clock menu more-arrow no yes

Filed under:

Don’t Freak Out About FC Dallas (yet)

New, 13 comments

The computers like FCD more than you may realize – hopefully it’s not too late for them to be right.

MLS: Real Salt Lake at FC Dallas Kevin Jairaj-USA TODAY Sports

I get it. It looks bad. FC Dallas has six points from the first seven games, 0.86 points per game, a rate which would be in the neighborhood of the Wooden Spoon in most years.

The club’s marquee $3 million striker is struggling badly to score from open play. All three rostered goalkeepers are injured to one extent or another. The defense keeps making bizarre unforced errors. The two new starting wingers seem allergic to the opposition goal. The head coach is coming under increasing scrutiny for his tactics and lineup choices. All this while ownership plays their violins and sells exciting prospect after exciting prospect overseas before they play real minutes with the first team.

I get it. It’s not a fun time to be an FCD fan.

BUT – FCD has been WAY better than their record would suggest in ways that matter

Per ASA (from whom I’ll be getting all the data in this piece unless stated otherwise), FCD has picked up a little more than half a point less per game than they would be expected to based on the shots they have taken and conceded. In other words, expected goals (“xG”) suggest FCD has gotten pretty unlucky so far this year. In case there’s anyone out there protesting with Bill Parcells that “you are what your record says you are”, I’ll remind you that past goal differentials are better predictors of future results than past records, and that expected goal differentials are better still.

In fact, by non-penalty xG (“npxG”), FCD has not only been very good: they’ve been balanced. FCD ranks in the top-10 in MLS in 2021 in both npxG for and against. The only teams in 2021 with a better simple sum of ranks (FCD is sixth in npxG for and ninth in npxG against, so the simple sum is 6+9=15) are NER (2+10=12), NYC (7+1=8), NSH (1+2=3), and LAFC (8+5=13).

It’s not all that common to be top-10 in both metrics. 31% of MLS teams in ’13-’19 finished the full season with such a record. 84% of those teams went on to make the playoffs with an average post-season seed in-conference of 2.8. A quarter of that 31% was the top-seeded team in their conference.

Teams that put up numbers similar to FCD’s so far this year are overwhelmingly playoff-caliber teams.

BUT – xG reversion requires at least average finishing skill and FCD doesn’t have any of that!

There are hundreds of examples of people who thought the exact same thing about a particular player or team (or the opposite: that a given example was the positive outlier) and were proven wrong in time. Using out/underperformance of xG to predict future variances from xG is a fool’s errand in the vast majority of cases. It’s that old proverb: chance creation tends to recur, but chance conversion does not.

As a reminder, xG models freeze time as the shot is being taken, judging its chances of being a goal by the shot’s location, the type of shot, the speed of the attack to that point, the action that led to the shot, etc. There are two factors past that point we can look at to bolster our understanding of why a team might or might not score: the quality of the shot, and the defense and GK’s attempts to stop it going in. We can measure the former by the difference between standard and post-shot xG models, and the latter by the difference between post-shot xG models and actual goals scored. Results of that analysis below (all data in this section from FBRef and includes PKs). Big takeaways:

  1. FCD’s finishing has lowered its chances to score by about 0.2 goals on aggregate in 2021. Before you say that proves Jara and co. can’t finish because the number is negative, note that it’s good enough for tenth in the league. As the magnitude of 0.2 would suggest, FCD has been an average finishing team in 2021 by this measure.
  2. Surprisingly given all the injuries to the FCD GK corps, Maurer and the Maurerettes have still put in great work keeping the ball out of the net, lowering the opponent’s chances to score by about 2.3 goals on aggregate, fourth in the league behind the Galaxy, the Sounders, and the Union.
  3. The real story here, in my humble opinion, is how well FCD’s foes have played so far this year. If “FCD’s Opponents” was an MLS team, they’d rank third in the league in how much they have added to their chances of scoring with their finishing (1.7 goals) and fourth in how much they have lowered their opponents’ (FCD’s) chances of scoring with their shot-blocking (2.8 goals). I’d expect both of those to average about zero going forward given the nature of “FCD’s Opponents”.

Why is FCD underperforming their chances? It’s more to do with the herculean efforts of their opponents relative to league average than it has to do with FCD’s shortcomings.

BUT – xG doesn’t see the whole field!

It doesn’t, but the non-shot models like FCD too. G+, ASA’s metric that combines every on-ball touch into a single number, says FCD so far in 2021 is worth about 1.4 goals-for per game and 1.2 goals-against per game, with their 0.2 G+ differentials twelfth in the league and sixth in the West, again painting the picture of a solid playoff team, and G+ outperforms xG in predicting future performance.

