Sounders’ remaining schedule setting up nicely

Why hello, Sounders at Heart! I hope you’re all in as good of spirits as I am, because the last few weeks have been pretty good to us. The Seattle Sounders set a record for a comeback, signed a few reinforcements, and have strung together three and a half games of shutouts with a +11 GD.

We’ve now crossed the two-thirds mark in the season, with 11 games remaining, and sights are set on the finish line. Seattle currently sit in 4th place in the West, in a four-way tie at 34 points. Needless to say, it’s a crowded conference. So who has the best prospects for a good end to the season? Who has the hardest schedule? What are the Sounders’ chances of making the playoffs? Do we have a shot at the Shield? The answers to your questions are here.

Welcome back to Sounder Data, the semi-regular series where I use data, statistics and visualizations to tackle interesting questions about the Sounders and MLS.

The obvious question in all of this is strength of schedule. How hard are our remaining games, considering that some are home, some are away, some opponents are good road teams and some are bad home teams? I computed Points Per Game (PPG) for each team at home and away, them matched that up to ours and everyone else’s remaining schedule. At this point in the season, the average of our opponents’ PPG (conditional on home or away), should be a pretty good indication of how hard our remaining 11 games will be. Here’s a graph comparing us to the rest of the league:

As you can see, we’re in good shape. The numbers inside the bars are the average of all our opponents’ points per game for the rest of the season at home or away, depending on where each game will be. We’re right near the bottom with an opponent-PPG of 1.23. That’s the 4th easiest schedule, above Vancouver, San Jose and RSL (who have a remarkably easy schedule somehow). Note, however, that some of our rivals have games in hand.

At the other end is the poor Loons. The heart-warming underdog of MLS hasn’t fared well in their first season. Sadly for them, they have the hardest remaining schedule by a lot, at 1.71.

So there you have it. Things look promising for the final third of this season for the Sounders, and less promising for some of our competition...

But why stop there? What we really want to know is how ours and everyone else’s remaining schedule is expected to translate into points and standings come October. And what kind of data analyst would I be if I didn’t oblige that curiosity!

So I took all of the matches so far this season, and fit a multinomial logistic regression to predict win/draw/loss based on the teams’ current PPG at home and away. Then, I simulated the outcomes of every future match 1,000 times. Basically, I just played out the rest of the season on auto-pilot using all of the evidence we have so far this season, and repeated it a thousand times to see how everything turns out under stochastically different re-runs. This way, we can look at how often (among the 1,000 simulated end-of-seasons) various interesting results occur, such as winning the Supporters’ Shield, making the playoffs etc.

Here’s the results for the Western conference:

This graph shows the total points each team is expected to have by the end of the season, based on who they have left to play, whether they play them at home or away, and how they (and their opposition) have played so far. The curves show the range of results that came out of the 1,000 simulations.

The Sounders are expected to get between 50 and 51 points by the end of the season. But as you can see in the graphs above, a few of the simulated finishes had us coming in at around 40 points, and a few had us up in the upper 50’s. 95% of our simulated results fell in the range of 43-58, 80% fell in the range of 46-55 points.

How that compares to everyone else is what’s interesting though. Out of all 1,000 simulations, the Sounders made the playoffs 853 times (85%). That’s pretty good, but it does leave 147 simulated outcomes where we didn’t make the playoffs.

Among the simulations, we tended to rank 4th in the West the most often. The range was moderately wide, with a 5th percentile ranking of 2nd and an 95th percentile ranking of 7th. In only 38 of the 1,000 simulations did we finish 1st in the West, and there was exactly 1 occurrence of a Sounders Supporters’ Shield victory (so I’m telling you there’s a chance...).

The trouble is that FC Dallas, Vancouver and San Jose have a pretty easy schedule too. To add to that, Dallas’ schedule in particular puts them a bit more in the “driver seat” in terms of where they end in the standings. They also have the edge on us in probability of winning in our two match-ups left this season. The other challenge is the parity. Houston, Kansas City, Vancouver and to a lesser extent Portland and San Jose balance out to very similar finishing positions to us, leaving only a few points to determine the difference. A similar plot to above -this time showing standings- makes it pretty clear:

The Sounders distribution peaks around 4th place, but it’s pretty low and flat. That means every point counts. The same goes for our closest competition too. Once you take into account the fixtures, Kansas City isn’t expected to finish much higher than us, and Vancouver and Houston are nearly the same. Portland and San Jose aren’t far behind either. You can see that Dallas is the strong favorite to come in 1st, as their distribution skews heavily in their favor.

In fact, there was no simulated result in which FC Dallas wasn’t expected to make the playoffs. The same was true for three other teams: NYCFC, Chicago and Toronto FC:

The Sounders ranked 9th in terms of chances of making the playoffs. A few teams (you can guess who) had a zero percent chance. That’s not to say that they’ve been eliminated, just that it’s so unlikely given where they are now and who they have yet to play that it didn’t happen in a thousand trials.

As for the Shield, we’re basically out, along with most other teams:

Again, technically nobody has been eliminated from the running, but probabilities are low for all but a few.

So what does it all mean? It means our prospects for the playoffs are good, but we’re not in the clear just yet. Just a few dropped points and we’ll be biting our nails. But if the next 11 games pan out similarly to the first 23 we should keep up our 100% history of making the playoffs.

Of course, this analysis is only based on what has happened so far in 2017. Signings, coaching changes, or just mental turnarounds could drastically alter the future in unpredictable ways. And of course when the playoffs do come around, it’s anyone’s guess! Hay fe.