MLS Playoff and Supporters Shield Odds at the Halfway Point
We're now halfway through the 2011 MLS season, more or less, and the playoff picture has started to shake out. As the last few seasons have shown, championships aren't won by the best regular season team. The secret to winning an MLS championship is making the playoffs by hook or by crook and then managing your matchups (and a little luck) in the postseason. The MLS brass tweaked the formula a little this season to make a higher playoff seed slightly more valuable by forcing the last 4 playoff teams into a preliminary round, which gives a strong incentive to make the top 3 in each conference. All else being equal, avoiding that first playoff round effectively gives the top 6 twice the odds of winning it all as the wild cards.
Regular readers know we run a season simulation that takes into account goal-scoring and -preventing ability, home field advantage, and a recency bias to project out the remainder of the season. I've been running it since opening day, but now the results are baked in well enough that I think the results are starting to become interesting. The simulation has consistently shown from almost the first week that the point cutoff for the playoffs this season is somewhere around 43-45 points (currently 43.8). This is a big jump from the 40 points that would have got you in last season over the then Kansas City Wizards due to the increased number of games with 18 teams, but not as big a jump as it would have been due to opening up the playoff field to 10 teams.
We're also calculating the odds of winning the Supporters Shield this season. That's a race that collapses very quickly to just a few teams. Even now, just over halfway through, we calculate only two teams has having more than a 10% chance and five teams with more than a 5% chance to win it. The full table of projected results is after the jump:
| Team | Avg Pos | Avg Points | Playoff % | Shield % |
|---|---|---|---|---|
| Los Angeles | 2.5936 | 58.6429 | 0.9986 | 0.3581 |
| FC Dallas | 2.7541 | 58.4971 | 0.9968 | 0.3367 |
| Seattle | 4.7864 | 53.4099 | 0.9756 | 0.071 |
| Real Salt Lake | 4.7965 | 53.4151 | 0.97 | 0.0942 |
| New York | 5.0627 | 52.4788 | 0.9681 | 0.0695 |
| Columbus | 6.0821 | 51.1635 | 0.9292 | 0.0344 |
| Philadelphia | 6.6339 | 49.9788 | 0.9016 | 0.0225 |
| Kansas City | 7.7 | 48.2063 | 0.8186 | 0.0102 |
| San Jose | 10.2476 | 43.961 | 0.5207 | 0.0016 |
| Colorado | 10.598 | 43.5932 | 0.4691 | 5.0E-4 |
| Chivas USA | 10.6594 | 43.128 | 0.4668 | 4.0E-4 |
| DC United | 11.3251 | 42.4583 | 0.3866 | 8.0E-4 |
| Chicago | 12.5847 | 39.7803 | 0.2313 | 1.0E-4 |
| Houston | 12.5799 | 39.7873 | 0.2269 | 0.0 |
| Portland | 14.2646 | 37.1926 | 0.0835 | 0.0 |
| New England | 15.4354 | 34.2539 | 0.0356 | 0.0 |
| Toronto FC | 16.0564 | 33.2253 | 0.0162 | 0.0 |
| Vancouver | 16.8396 | 30.359 | 0.0048 | 0.0 |
| Avg Pts for Playoffs | 43.8378 | |||
Note that despite the continued domination of the Western Conference, it's very unlikely that the top 10 won't include 3 Eastern Conference teams. Which means we'll likely avoid the nightmare scenario team of a West team missing the playoffs entirely despite having more points than an East team. However, it does mean that the wild card slots will again be dominated by the West. The current projection is for 3 of the 4 wild cards to be from the West, and if Sporting Kansas City slips or Chivas USA goes on a run it could be all 4.
The top 7 teams are all 90% assured of making the playoff at this point. The LA Galaxy already have 36 of the magical 45 points needed with 14 games left to play. They, FC Dallas, the Seattle Sounders, and Real Salt Lake could average under a point per game in the remaining season and still get to 45. The Philadelphia Union, New York Red Bulls, and Columbus Crew aren't far behind that pace.
And there are only 4 teams with less than a 10% chance to make the playoffs. That includes both expansion teams (despite an early season run that made the Portland Timbers look like serious playoff contenders) as well as Toronto FC — who look like they're hitting rock bottom on the pitch — and the New England Revolution — who look like they're hitting rock bottom off it. Portland can't be discounted completely. A 1.5 point per game average — which is a strong but not ridiculous pace — would put them exactly at 45 points and a shot at the playoffs.
So the jockeying for a playoff spot over the next couple of months will mostly take place among the remaining 7 teams. Kansas City is in the healthiest position among them. We were cautioning people not to underestimate the impact of Sporting's ridiculous road trip to start the season, and now that Livestrong Sporting Park has opened up and the team has a home venue, they're starting to rack up points and results and now project out at as a wild card team with a decent shot of taking a top 3 position in the East. Below them sit the two Eastern Conference Finalists (despite being from the Western Conference) from last season, including the defending MLS Cup champions. Both San Jose and Colorado are even money to make the playoffs. And below them Chivas, DC United, Chicago Fire, and the Houston Dynamo are longer shots.
