Thursday, October 04, 2012

Playoff Meanderings

* I always like to write a little something about playoff odds. The playoff odds that I publish are not intended to be the most accurate. They incorporate only full seasonal team data, haphazardly thrown into a weighted average. They assume that win probabilities are constant from game to game. There are other nits I could pick, but those are the huge ones.

So why bother? The main goal is to once again make the point that attempting to predict playoff outcomes is largely a fool’s errand. This is probably an obvious point to anyone who reads this blog, but it is one that I feel compelled to come back to each October regardless.

The methodology here was to use my crude team ratings--based on estimated win ratio and adjusted for strength of schedule. I figured three sets of these--based on actual wins and losses, based on runs scored and allowed, and based on runs created and allowed. Then I combined them and built-in some regression to the mean, with no real rhyme or reason to the weighting (the win ratios fed into the rating are based 30% on actual record, 30% on R/RA, 20% on RC/RCA, and 20% on .500).

These then go into this spreadsheet, which calculates the probability of a team winning a playoff series using Log5 and an assumed home field winning percentage of .545 for a .500 team. First, here are the ratings for each of the playoff teams:

As you can see, the rating system still believes that the AL is superior to the NL (the overall AL rating is 106 versus 95 for the NL). Baltimore comes out better than I expected, and it is worth remembering that playing in the AL East means that their schedule was very tough (after the season, I’ll have a post with full rankings). You may be surprised to see Cincinnati so low, but they exceeded their expected W%s by a fair amount and along with St. Louis were the strongest teams in the weakest division, meaning they faced the easiest schedules in MLB.

Feeding those rankings through the playoff odds spreadsheet, here are the crude playoff probabilities:

I would suggest that Oakland and Washington are the teams that benefit the most from the crude nature of this approach. No team has much better than a 1 in 3 chance of winning the pennant, which is typical but completely out of line with the mainstream notions of anointing favorites. It should come as no surprise that the four wildcards rate as having the lowest odds, but it’s worth noting that the wildcards combined have a higher estimated probability of winning the World Series than the #3 seeds in each league.

Whichever team comes out of the wildcard game will be in decent shape, although as the media will remind you many times will be a bit disadvantaged in their starting pitching options for the Division Series. Here are each of the wildcard’s probabilities assuming they win the game (again, with no penalty for fatigue):

Collapse aside, Texas remains one of the strongest teams on paper. Atlanta compares favorably to Cincinnati or San Francisco, and both Baltimore and St. Louis have respectable chances (in fact, the Cards 8% is better than their 7% from a similar methodology last year).

* The 2-3 Division Series format is already being set up as a ready-made excuse for any of the higher seeded teams that lose. I’m not saying it’s the optimal format, but the importance of having the first two games at home can be overstated. Of course, the model I’m using here can’t account for any psychological effects, but there is no difference in the expected outcome of the series as long there are three home games scheduled.

Theoretically, assuming evenly matched teams in each game of a series, 37.5% of five-game series should go the distance and only 25% should be sweeps. But empirically, 41 five-game series since 1969 have been sweeps, while only 27 have gone the distance. Of course, those empirical results include teams that benefitted from jumping up 2-0 at home.

While the 2-3 format is not ideal, and offers the possibility of what I call a reverse home field advantage (the lower seeded team actually playing more home games), I don’t see any reason to accept it as an excuse. A couple of the higher seeded teams will probably lose, but that happens in typical seasons as well.

It’s also worth noting that a 2-3 format has been used before, most recently in 1995-96. In those two seasons, the cross-divisional matchups were pre-determined. In 1995 for example, the 100-44 Indians opened against the 86-58 Red Sox with two games at Jacobs Field and the final three games schedule for Fenway Park (Cleveland swept). Meanwhile, the 79-66 Mariners played the 79-65 Yankees. The 2-3 of 2012 may be a mild annoyance, but playoff formats have previously reached colossal levels of stupidity to which it can only aspire.

* For the last few years I have also expressed my playoff rooting preferences. There’s no reason why you should care, but it’s good to get it off my chest.

Would be happy if they win: Yankees, A’s, Reds, Nationals, Braves
Would be indifferent if they win: Rangers, Cardinals
Would be annoyed if they win: Orioles, Tigers, Giants

So expect a BAL/SF World Series (1.8% chance based on the crude odds).

No comments:

Post a Comment

Comments are moderated, so there will be a lag between your post and it actually appearing. I reserve the right to reject any comment for any reason.