PDA

View Full Version : 2004 Yankees; Statistical Anomoly???



SODM
10-16-04, 11:17 AM
The 2004 Yankees exceeded their expected win total by an astonishing 11.5 wins based on pyth.
Substituting Eq and aEq for R only explains away one or two of the wins.

So the question is WHY?????
Can anyone come up with any statistical justification or can we chalk this up to the intangible qualities so many baseball observers claim we have?

Right now I am entertaining the notion that this year our lack of pitching depth has caused us to give up an excessive amount of "runs that don't matter" in lopsided games.
I'll let you know what I come up with.

Anyone else got any other theories?


Just to put this anomoly in a historical perspective the 2004 Yankees would be the third most overachieving team IN THE HISTORY OF BASEBALL behind the 1905 Tigers and the 1984 Mets.



Here's a BP article that lists the greatest pyth disparities through 2001.

http://www.baseballprospectus.com/article.php?articleid=1019

WiffleWOOD
10-16-04, 03:15 PM
2 reasons:

1. As you said, the middle relief has greatly inflated the number of runs we've allowed. Those guys also came in during blowouts and put them further out of reach. There are also those 3 horrifying scores (the 22-0 game, etc) that had a small effect.

2. Mariano Rivera.

Irony Of It All
10-16-04, 03:28 PM
Originally posted by WiffleWOOD
2 reasons:
2. Mariano Rivera.

2. Mariano Rivera and Tom Gordon.

WiffleWOOD
10-16-04, 04:46 PM
Originally posted by Irony Of It All


2. Mariano Rivera and Tom Gordon.

right.

Eric Van
10-18-04, 11:18 AM
There are three things going on here.

First, someone at SoSH (not me! But I did interpret his results) did a study to try to refine the Pythagorean. What we discovered was this: an even better prediction comes if you subtract some part of the Standard Deviation of Runs Scored and Runs Allowed from the totals.

What this means in plain English is that teams with inconsistent offenses will underform their straight Pyth, and teams with inconsistent pitching will overperform.

Think about that, it makes perfect sense. An inconsistent offense means lots of blowout wins and lots of low-scoring games. That leads to fewer wins than a more consistent offense. An inconsistent defense means you allow lots of low scoring games, which leads to more wins, and get blown out a lot, which artificially inflates the Runs Allowed.

You guys have already figured this out. You'd be hard pressed to find a team where the last 2 (second half of season) or 3 (first half, before Quantrill's arm fell off) guys out of the pen were so great and the rest of the guys were so awful.. This means that once you're down by a few runs, you have a tendency, very unsual in a team otherwise this good, to get your asses royally kicked.

So, if you used the perfected version of the Pyth formula, I'm confident that that would knock a good bunch of games off the overperformance.

The second factor is just plain luck. My buddy Cecilia Tan, author of the forthcoming book The Fifty Greatest Yankee Games Ever and both a Yankee and SABR diehard (and for the one person out there who might be wondering, is that the same Cecilia Tan who's the world's number one writer and editor of science fiction erotica, yup), assures me that the team has indeed just been lucky this year.

However, the Yankees have now overperformed Pyth for an astounding twelve straight years, averaging +3.8 wins per year. I suspect that doesn't go away if you use the revised Pyth with the inconsistency adjustment. This is one of the great mysteries of sabermetrics. The easy explanations -- it's Mo, it's Torre -- run into the obvious problem that there have been other great closers and other great managers, and no one has ever duplicated this run. I've come up with other really interesting explanation, which I'm keeping under my belt for the time being, except to say that it involves roster construction and finding a certain style of player.

My best guess is that having a truly great closer helps you beat Pyth, having a truly great manager ditto, and ditto (tritto?) for my X factor. And that the Yankees are simply the first team in MLB history to have all three happening simultaneously. Other teams that have had one or two factors will have a tendency to beat Pyth in a given year, but then you add in luck, and they don't succeed in putting together a run like the Yanks.'

Hope that sheds some light on the matter.

STNYY
10-18-04, 11:42 AM
Originally posted by Eric Van
However, the Yankees have now overperformed Pyth for an astounding twelve straight years, averaging +3.8 wins per year. I suspect that doesn't go away if you use the revised Pyth with the inconsistency adjustment. This is one of the great mysteries of sabermetrics. The easy explanations -- it's Mo, it's Torre -- run into the obvious problem that there have been other great closers and other great managers, and no one has ever duplicated this run. I've come up with other really interesting explanation, which I'm keeping under my belt for the time being, except to say that it involves roster construction and finding a certain style of player.

My best guess is that having a truly great closer helps you beat Pyth, having a truly great manager ditto, and ditto (tritto?) for my X factor. And that the Yankees are simply the first team in MLB history to have all three happening simultaneously. Other teams that have had one or two factors will have a tendency to beat Pyth in a given year, but then you add in luck, and they don't succeed in putting together a run like the Yanks.'

Hope that sheds some light on the matter.

Why keep it under you belt? ;) My intuitive, non statistical sense of the Yankees success in the current run also includes a "certain style of player" theory. The O'Neill/Tino/Brosius Yanks of 98-01 not only fit into the grinding out ABs, high OBP patient hitter mode (I remember the biggest knock against the '98 Yanks was how godawful long their games were) but a kind 'kick 'em when they're down' mentality. Those Yankee teams just stomped on bad teams, the unwritten rule against humiliating another team meant nothing to them. They piled on the runs in a sort of inverse of the current bullpen situation - they would exploit any weakness in opposing middle relief and show no mercy. Not unlike Game 3. Approaching every AB determined to get on base or score a run, even when you lead by 10 in the 7th, is bound to carry over into tight game situations against superior pitching, or at least put you in a better position to capitalize on good luck.

That 'go for the kill' attitude has also always been reflected in Torre's bullpen management, much to the annoyance of many on this board. When you have the personnel to make it work, you can win an historic number of games. When you don't, you have alot of tired arms in Sept/Oct.

Game 3 was the first time since BK Kim in 2001 that I saw the Yanks play as if they were trying to make an opposing player cry and I liked it.

ctan
02-22-05, 10:14 PM
Eric, WTF are you doing here? :) You've found my longtime online hangout.

Anyway, this all just reinforces my ongoing theory that the main difference between the Yankees and all other teams is we're luckier. It *can't* be possible, I keep telling myself, that we keep beating the odds/stats--it has to even out sometime, doesn't it? But then it keeps going. I didn't realize it was 12 years now in a row that the Yankees beat the Pythagorean win total, though. So that pre-dates Torre, back into the Showalter era. Thanagain, if there is no strike in 1994 and Showalter used his bullpen a little better in October in 1995, maybe the dynasty starts a few years earlier...

See you around SABR.

-ctan

VFBundy
02-25-05, 06:31 AM
If you look back all the way to the 1998 season, each year the Yankees "out-pythag" the Red Sox, meaning the Yankees exceeded their pythagorean win % by more games than the Red Sox did (who, many times, finished below their pythagorean win %). The one exception was 1999, where they were tied; each team exceeding their pythagorean win % by two games. In three of the past eight seasons, the Red Sox pythagorean win % is actually higher than the Yankees.

The reason has to be Mariano. During that stretch, the Yankees winning % is close to .600 in one-run games, whereas, during that same stretch, the Red Sox winning % is below .500 in one-run games.

I posted about this last year in a Yankees newsgroup. The numbers were amazing. Maybe I can find it later. (Gotta get ready for work now. :()