure of goalkeeping is a stat called PSxG+/-, which subtracts goals allowed from post-shot expected goals, which itself measures the likelihood of a goalkeeper saving any shot. Think of it this way: this stat tells you that a team has allowed X goals from Y expected goals, and if PSxG+/- is positive, you’ve got above average goalkeeping. D.C. United has allowed 10 goals on 9.2 expected goals, a differential of -.80, which is seventh worst in the league. Sadly, this is still an improvement from last season, when they were second worst. This is not wholly Tyler Miller‘s fault, because he’s facing 1.84 PSxG per game, fourth worst in MLS, but two of the teams with worse numbers, Austin and Colorado (1.98 each), have allowed just 1.4 and 1.6 goals per game respectively. Those teams are getting good goalkeeping, as Austin leads the league in PSxG+/- with +.58, and Colorado is third with +.38.
Individual Performers (among players with 180+ minutes played)
- Mateusz Klich is 13th in MLS in SCA (Shot Creating Actions, which are offensive actions like passes, dribbles, or fouls suffered, that lead directly to a shot) with 5.2 per 90 minutes. Atlanta’s Thiago Almada leads the league with 7.75. Unfortunately, more than half of Klich’s SCAs come on dead ball passes. Chris Durkin is the next best United player with 3.9 per 90 minutes, 36th in the league.
- Russell Canouse leads the league with 1.33 attempted tackles in the attacking third per 90 minutes, and he’s second in the league in interceptions with 3.11 per 90 minutes, behind LAFC’s Timothy Tillman with 3.33.
- Steven Birnbaum is known in this space for being good in the air. This year he’s fourth in the league in aerial duels won, with 5.37 per 90 minutes. While that number is very good, it’s dwarfed by Christian Benteke, whose 8.40 leads the league.
While not everything looks bleak for D.C. United, I would expect that even these good numbers will start to come down to Earth if results don’t improve. Regression to the mean can be swift once we get beyond small sample sizes like five games, so real improvement in the columns that matter most needs to come soon.
Great analysis. The defensive numbers are particularly telling. It was hard for me to quantify what we’ve been seeing with teams seemingly having more space once they get into our defensive third than they should. Being too aggressive up field makes sense. Nicely done! Looking forward to more of these!
To my eye, Canouse has to be one of the worst mids in the league at passing – he is almost guaranteed to turn it over or make a backwards/useless pass. Any chance you’re able to take a look at the numbers on this? It’s a long simmering frustration and have always wondered how the numbers would show up…
He’s turned more into a destroyer I think.
Hey John!
I can jump in here with some Fbref stats on Canouse to backup your points. Canouse is in the 23rd percentile for progressive passes (passes going forward), completing 3.94 per 90. For passing in general, he is in the 17th percentile for attempted passes (37.8 per 90) and 32nd percentile for completion (78.8% completion rate per 90). So tl;dr, he does not complete many passes in general, much less progressive passes, and his completion percentage is not great on his passes in general. The percentiles basically show how he is in relation to other players. For instance, 23rd percentile for progressive passes means that 77% of other midfielders make more progressive passes than Canouse does. I hope this helps! Below is the link to his Fbref data.
https://fbref.com/en/players/ec730d9a/Russell-Canouse
I think Canouse’s biggest problem has nothing to do with passing. His awareness this season creating a cover shadow. Maybe it’s because the double pivot allowed him a little more leeway, but this has been poor and is likely why the line-breaking passes are such a large problem. DCU look more structured, but they have a huge flaw there that needs to be remedied in their defensive setup.
I did a quick comparison on FBref, and noticed these things about Canouse side by side Aiden Morris, Kellyn Acosta, and Noel Buck:
At a brief look, it seems Canouse is being used less as a circulator this year and more of a forward connector in a comparison of 2022 to 2023. While has 0 xAG (expected assisted goals), the lowest shots created of the group, and 0.1 xG. He does have the second highest successful progressive passes, and highest number of passes into the final 1/3. While he passes long more than any of the other players per game, his completion % is pretty low (60s) but these are lower risk higher reward typically.
Passing-wise, Aiden Morris is having a great year. Acosta is not. Canouse is upper-middle of the group outside of long passes. Progressive passing is around 1000 yards towards the opponents net, Acosta and Buck are in the 700s and Morris is in the 1100s. So when you think about circulators, he’s not a pass back to Birnbaum/Williams player, he’s moving it up field the majority of the time. He’s >80% completion for short and medium passes (5-15 and 15-30), but his long passes are around 60% completion. I don’t have a pass map, but from watching him play I suspect this is switches or long diagonals. With Durkin being a receiver for most of the year, I wonder if that had a negative effect due to speed. Benteke typically receives from CBs and GK in our patterns of play so he doesn’t benefit from our forwards absurd ability to receive.
He’s also excellent in the air as a DM so far (5 of 6 duels won). Noel Buck is pretty abysmal winning no aerials.
