Commentary / Union

On luck and chemistry: Regression to the mean?

Photo: 215pix

Philadelphia Union are the among the hottest teams in Major League Soccer. With all due respect to the Canadian juggernaut that is Toronto FC and the suddenly competent and entertaining Chicago Fire, there is no more stark a 2017 turnaround than what the Boys in Blue have done.

The Union started the season with two draws followed by four losses. The Bizarro Union then earned two draws followed by four wins. It’s a clean bit of mirror-imagery that both Jerry Seinfeld and his Bizarro World counterpart, Kevin, would be interested in learning more about.

Are the Union a “better” team in the last six matches than they were in the first six, and does any statistical narrative bear that out?

This website keeps a log of league-gathered statistics from every match. Picking a few of the most important and more unique ones, plugging them into a rudimentary spreadsheet, and then doing some “first six” and “next six” digging uncovered some intriguing patterns.

  1. The Union are taking more shots, earning more corner kicks, and drawing more fouls in their last six matches than they did in their first six.
  2. They’re passing the ball less and holding less possession on their winning streak than they did during their drought (quite astonishing considering the massive outlying data points in both categories from the Colorado match). However, they are making more accurate passes and more efficient use of the possession they have.
  3. The Union are not winning more duels or more tackles in their victorious second act, but they are more severely limiting the duels and tackles won by their opponents.

There are two narrative paths that begin to present themselves out of these statistical woods.

Narrative 1: Team finds itself

The first one is a story of a team finding itself and its identity, of a group of journeymen, outcasts, draft picks, and national teamers banding together to win as a unit, as a sum greater than that of their disparate parts. The Union are being efficient with the ball and assertive in their possession. They are becoming more cohesive as a unit, and thus supporting one another in such a way that fewer 50/50 duels seem like all-or-nothing challenges.

This story also includes a coach who is beginning to find his voice, using the deepest lineup he’s ever had to incrementally build a winner, one week and one opponent at a time.

Perhaps the most supportive data point for this story is the simple eye test. Most Union fans would agree that the recent version of this team and its coach are doing these things better than the early season version of either, and that players’ positional preferences and tendencies are being more directly augmented than perhaps they have ever been.

Narrative 2: Regression to the mean

The second narrative is more sterile and eminently less inspiring:

The Union are an average or slightly above average MLS side and, after underachieving their expected points per game for six matches, have regressed toward their mean by overachieving.

They had more shots, corner kicks, and fouls won in the their second six matches because they had one more home game in that group than in the former. MLS teams play far better at home than on the road in addition to referees awarding home teams more favor than they do away teams.

Regression to the mean also accounts for luck, which the Union had little of to start the season: David Villa’s audacious goal from midfield, which Andre Blake stops ninety-nine times out of one hundred, or Fabinho’s failed bicycle clearance against Portland that didn’t decide the match but certainly sealed it. They have had a lot of since then: Caleb Calvert’s surreal and absurd double yellow card or the multiple, inches-wide chances Red Bull slid past Andre Blake.

Narrative 3: The likely truth

The truth is probably somewhere in between.

The Union have played better soccer in their winning streak because players have been put in spots where they are comfortable and where they are able to maximize their ability to contribute. Individual performances that leave the greater unit exposed have been minimized through patient but deliberate squad rotation and, for what seems like the first time in the team’s existence, timely and issue-focused substitutions. This is a credit to both the individuals and the group as a whole.

The Union have also been on the good side of favor in the past few weeks.

Two examples include C.J. Sapong’s first goal against Red Bulls, which was assertively taken in ways that fans of the team have wanted for years, but it was also a savable shot that was fortuitously deflected around one of the best keepers in the league. Similarly, Joao Pedro’s low driven strike during the team’s game against the Galaxy hit both posts and bounced back into the field of play, directly into the waiting arms of Andre Blake.

Without any counterfactual to observe, there is no way of knowing how the team would have responded if those moments had gone differently. Further, none of this is meant to diminish the team’s recent results, which have been stellar.

Rather, it is simply to point out that so much of the outcome of a low-scoring game like soccer is predicated on luck, and as one wise man put it, “It’s hard to detect good luck. It looks so much like something you’ve earned.”


  1. Just wondering, who on this team would you describe as a journeyman?

    • hobosocks says:

      I know this question is for Chris, but here is my take. I think of a journeyman in sports as someone who is experienced and competent but who has not and is not likely to achieve true mastery. Fabinho is the first name that jumps out at me to meet this criteria. Gaddis fits in there as well. Creavalle, maybe, in his limited role (Carroll would have been a perfect example a few years ago). I’d add CJ to that group. Maybe even Haris, considering the level he’s generally played at.

      • Thats fair. I take journeyman to be more negative. As in that mediocre player who jumps from team to team and is nothing more than depth.

      • For me it was more that this is Jay Simpson’s 8th club, Haris’s 8th, Ilsinho’s 5th, Fabi’s 10th, and Gooch’s 12th. Nothing negative, but that was some journeying.

  2. I think this piece is spot on. And I think the Union are, in fact, an average to slightly above-average MLS team, building some cohesion due to exactly the factors you have mentioned. Now if Earnie lands us a #10 during the transfer window…

    • OneManWolfpack says:

      Agreed. If this team is enhanced by a true #10, they become a notch or two better. A solid 3-4 place finish is possible, instead of the 5-6 spot.