Another way I’ve found recently to assess a team’s play outside of its shots is what I call the “Misery Index”. The Misery Index is a passing metric: if your team is better than average at doing what you want to with the ball and better than average at preventing the other team from doing the same. I call it the “Misery Index” because teams with high scores tend to be miserable teams to play against: they suffocate you and elude your attempts to return the favor.

Turns out the Misery Index is a really strong predictor of team points-per-game, and teams with similar scores as FCD (about +2.9) tend, on average, to accrue points at rates that will almost always get you into the playoffs. Very few second quartile teams (FCD’s bucket) have been as bad over the course of a season as FCD has so far in 2021.

BUT – what about those PKs?!?

Yes, we’ve been referencing non-penalty xG so far. Obviously, PKs change matches, as they have for FCD so far this year. You shouldn’t just exclude them when deciding if a team has played well. That said, they also don’t happen all that often, seem to cluster randomly, and tend to cloud how teams perform in the run of play.

But we still need to address them, because if FCD keeps giving them up at this rate, there’s no chance for this team to move forward. There’s not an agreed-upon statistical norm for what “PK Concession Rate” should mean. While some people might use PKs conceded per game, I’ve chosen to look at PKs compared to touches that you allow your opponent to have in your box. The basic thinking is that teams that spend less time defending in their own box should concede fewer spot-kicks. And indeed, by that measure (and basically any measure), FCD is an outlier to a huge degree. So far in 2021, they have conceded a PK for every 37 opponent touches in the box. For context, the next lowest such rate in FBRef’s data set is TFC in 2019 with 55. The MLS average is ~145, which means FCD has conceded almost exactly three more PKs in 2021 than the average MLS team would, given how much FCD defends in their own box (about average). This team has been killed by PKs so far this season, and yet that doesn’t mean they’ll continue in that fashion.

There is some reason to believe PK concession rates don’t recur in the same way as xG over/underperformance. For example, there’s effectively no correlation year-on-year (see below), which suggests it’s more a matter of luck rather than skill. That said, there’s not nearly as much literature on this topic, and I can’t shake the image of Jader Obrian tripping Cristian Espinoza in San Jose from my brain. That’s a bad decision, not bad luck. Maybe we just have to believe that FCD will make an average amount of bad decisions going forward.

STILL – mean reversion going forward doesn’t erase the issues thus far

Analytics website 538 thinks, despite the early season struggles (which they weigh heavily in their model), FCD is still the eleventh best team in the league and fifth-best in the West. By the way, those models take way more into account than we have discussed here. Yet, despite that status, 538 also thinks FCD only has a 42% chance of making the playoffs because their results have lagged their performances so badly. Oh, by the way, FCD’s two inter-league opponents are New England and NYCFC, who might be the two best teams in the Eastern Conference, which contrasts with Houston, for example, who get to play Cincinnati and Montreal.

While mean reversion says that, over long periods of time, data tend toward the mean, that doesn’t mean that FCD is somehow “due” for some good luck to reverse out all the PKs and the finishing issues from the first seven games. That’s the Gambler’s Fallacy. Instead, we should predict average luck for FCD.

Take the PKs, for example. Mean reversion doesn’t mean that FCD’s dot above will most likely end up on the 50% line by the end of the season. It means that FCD’s dot will most likely end up on a line with the same slope as the 50% line extending from where FCD’s dot is today. Even with average luck going forward, FCD is still likely to have a really bad season of PK luck because it’s been so bad already.

The same logic applies to the standings: although, based on the above discussion, I’d expect FCD to accrue points at a rate of 1.5 or so going forward, that doesn’t mean they will end up with 1.5 PPG at the end of the season because they’re so far below that right now. A 42% chance to make the playoffs sounds about right.

AND STILL – this doesn’t mean FCD shouldn’t make personnel or tactical changes

Club leadership has been very clear that they have ambitions to be a team that earns home playoff games. While I’ll say FCD’s a playoff quality team until I’m blue in the face, I think it’s a stretch to say they are all that close to the SKCs or Seattles of the conference. Players like Jara, Obrian, Vargas, Munjoma, Tessmann, and the center backs need to either step up their games or find a seat on the bench. With Pomykal and Ferreira returning from injury soon and a whole bunch of money to spend this summer (remember that Reynolds’ $10+ million actually goes through in the coming months), Luchi Gonzalez and Andre Zanotta should be holding starters to a higher standard of performance than ever.

There’s too much ground to make up not to be ruthless. Making the playoffs is table stakes at FCD – we’re not perpetual cellar-dwellers like the Dynamo. In many ways, FCD has played to that standard so far this year. To meet it, however, they have to exceed it going forward.