Covering the Front and Back Pages of the Newspaper
April 9, 2008
BASEBALL: Final EWSL Predictions
Now, lest I be accused of predicting the major leagues to finish above .500, I noticed that if you add up the W-L records in my preseason EWSL reports add up to have all of MLB over .500. The reason for that, of course, is as follows:
1. EWSL - by rating only 23 players per team, whereas the average team uses something like 35-40 players in a season - tends to underreport the total number of team wins.
2. To fix this in converting team EWSL to a W-L record this season I applied an average adjustment of plus 12.853 wins per team. That's the average number of wins you get from 1/3 of the average number of Win Shares per team earned in 2005-07 from players I didn't rate in a team's preseason 23-man EWSL roster.
That's a reasonable enough fudge factor, and I was doing one division at a time; but now that I have all 30 teams done, I need to rebalance the numbers to get them all out at .500. Also, I made two adjustments for roster changes between the writing of the previews and the start of the season: I replaced Kelvim Escobar, who is out for the season, with Dustin Moseley, thus dropping the Angels team EWSL from 250.31 to 247.08, and I replaced Reed Johnson (who got rated on both the Blue Jays and the Cubs) on Toronto's roster with John McDonald, dropping the Jays from 209.93 to 207.68. I stayed away from less drastic tinkering, but of course you can expect a downgrade on Detroit's full-season outlook, for example, from being without Curtis Granderson for the early part of the year (not that I'd blame his absence for everything that's gone wrong so far for the Tigers).
With those two adjustments made, we get a major league total of 6193.10 EWSL, which is enough for 68.81 wins per major league team. Now, there are two ways I could get that up to 81 wins per team - proportionally, as I did in 2005 and 2006, or by sticking with the straight addition per team approach. I'm using the latter because (1) historically, I have not observed any notable positive relationship between a team's preseason EWSL and how many WS it generates from players outside the 23-man roster and (2) adjusting proportionally gets us into some question-begging issues about the unbalanced schedule...I just don't want to get into that. So I'm now using a standard adjustment of plus 12.188 wins per team. Of course, for all that math it's an adjustment of less than half a win per team, so the end results here should not be all that dramatic.
Without further ado, here are the final standings according to EWSL:
NL Wild Card: Phillies.
AL Wild Card: Indians.
A few final notes, bearing in mind that in the division previews I already went through where I subjectively expect particular teams to depart from their EWSL baseline expectations. As noted in the divisional previews, EWSL is furthest out on a limb, compared to the general consensus among preseason analysts, in being pessimistic about the Red Sox, Cubs and Rays - the Cubs mainly because of their age, the Rays mainly because of their reliance on unproven youngsters, the Sox because of a mix of the two. The disadvantage of a system like EWSL that is not at all individualized is that it can't target the particular players who are likely to do a lot more than their prior major league accomplishments, as more refined systems like Baseball Prospectus' PECOTA system can. But prediction isn't an exact science anyway; in looking over where things stand entering a season, there's something to be said for considering the discipline of a remorselessly depersonalized system such as this one, which cautions that unproven youngsters should be valued as such until they show us otherwise, and that age cuts down everyone sooner or later. The early injury to Matt Garza is perhaps one indicator of the wisdom of this approach. That said, as an empirical-testing matter, I'll be interested to see whether EWSL turns out to be a better guide as a whole to the direction of those three teams.
Posted by Baseball Crank at 9:25 AM | Baseball 2008 | Baseball Studies | Comments (8) | TrackBack (0)