</font><blockquote><font size="1" face="Verdana, Arial">quote:</font><hr /><font size="2" face="Verdana, Arial">Originally posted by cincinnati chili:
<strong>Ayala, if memory serves, hurt the Mariners badly in at least one of his years. The Mariners would have been better off picking up some righty at random off the waiver wire. So if win shares were tweaked to account for this fact, I don't think that the Wilson and Ayala trade would have been so bad for the Reds.</strong></font><hr /></blockquote><font size="2" face="Verdana, Arial">Here are Ayala's win shares totals from 1994-1998:
94: 10 WS
95: 7
96: 4
97: 9
98: 0
And here are his raw numbers from those years:
94: 56 innings, 76 Ks, 2.86 ERA, ERA is 71% better than league average
95: 71 innings, 77 Ks, 4.44 ERA, ERA is 10% better than league average
96: 67 innings, 61 Ks, 5.88 ERA, ERA is 16% LOWER than league average
97: 97 innings, 92 Ks, 3.82 ERA, ERA is 19% better than league average
98: 75 innings, 68 Ks, 7.29 ERA, ERA is 36% LOWER than league average
Based on this basic data, I tend to think that win shares does a pretty good job of assessing how many wins Ayala contributed, with his 1996 season as the only possible exception. I haven't done all the math, but there could be other possible explanations for why Ayala got 4 win shares in 1996--maybe he was unlucky (unusual # of hits on balls in play), maybe there's a ballpark effect that distorts the raw numbers, maybe other relievers let all his runs score (i.e., he was hurt by the rest of the bullpen). I assume James has run the numbers on all this stuff.
</font><blockquote><font size="1" face="Verdana, Arial">quote:</font><hr /><font size="2" face="Verdana, Arial">Originally posted by cincinnati chili:
<strong>I think Win Shares are a good, but flawed analytical tool. My problem comes in the fact that all values are positive. Players don't get a negative score if they have a season substantially below replacement level. </strong></font><hr /></blockquote><font size="2" face="Verdana, Arial">True. And I can understand your concern. Ayala's 1998 was indeed horrific, and in your view, he should've been penalized for that.
But one issue I've always had with replacement level statistics (which won't preclude me from using them in the future, anyway
) is: how do you come to a final conclusion as to where the "true replacement level" is? Personally, I haven't seen anything that has left me 100% confident with it.
Plus, I like the fact that win shares uses simple numbers.
<small>[ 07-11-2002, 02:22 PM: Message edited by: D-Man ]</small>