Two comments. I think karl started all of this, and I appreciate karl's skepticism, being something of a skeptic and curmudgeon myself (as you know). But, karl, when you say that the MVP should obviously and self-evidently be Guerrero or Ramirez and any metric that doesn't come to that conclusion must therefore be suspect, well, there are many ways to suggest that WS and/or WARP aren't cutting it. The fact that they don't end up with Guerrero or Ramirez is not one of the stronger ones.

Two, I also like that WS distributes real wins. I mean, just because all the things that affect winning and losing aren't being measured doesn't mean they don't exist. They do, otherwise every team would hit their Pythag, wouldn't they? So, IMHO, distributing the marginal wins (+ or -) proportionally is better than pretending they didn't happen. And if you distribute the Yankees' 12 wins (36 WS) that way and you come up with, say, 10 percent of them (26/267) of them to ARod, or 4 more (for the actual total of 30), what is your margin of error, really? 0 to 3 in either direction?

Now I acknowledge that that (ARod could be anywhere from 27 to 33) does not make for a good tool for naming an MVP. But over the course of a career, I would guess it evens out--not the over- and under-performance relative to Pythag. A player whose teams consistently outperform their Pythag through his career probably had something to do with it. But rather the margin of error (+ and -) in the assignment of marginal WS would be expected to balance out. So for a career, the distribution of real wins vs. Pythag wins seems like a giant step in the right direction.

But even then, WS is a good way to construct a consideration set of MVP candidates, and especially a way of identifying the bogus, sentimental candidate.

As sort of an aside, then, my consideration set for MVP has to include the three SSs who hit like OF this year--Tejada, Guillen and Young. I'm not sure one of them isn't the MVP.

None of the grand metrics spread out the plausible candidates for 2004 AL MVP wider than the likely errors in those metrics. They inform the discussion, but do not decide the case. So I'd say to anyone arguing for Vlad, or for Tejada, or for Sheffield, or for any of several other candidates (Santana, anyone?): I can't prove you're wrong, and if your guy wins, I won't call it an injustice.

Of course the margin Bonds enjoys over Beltre/Rolen/Pujols/Edmonds is a different matter, even though each of the latter four has performed at a typical MVP level.

The comment that Vlad or Manny should be MVP was my prompt to start disparaging WS, but not really a well grounded reason, I accept. Having all of the top 10 in the NL above anybody in the AL is a much better ad hominem reason. There are then a bunch of analytical reasons, to which I refer you, except that if the bloody thing doesn't use Pythag I accept that my entirely well grounded denunciations of Pythag are utterly irrelevant.

WS is also a good way to construct a consideration set for HOM candidates; you can be pretty sure that below the 180-200 level there really won't be a good case (unless you're talking an 1875 player with 40 game seasons.) But there appear to be plenty of 250s who are better than most of the 320s and plenty of 350s who are really no more than a 280.

My bottom line is that WS is a good shorthand, but far too inaccurate to make the fine distinctions that we need to make once you try to choose between the candidates that aren't named after Presidents.

Two, I also like that WS distributes real wins. I mean, just because all the things that affect winning and losing aren't being measured doesn't mean they don't exist. They do, otherwise every team would hit their Pythag

Right.

I guess I'm with Branch Rickey on this: "Luck is the residue of design." To me, it's not enough to say the Yankees got really lucky. The rational mind struggles to find reason amidst the chaos.

How about this line of thought. How do teams exceed their Pyth Pct by a lot? Isn't it usually by winning a lot of close games? If this is true, then what are characteristics commonly found in teams that win a lot of close games? The first thing that comes to mind is a great bullpen, right? If this is true, then perhaps Rivera and Gordon deserve a greater portion of the Yankees extra win shares.

Of course, there's always the possibility the Yankees won due to things unmeasurable by statistics. Does that great bugaboo of sabermetrics, team attitude/chemistry, count for anything? Is it worth touching on the intangibles?

Why should it be assumed this "overperformance" is due to luck? Has there been a systematic study done of teams that over/under perform their Pythag Pct by big margins?

My working theory, based upon the two examples of the 2004 Yankees and the 1970 Reds, is a perfect storm of mediocre starters, great closers, and crappy middle relievers.

When the mediocre starter has a good game, and the team is up by a run or two, the closers can nail it down and you get the win.

But when the mediocre starter has a bad games, and you're down 6-1 in the fifth inning and need to go to your bullpen, you could end up losing 22-3 with your crappy middle relievers.

In terms of why it doesn't track year to year -- it's likely relatively easy to improve your team by replacing crappy middle relievers, so you don't get the perfect storm two years in a row.


I haven't worked out yet what it would mean for Win Shares if my theory is true.

Having all of the top 10 in the NL above anybody in the AL is a much better ad hominem reason.

Except that this is a product, not of Win Shares, but of "Hardballtimes modified Win Shares". (It might also be a product of Win Shares, but I don't know that, because I can't find an independent source for them.) And if it's solely a product of "Hardballtimes modified Win Shares", then it's a good argument against them, despite what their authors may think of their modifications.

Post from Joe Dimino:
This might sound arrogant, but what the hell, I've been accused of worse . . . the WS values THT publishes are better than James' so there's no reason to publish them both. James flat made a mistake not including negative Win Shares, there is no need to maintain that error.

No comment needed.

I realize that this is not the forum to do this but...
EQA RARP VORP FRAA
Michael Young .269 35.6 61.0 -15
Derek Jeter .280 44.9 60.3 2

So why again is Mike Young getting MVP play and Derek Jeter is not? I guess I am not stumping for Jeter on MVP ballots, but that Young not get on as many. Also I would be interested in what UZR says, last I heard, Jeter was at -4 in september.

Again, sorry for the hijack.

I would hope nobody assumes over/underperformance vs. Pythag is due to luck.

Now, somebody might *hypothesize* that it is due to luck. But of course another person might hypothesize that it is due to random events, which is sort of the same/sort of different. Another might hypothesize that it is due to things that players do that we just can't measure, or haven't yet measured, or even identified.

These are all hypotheses, I think. I guess for me there is an intersection among luck, randomness and poorly measured skills/occurrences and the answer is somewhere in that intersection. IOW we're talking about subtle skills/occurrences, and what is random or lucky about them is that 1) unlike hitting a high fast ball, they are skills and occurrences which only happen very rarely, so they only affect the outcome of very rare and particular games, and so we ignore them most of the time, and 2) when they happen--ie. if they happen in the 1st inning, then we ignore them, and if they happen in the bottom of the 9th in a tie game then we oooh and ahhh.

But what exactly are *they*? Hell if I know.

As to M. Young--karl and all you saberskeptics, help me here. M. Young is an MVP candidate in the purely old-fashioned sense that he is the best player on a team that surprisingly contended for a pennant. Jeter is NOT a candidate because he is only the 4th or 5th best player on his team.

John:

A thread for Dick Redding? :-)

EQA RARP VORP FRAA
Michael Young .269 35.6 61.0 -15
Derek Jeter .280 44.9 60.3 2

So why again is Mike Young getting MVP play and Derek Jeter is not?
---
Because Young is looked on as the only viable candidate who played for Texas, the surprise pennant contender, and Jeter is one of many who wears pinstripes.

IOW, it's how voters' minds work.