Dearest Sounders at Heart,
Ow. That was an ugly loss. I think other writers here (including you fine commenters) summed up the bad performance and feeling of despair that’s on all of our minds right about now. The Sounders aren’t looking good, and the next few games suddenly feel daunting: away at LA Galaxy, away at Portland, then home versus LA Galaxy. Keep those seat belts fastened for this period of turbulence, friends. It’ll probably be ... bumpy.
So, in amongst all this doom and gloom, I thought I’d bring you something that always cheers me up: statistics!
We’re back with another edition of this occasional series I call Sounder Data. It’s the space where I use data, statistics, and visualizations to tackle pressing questions about the Sounders and MLS. Today, a re-run of one I did the last two years and which feels quite relevant given our collective angst: strength of schedule.
I do this now because sure, the next three games seem hard, but what are our chances after that? And what are everyone’s schedules shaping up like? We know Portland has the luxury of rounding out the season with a bunch of home games, but surely some of our other competition has tough times ahead of them too. When all the remaining schedules balance out, where are we likely to land? I’m here to see that all your questions get answered.
The first step is to measure strength of schedule. I did this in the same way as before: by counting up the points-per-game of each of our remaining opponents based on the 2019 season so far. I looked up everyone’s PPG at home or away (depending on where we will play them), and then I took the average. The news isn’t great, but not as bad as it could have been:
There you see, as expected, the Portland Timbers with a phenomenally-easy remaining schedule. The combined PPG of their opponents — most of whom have to travel to Providence Park to play them — is a measly 1.08. At the other end of the spectrum are the poor ol’ San Jose Earthquakes who have 10 games left with an average opponent’s PPG of 1.7. Sucks to be them. Fortunately, Minnesota and LA Galaxy (both currently ahead of us in the standings) also have a very hard schedule. RSL and FC Dallas are our other major concerns, and they have moderately easy schedules.
The Sounders are just a little above the median, with an upcoming PPG average of 1.45. That’s not great but not bad. It’s all because our remaining schedule is a mixture of “easy” and “hard” games (in quotes because the margins are pretty narrow both ways) as you can see in this figure:
This is showing how likely each team is to win each of their upcoming games. Green is likely, red is unlikely. On paper, the Sounders have two harder games, followed by four easier games, followed by two harder games, and end with one easier game, roughly speaking (see exactly who those opponents are here). And you can compare those probabilities to our competition, for example Portland (basically all green) and San Jose (lots of red) like I mentioned.
How did I estimate these probabilities you ask? Great question, smart ass. I again used PPG home and away, but this time for both teams in the equation. I threw those two variables into what’s called a multinomial logistic regression, basically a statistical model for predicting a categorical variable like win/draw/loss. It predicts those pretty well because stats are cool. Then I made it simulate out the rest of the season and pick winners/ties/losers for every game for every team. It’s important to actually simulate it all out because every winner creates a loser somewhere else in the league. The simple probabilities aren’t enough. Then, I repeated the simulation 1,000 times because it’s a statistical model and it’ll occasionally predict the less-likely result, perfectly in proportion to how likely it is (which is what we want). Computers.
Using all 1,000 simulated finishes to the season, we can then do fun things like count up the number of times the Sounders make the playoffs, the number of times they get a home seed in the playoffs and the average points they ended the season with across all the simulations.
Here’s the average league table among all 1,000 simulations for the Western conference:
Western Conference Expected Rank 2019.csv
|Team||Average Finish||Average Points|
|Team||Average Finish||Average Points|
|Real Salt Lake||4.3105||52.19|
You can see LAFC looking unstoppable at #1, with an estimated average points of 77. In fact, in exactly zero simulations did they fail to win the Supporters Shield. Apparently they haven’t mathematically clinched the shield yet, but given their current points and schedule, it’s over. Refreshingly, there we are at #3 in the West. The simulation is guessing that we’ll end up with about 53 points, and only Portland and LAFC will end up with more.
But you might notice an oddity, that the simulation also says our average position is 4th. You might also notice the same for Real Salt Lake and (if you round) Portland too. This is because the race super is tight. The simulations are saying that there’s likely to be only one point separating us three teams in the end, if any. There’s also the small matter that I didn’t split ties in the rankings, mostly because it’s harder to simulate, so that figures into those averages too.
Nevertheless, this is good news. Our most likely finish has us in third or fourth place, good enough for a home game in the first round. To look at it another way, 62.4% of the 1,000 simulations ended with the Sounders above the 5th seed (and thereby playing at least one home game in the playoffs). Encouragingly, 92.9% of the simulations ended with the Sounders at least making the playoffs. You can see the distribution of results here:
These are histograms showing the rank each Western Conference team ended the season with in each of the 1,000 simulations. The red line is at 7.5, or the cutoff to make the playoffs (again, I didn’t break ties as you can see in the small bars at 0.5 increments). You can see Colorado, Kansas City and Vancouver with the bulk of their simulations ending below the line, and teams like Dallas and Minnesota teetering on the edge. There’s a notable number of simulations that appear to leave us in 8th place, but again, the majority of the Sounders’ results are on the right side of the red line.
So to recap, the Sounders have a moderately hard remaining schedule (based on 2019 PPG so far), but given that and everyone else’s schedules, we’re still more likely than not to make the playoffs, and even to get a home seed. But according to the simulations, we should prepare for the seeding to be very, very close.