The Supporters Shield race is mostly a two-team race at this point, with about a third of the simulations going to the Galaxy, a third to FC Dallas, and a third to the field. The best shot outside those two teams goes to Real Salt Lake at about 9%, and then the Sounders and New York around 7%.
Of course all of these projections are only based on past (mostly recent) performance and can't account for any significant changes to the teams. A big injury, a big return from injury (or suspension, as in the Brian Mullan case), big signings, or a big player exodus could all significantly impact a team's performance.
16 comments
|
1 recs |
Do you like this story?
Comments
zeroes
The zeroes in the last column after Portland and Vancouver’s name bring me much joy.
by Derek Young on Jul 7, 2011 12:06 PM PDT reply actions 5 recs
Keep it up
Keep up this kind of analysis! It’d be awesome to be able to take a peek after each week to see how the %’s and odds are changing!
Widget anyone?
"But who would listen to Little Old Me anyway?"
-by -Dave Clark
and -thehemogoblin
by Little old me on Jul 7, 2011 3:55 PM PDT up reply actions
Hmmm...
Seems like the Shield odds are a bit too high/low at the extremes for halfway through the season. Can you point to a link on how your “recency bias” works? I’m curious how big an effect that has on these results. Thanks.
The recency bias
is a straight linear weight from 1.0 for the most recent game back to 0.0 for the first game of the season, so it doesn’t overly bias recent results. A game in May counts half as much as a game yesterday.
One reason that Shield probabilities collapse so suddenly is that teams that win a lot tend to continue to win a lot. People like to look at standings and think ‘oh, if we start outpacing the leader by a point a game, we have a shot!’ But the team that’s leading is leading for a reason and it usually continues to happen.
Nos Audietis
I guess I still put a lot of weight in regression to the mean at this point in the season
Yeah, the top teams at this point are going to tend to do better than worse over the balance of the season, but in the grand scheme of things, 16-18 games isn’t a very big sample size (especially in a league with so much parity) and there ought to be plenty of regression to the mean when we are figuring expected record based on current record and current goals scored/allowed.
I guess you'd have to run the numbers
But anecdotally, I don’t remember this happening much. The teams with the best records at midseason tend to be the teams with the best record at the end of the season. I see what you’re saying, but I don’t know that it holds up to actual analysis. Would be interesting theory to test, tho.
Editor/writer at Sounder at Heart, MLS editor SB Nation. Follow me on Twitter
by Jeremiah Oshan on Jul 8, 2011 9:39 AM PDT up reply actions
Just as a quick check
The first year I could find mid-season standings for was 2009 (happened upon a week 15 power rankings page), so I could compare first-half PPM to second-half PPM. Yeah, it’s just one season, but if it’s a strong, consistent relationship between first-half play and second-half play, then it should show up nearly every season.
But plotting the first-half PPM against second-half PPM, we see that there is very little relationship between the two. It’s true that the better teams tend to play better, but there is a big variance in play. Doing a linear regression (obviously rough but this is a first approximation) I got an R-squared of .046—really small. I would post the graph here, but I don’t know if there’s a way to upload images to a comment.
Another number I can put to this is that in 2009, the standard deviation of first-half PPM vs. second-half PPM is 0.41—practically the difference between being an average team and a Shield contender. As an anecdote from that season, LA after 15 games had just 1.2 PPM—4th from the bottom of the table—then came within a point of the Shield with a second half of 2.0 PPM.
So overall, my feeling is just that the race is more wide open than sidereal’s odds suggest.
Couple things on regression
First, regression to the mean happens in the proportion that luck factors into results. I take some of that out by basing the results on goals rather than wins and could probably take even more out based on the work we’ve done with shots, but regardless I don’t think anyone seriously thinks that all of the teams in MLS are league average and standings variances at this point are just due to randomness.
Second, even with regression to the mean, points already earned have already been earned and count towards table position and playoff odds. The Galaxy could be tediously average for the rest of the season and are still almost locks to make the postseason simply because they’ve already baked in so many points, as I alluded to in the article. That’s also applicable to the Shield. Even if you argue that all teams are effectively equal in quality, the Galaxy/Dallas combo will still dominate the probabilities because they effectively have a 4-5 point head start on everyone else. It’d be like flipping a series of coins but giving yourself a couple of heads before you start. Heads is going to dominate the probabilities until you get really far out in terms of number of flips. And again, Dallas and LA aren’t average. So you’ve got two good teams who are going to stay good and who have a 5 point head start. So it doesn’t seem that out there to calculate that it’s better than even that one of those two will win the Shield.
Nos Audietis
"Both San Jose and Colorado are even money to make the playoffs. And below them Chivas, DC United, Chicago Fire, and the Houston Dynamo are longer shots."
Am I reading the table wrong? Doesn’t it say Chivas has about the same playoff chance as Colorado?
Odds
The not-quite-degenerate gambler in me looks at this and wants to find someone laying 14-1 odds on the Sounders winning the Shield…though if given truth serum I’d put Dallas as pre-transfer window favorites.
For that matter, I don’t really like this but getting the same odds on Portland making the playoffs would be a tempting bet as well.
updates?
Any chance at updating this? I’d love to see the changes that have happened over the last couple of weeks!

by 



