All of this said, he’s also 5 games in and hasn’t been hurt yet this season so thumbs up there!
Outstanding stuff! Excellent contribution.
Nice – thanks for digging, super insightful. He’s an absolute destroyer in the tackle department (though has lost his mark on a couple goals this year…), i just get so frustrated after he wins the ball and then turns it over right away… like clockwork.
I guess I think he wants to be a ‘pass the ball up the field’ type player, but maybe he should be trying to get it to klich or somebody better than him at doing so… I just feel like DC has a huge weakness right now (and last year) in the middle of the field in terms of transitions and building an attack. They are a bit of a one-trick pony right now in terms of sending to wings, crossing in to benteke and seeing what happens. Willing to give them a chance now that taxi is back, but need more from the klich/canouse/center mids if we’re actually serious about becoming a credible team that doesn’t squeak out a win every once in awhile by getting lucky late, which seems to be rooneys MO right now.
Excellent analysis, Blasian. I’m looking forward to more analyses explaining how DCU is now racking up the wins.
Great work on this! These numbers are actually fairly encouraging, and suggest the team has a higher ceiling… unfortunately the floor is extremely low. It requires us to have stellar emergency defending to counter the fact that Canouse isn’t hovering above the backline and is playing destroyer at the top of the midfield circle. … and that just hasn’t been the case.
Thanks for this, it takes quite a while to spend on any of these metrics to do a write up so it’s appreciated.
Just like xG, the PSxG+/- is good to use at the end of the season or after a lot of games. Thinking back on the goals, I believe only 1 Columbus, 1 NYC goal, and maybe one other that Miller really had a decent chance to do better and stop. I suspect we’ll see this level out and move positive later in the year.
Okay, I’ve got a bunch of thoughts.
First, kudos for diving in to the data and then presenting it. This is meaty, it’s insightful, it’s good stuff. We need more of this. Very good job. Thank you for submitting this.
Okay, here’s my evaluation of the data and thoughts on Blasian’s excellent analysis.
First, I’m encouraged by the data on tackles in the attacking third and middle third. I’m not so sure this means we need to be less aggressive in those areas in order to shore up the defensive third. How I interpret that data on tackles in the defensive third is not that we’ve been over committed. It’s that we’ve had a combination of defending counters poorly–allowing ourselves to be surprised. And we’ve had some sloppiness (prime example, Durkin being faked out and over committing on the Bou goal in the Revs game. The coaches sometimes “sound” like coach speak but I think it’s accurate in this case. We’re getting sloppy on 1-2 quick plays per game and being badly punished for them. I would be interested to see if there is data on the average amount of possession on all of the goals scored against us (and how that stacks up against other teams in the league). What I “suspect” (but do not know) is that we’re giving up quick goals. We’re either failing to pressure a distributor in our attacking third (who has time to pick someone out) or we’re getting sloppy and allowing someone to get separation or a step and then punish us. I think that’s solvable.
Also, related to this, the “eye test” tells me our outside backs have been bad defensively. Sometimes caught forward, other times (Andy Najar, I’m looking at you) allowing an attacker to gain a step on them or get some separation. I’m wondering what a good metric is to test this factor.
Second, I’m impressed at how strong Benteke is in the air. Yes, I knew he was a beast. But frankly, he’s winning balls and if he was putting more of them on frame, we’d probably have another 4 points. I don’t mean to make this a game of “if,” I’m only saying the data on Benteke gives me an even stronger impression of him then I had with the “eye test.”
Third, I’d need to look at more data before concluded we’re just about crossing it in to the box and letting Benteke do his thing. The data “suggests” that. But I think that’s more of a factor that we don’t have a striker playing off of Benteke who invites through balls and an in the box pass. Taxi is probably that answer, especially given his late runs in to the box after the defense has collapsed on the striker. I like me some TKDP but he’s not the offensive threat that is going to generate all of those non-crossed chances the data says we’re missing. So I think we’re playing balls in to the box on the ground but doing so ineffectively. Related to this, I think in the offensive third, there isn’t a lot of one-touch play (so non-crossed balls aren’t played as quickly and the defense has more time to adjust and intercept/deny the chance). That’s my impression–not based on any hard data.
Fourth, on Miller, he’s clearly not a brilliant shot stopper. My gut reaction is the data and his ranking is more a factor of the defense in front of him than his play. Our defense is mostly doing a good job (so not a lot of xG that he’s facing and having to stop). But we’re letting down at times and the other side is burying those chances. So his score is low. Is he just not a good shot stopper? Or is the data early and too small of a sample? That I can’t say. I do think our trade off right now is getting a good sweeper keeper but maybe giving up some shot stopping ability–that’s my initial take on the data.
Again, good stuff–really appreciate the effort to pull this together and the analysis on what it all means. Please continue.
[…] Numbers Only: DCU’s early returns (The DP)For some more statistics, check out Blasian’s break-down of how the Black-and-Red are performing this season. I look forward to read more of these as the season goes on (and as we get some Spirit stats)! […]