MLS Playoff Hopes: The Process of Elimination

As the 2011 MLS Season rounds the turn to the final stretch of the regular season, we've been talking about playoffs quite a bit here. We've looked at playoff position and playoff chances based on the current MLS table, based on a simulation, based on a target number of points, based on points per game standings, and so on. But another perspective is that a team makes the playoffs when 8 other teams don't. So instead of focusing on which teams are going in, we can take a look at which teams are out.

Last week I tweeted out our playoff odds for Cascadia teams, which at the time were: Seattle Sounders 99.7%, Portland Timbers 12.8%, and Vancouver Whitecaps 0.02% (the Timbers have since dropped down to just over 8% after losing in Kansas City). A few people responded with some surprise that the Whitecaps had any playoff chances at all and were wondering if they were mathematically eliminated. The short answer is no they're not mathematically eliminated and in fact they're not really even close. This is a reminder that mathematical elimination is a pretty useless standard, and I wanted to discuss it along with some other standards we could (and should) use when considering teams eliminated from playoff races.

Mathematical Elimination

The only attraction of mathematical elimination is that it's definitely, absolutely correct. You never want to say someone is eliminated and then have them suddenly not be, and barring a change in the rules or league structure midseason, a team that's mathematically eliminated is reliably out of the playoffs. But the downside of a such a high standard is, as I've pointed out, that it takes forever to happen and it happens well beyond the point at which even a committed fan or team executive would acknowledge that a team is no longer in a playoff race. The Whitecaps, which are the least likely team in MLS to make the playoffs, are in fact only 13 points behind the New York Red Bulls, who are currently in playoff position (ignoring the complexities of games in hand for a moment). Thirteen points is a run of 4 wins and a draw, and Vancouver has 10 games remaining in their season, so it's certainly physically possible for them to get those points.

To do so they would not only have to play way, way above their demonstrated level of ability thus far this season, but the entire rest of the league would have to cooperate by having every other near-playoff team play way below their demonstrated level of ability by effectively getting no more points. A committed fan might hope for the former, but hoping for the latter is untenable.

Effective Elimination

That brings us to a more reasonable standard of elimination. Instead of comparing a team to where its opponents and the playoff standard are now, we can compare them to where the playoff cutoff is likely to be, given the normal standard of play in the league. In other words, if the Mathematical Elimination standard requires both that a team play like the best team in the world and the other contenders play like the worst teams in the world, the Effective Elimination standard removes the second requirement and instead assumes that the rest of the league will continue more or less as normal, but allows for the team we're looking at to improve dramatically. For example, the MLS playoff cutoff over the last few years has reliably been around 40 points. This season, with the greater number of games, we're projecting it to be around 44. But we could accept 40 as an extremely conservative guess as a target for a team to make the playoffs. It's not outside the realm of possibility that a team with less would get in, but nobody should make an important decision based on that hope.

So if we take a look at teams that don't have a chance to make 40 points, we can call them Effectively eliminated. For example, Vancouver can get to 48 points if they win out their remaining schedule. Of course they won't, but that's not the standard of Effective elimination, which is still pretty high. And so Vancouver isn't Effectively eliminated yet, but that will come sooner than Mathematical Elimination. Every game they don't win drops their potential maximum by 2 or 3 points, and so we're likely to see effective elimination after 3-5 more matches, with Chicago, New England, and Toronto following soon after.

Projected Elimination

If we say that Effective Elimination removes the assumption that the rest of the league will play like crap, then Projected Elimination removes the assumption that the team we're looking at will play like superstars. Instead of picking a target that the team might achieve if they play really well, we instead project out based on how they've already played. There are a few ways to do this. . you could use PPM standings if you assume that the team will continue at their current PPM rate. Instead I'll use the simulation that we run anyway which has the advantage of being probabilistic, so we don't just say that Chivas USA is eliminated by projection because they're currently outside the top 10, but instead assign them a probability of making up that gap. The probability cutoff that determines whether we want to call a team eliminated is obviously arbitrary, but I think that playoff odds of less than 1% are safe enough. You might get that wrong once in 10 years or so. So which teams have reached projected elimination? Right now that list includes Vancouver, New England (at 0.13%), and Toronto FC (at 0.15%). Chicago and San Jose are barely holding on at more than 1% but less than a 2% chance to make the playoffs.

For a team that's reliably in the playoffs, like the Sounders, this doesn't matter as much. But a team like Chivas or New York Red Bulls will be keenly aware of which other teams have been eliminated, since they only make it in when that number hits 8. I'll probably post updates as the weeks go on regarding which teams have reached which standards of elimination.