  3. Adam Schorr says:

    There is definitely some regression to the mean, but regression doesn’t explain “taking more shots, earning more corner kicks, and drawing more fouls”. That is a tangible change. Regression to the mean would see “all metrics are the same except we’re scoring on more shots and allowing on less”. So while there may be some regression, I would not consider it a good explanation here.
    I also don’t think “team finds itself” is a good explanation. You would expect to see more gradual progress rather than “we’re the worst” followed by “we’re the best”.
    No, I think the most likely explanation is also the simplest: Jim Curtin (who is still a terrible coach) got his opening day XI so hilariously, hilariously wrong that it took 6 games, multiple injuries and illnesses, and his best player openly complaining to the media to end up with a workable XI. Playing Bedoya at the 10 was never going to work. Playing 2 wings and a striker that all lacked in speed was never going to work (it is no coincidence that the offense ticked up after inserting Fafa). Playing 2 CBs with no ball skills was never going to work. Only two players right now are playing the same role they played on opening day (Blake in goal and Fabinho at LB). Everybody else has either changed position or wasn’t starting at the start of the season. Kinkead had a good article about Curtin “doing what he hasn’t done before”, but reading through it, there was a lot of “Curtin’s hand was forced due to the player being unavailable.” Curtin lucked into a good starting XI. Making basic fixes to problems that never should have existed turned this team around.

    • “They had more shots, corner kicks, and fouls won in the their second six matches because they had one more home game in that group than in the former” was in this article.

      You’re incorrect in your assumption that luck does not affect stats except for goals. An unlucky red card will see a team’s stats deteriorate across the board. An unlucky goal-against may cause a change to more desperate tactics, also causing stats to deteriorate across the board.

      Also, in “Only two players right now are playing the same role they played on opening day (Blake in goal and Fabinho at LB)”, you seem to have skipped Onyewu, Pontius, Medunjanin, and part of Ilsinho (since Ilsinho only switched roles 2 matches ago)

      • Adam Schorr says:

        Onyewu moved from RCB to LCB. Pontius moved from LM to RM. New roles for both. Medunjanin moved from the 8 to the 6, new role for him. Ilsinho didn’t start opening day.
        Also, rather than going with “there may have been an unlucky red card”, a simple “6 games isn’t much of a sample” is probably a stronger argument, but I’ve already written about how soccer stats basically don’t exist. Given what we have, no, an “unlucky red card” (or a stupid one, as the Union benefited from) won’t have that kind of outsized impact.

      • I don’t get your point. I provided two counter-examples to demonstrate your statement (about stats needing to stay the same for regression-to-the-mean) was wrong. I wasn’t suggesting a red card had anything to do with Union stats, only that one could.

        As I pointed out, you seem fixated on goals/wins being the only lucky stat, but fail in recognizing that many available stats are also heavily influenced by luck

    • What was wrong with the opening day roster. Fafa, Creavalle, and Ilsinho were hurt. Alberg was so outta shape he was out too. Honestly what was he supposed to do?

      • Adam: In your posted link: “Occam’s razor is not considered an irrefutable principle of logic or a scientific result”

      • Adam, re: Occam’s: Not saying the lineup changes weren’t the most probable explanation for the change in form, just that your narrative for when/why they occurred confirms your belief that “Curtin is a terrible coach.”
        Surely, some of the changes were due to injury, but, as pointed out above, so were the choices he had to make for the Starting XI in Week One. Fafa and Roland were both not fit, and there was probably pressure to start the $500k English striker signed in the offseason.
        Pretty sure if he started Jack Elliott and Gooch together on opening day and sat Keegan, you would have said that he was out of his mind and terrible at his job then too.

      • el Pachyderm says:

        Was Jack Elliot even on the opening day roster?

      • Tim Jones says:

        I assume, El P., that you mean the gameday 18?
        He was on the Union roster I think. I seriously doubt he was on the gameday 18 for Vancouver.
        He got a start for the Steel at #6 in their second game, April 8 or thereabouts, and was brilliant there.
        On March 25 he was the only Union player to go the full 90 in the scrimmage-not-called-a-scrimmage against the Steel.
        His first appearance for the Union was when Marquez went down with concussion just before halftime, against whom I forget.

  4. Essentially I agree with Adam’s analysis. The lineup is different since the streak. It resembles the one we anticipated during the off-season and low-and-behold it’s winning and attractive.

    It’s different than the team JC trotted out on opening day that looked like hell, were losing, and made me sick yet again to be a Union fan. The current turn-around seems despite JC rather than because of him. Like Ernie stepped in and said this is what this player is here to do for every transfer since he’s been with us. My .02.

    • Your 2 cents are worthless because you are literally basing it on what you want to believe. The first game was a west coast trip that we played conservatively in and got the draw. No one was complaining about that result at the time.

  5. el Pachyderm says:

    I love the debate up top….
    honorable, respectful and both chock with strong and likely valid points. As with all things its likely the truth resides somewhere in the political middle.
    Would love to know Curtin’s opening day roster had everyone been fit.

    • My guess is:


  6. The Chopper says:

    There’s a bit of all of the above that is likely the truest answer. But simply, moving Bedoya and having him and Medunjanin find a comfort zone and expand their roles together is the biggest factor in the turnaround.

    Add in the good fortune of Alberg’s injury allowing Ilshino to move central and time allowing Gooch to round into servicable form and you have a team that is performaing at a higher level.

    They right now are about where they should be. Just around the final couple of playoff spots. That’s where we expected them before the season and unless there are upgrades, it is where they will finish.

  7. scottymac says:

    I love the narrative that Alberg was penciled in as a starter after riding the pine so much last year. That’s an interesting revision. Curtin doesn’t like one way players. I don’t buy that Alberg was so unfit he couldn’t play. It was when he committed to tracking back on D.
    This just seems absolution for Curtin for getting Bedoya at the 10 horribly wrong. They’re still a mid table ish side

Leave a Reply

Your email address will not be published. Required fields are marked *