I’ve advocated for a long time that Shanley should quit his job (ok, he did that part) and move to Vegas to make his living betting on sports events. In the past, he’s done in depth statistical analysis on such things as the significance of winning the first game of the series in the NHL postseasn. Today, Shanley brings us an analysis of the MLB postseason. I’m not a betting, man, but if you are, this might interest you.
The problem with picking baseball winners is that most results are determined by luck, especially since postseason teams are strong for the most part and only a few wins differentiate them. The only team with the best record to win the championship in the last 10 years was the ’98 Yankees, and they won a then-AL record 114 games. The factors that do give a minor statistical boost to some teams are regular season record and what one analyst calls a “secret sauce” – several team stats that bear relevance in the postseason whereas most do not.
http://www.diamond-mind.com/articles/playoff2002.htm
This article describes a method for calculating world series probabilities based on won/lost records. Interesting to note that for most years, the best team has a 25% or less chance of winning it all. That article does the computation for 2002; in that year, Anaheim (10.4%) defeated SF (7.6%). Last year, Chicago (18.7%) defeated Houston (7.9%). I ran the numbers this year and got this:
- Mets: 20.71%
- Yankees: 17.43
- Twins: 15.96
- Tigers: 12.79
- A’s: 11.09
- Padres: 9.24
- Dodgers: 7.55
- Cardinals: 5.23
The Yankees would actually be slight favorites (51.3%) against the Mets due to identical records and home-field advantage, but they’ll have to face tougher competition to get to the World Series.
Now for other considerations.
- The secret sauce article is here: http://www.stltoday.com/forums/viewtopic.php?p=3936878. In the relevant statistics, the Twins and Mets do very well, the Yankees above average, and the Cardinals really badly.
- For the first time, one league (the AL) was much, much better during interleague play, and most fans believe that whoever comes out of the AL will be a strong favorite in the World Series. So W/L comparisons for teams from different leagues might be off.
- Offense has been shown to have NO statistical correlation with postseason performance, but run prevention (pitching and defense) has a slight correlation. So the Yankees’ mammoth offense doesn’t appear to be statistically relevant. However, some people think that their offense is so good, it might be enough to matter.
- So I’m not sure how to build these factors into the model, but here’s what I did: Give a boost of 4 percentage points to all AL teams, 4 points to the Twins and Mets, 2 to the Yankees (for run prevention, not run scoring), and subtract 4 from the Cardinals. Here are the revised numbers. The second column is the expected payoff on $1 based on the most recent odds I found.
- Mets: 24.18% 0.45
- Twins: 24.00 0.44
- Yankees: 20.19 -0.56
- Tigers: 10.86 0.09
- A’s: 9.15 0.01
- Padres: 5.98 -0.46
- Dodgers: 4.39 -0.61
- Cardinals: 1.24 -0.84
So, bet on the Twins (5-1 odds) and Mets (5-1) to win, and against the Yankees (6-5).
On a completely random note, because I didn’t read any of this post at all but am not allowed to post computerish stuff on Katy’s blog, is there any way to quote comments without having to cut and paste? Like in LiveJournal or the WoW forum?
There is a plugin to do that (which I’ve installed) but it doesn’t seem to be working. I might spend a little time debugging the code for it later, but it’s not very well written (one massive php file). Gotta love php.
I see it as working.
Ok, wow, I don’t see anything. Maybe it’s browser dependant. I’m using Firefox, are you using IE? I enabled it on katy’s blog as well.
I’m also in Firefox, but it hasn’t been updated for a while. There are nice little “Reply here” links and the messages are nested. Are you sure you don’t see it? Because your comments are nested too. Well, I guess it only nests up to four levels, btw.
Oh, it did next. I don’t know why it said that it wouldn’t.
*nest. Sorry. I haven’t figured out how to edit my posts yet.
I think that’s the threaded comments feature. Is that what you were looking for, or were you looking for the ability to select text in a previous comment or text from the post and have it automagically copied to your current comment? That was the plugin I installed, and it doesn’t seem to be working.
This will work. I just wanted something to show I was responding to a specfic comment. But Katy’s blog doesn’t seem to have this feature. I don’t know if yours always has.
I’ll add it to Katy’s blog. I’ve had the feature since dusty complained about the issue a while ago.
Bitching to get things done properly since 2004. ™
So, as much as it hurts, now that the playoffs are over, I’ll analyze the strengths and weaknesses of this model. For starters, the model identified six favorable series bets (Mets LDS, LCS, & WS, Twins LCS & WS, Tigers LDS). Betting 1 unit on each would have resulted in a net loss of 0.63. However, betting 1 unit on the favorite for every series would have resulted in a net loss of about -4. Perhaps the most important point was the first sentence in the original post: “most results are determined by luck.”
Some thoughts on certain teams:
- The model correctly showed that the Yankees’ high odds of winning were unjustified.
- I think the Twins are still a good postseason team, despite their performance. Just goes to show that if a team plays flat for a couple games, their season is over in a hurry.
- The Mets had a 52% chance of making the World Series and didn’t. They had two dependable starting pitchers get injured right at the start of the playoffs and their starting left fielder missed the NLCS. Even so, their seven game series against the Cardinals was essentially even.
- Nobody gave the Cardinals a chance, including this writer. Even with the luck factor, this model didn’t recommend taking the 12-1 championship odds for St. Louis at the start of the postseason. It was reasonable to think they could win the first round, but they were at least 2-1 underdogs against the Mets and Tigers. So what happened?
1. Injured players (Rolen, Edmonds, Eckstein) got healthy just in time for the postseason, and contributed on offense and defense.
2. Starting pitching suddenly got a boost when Weaver and Reyes came out of nowhere to pitch better than they had all season.
3. It might have actually helped that their mediocre closer went on the DL in September. The bullpen had given no indication all year that they could pitch as well as they did in October.
So, here’s the formula for playoff success – get in, and get hot.