Thursday, September 02, 2004

Congress Repeals Law Of Averages

I came up with the idea of a "Congress Repeals Law of Averages" article for some Onion-like publication. I thought it was funny but I've had mixed reviews.

I've had internet connectivity issues today so I haven't been able to complete research on this post (in large part because Baseball Prospectus has had a rare slip-up with SNWL numbers not being updated in four weeks), but I have enough to make the argument. When I posted about the Yankees and Red Sox, I developed a brief hypothesis that perhaps the Yankees were outperforming their run expectancy due to really bad starts and that Support-Neutral Win Loss statistics could shed light on the issue. SNWL creates an expected win-loss record for pitchers based on how many innings they go in a start and how many runs they permit. My hypothesis was that maybe the Red Sox wouldn't look so unlucky (they have by far the greatest positive run differential in the AL but don't have a matching record based on the Pythagorean formula) once SNWL was considered. I didn't have any data on this when I thought about it, so I thought I could take an hour or two today to compile it. Turns out that wasn't possible. Instead, I'll offer you the preliminary evidence. To facilitate that, we will pretend that today is August 6 since I don't want to try to make my own SNWL formula at the moment and August 6 is the most recent update in SNWL stats. My suspicion is that the Red Sox luck has been better lately and maybe the Yankees luck has been worse, but we're just looking at games through August 6 now.

The Red Sox have just lost four of their last six games while playing the Twins, Devil Rays, and Tigers. All four losses were by one run, and both wins were by three runs. At this point they have the best offense in the league even when adjusting for park factors. They are third in the AL in Support-Neutral Wins Above Replacement with 7.7, .9 behind Oakland and .1 behind Minnesota. It should be noted that SNWAR does not adjust for defense, and thus the Red Sox supposed defensive woes are accounted for. The Yankees, on the other hand, have a SNWAR of only 2.2 and a very good offense that does not match up with Boston's. The Yankees have gotten 191.1 innings of 2.35 ERA pitching from Rivera, Quantrill, and Gordon, but the Red Sox have gotten a 3.07 ERA in 179 innings pitched from Foulke, Embree, Timilin, Mendoza, and Williamson. Both teams have been equally disastrous when any reliever not in that group has entered the game. Furthermore, the Red Sox have a substantially better batting average with runners in scoring position than the Yankees (sorry I can't get the exact data, for August 6, to back this up, but as of September 1 the Red Sox lead in that category, .297 to .274). The Yankees, in spite of all this, have a 69-39 record while the Red Sox are only 58-49, ten and a half games back.

--BACK TO SEPTEMBER--

If you looked at it on August 6, my hypothesis looks terrible. The Yankees bullpen is worth maybe 2-4 extra wins, but the Red Sox should, without question, lead the division by at least a handful of games (that is, if you're looking at the contribution of individual players) and instead trail by seven games. I probably won't ever understand why many Red Sox fans seem so strange most of the time, but somethiing like this might be a factor.

You're probably thinking: isn't this where the plexiglass principle comes in? Hasn't the Red Sox hot streak and Yankees cold streak evened this up? Well, no. The Red Sox have gone 20-4 since August 6. Boston did manage to win 5 games where they allowed more than four runs, including one where they allowed nine, and they did only lose two one-run games and won four one-run games in that span; both of those facts might seem to indicate a reversal of forunes, but they really don't mean much. The Red Sox run differential has been huge, 167-106, which would suggest a 17.1-6.9 record over that span, so their 20-4 record isn't particularly lucky. The Yankees, meanwhile, have played pretty much to their expectancy if we decide to be nice and not count 22-0 losses in their run differential, going 13-10 in those games with an expectancy of 13.1-9.9 (if you count the big loss, they're 13-11 versus an 11.4-12.6 expectancy).

I'm not sure I believe in the plexiglass principle, but I do believe in the law of averages (Congress be damned). Over any future period of time, we're likely to see baseball teams and players produce in accordance with their true performance levels rather than in accordance to their past raw results. If the Yankees lose in the playoffs or lose the division race, it will not be because they collapsed but simply because they had a stretch where their productivity merely matched their performance level instead of exceeding it and because Boston produced in accordance with their overwhelming advantage in performance level.

Comments: Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?