Baseball Crank
Covering the Front and Back Pages of the Newspaper
February 11, 2008
BASEBALL: The EWSL Fudge Factor

It's the time of year again when I prepare for my Established Win Shares Levels (EWSL)-powered team previews by first reviewing how last year's real-world results varied from the method. Of course, since EWSL is a compilation of historical results adjusted by age, rather than a projection system, it's not supposed to flawlessly predict the future but rather to set a reasonable baseline for expectations, from which real life invariably varies.

This time, I will start off with the subjective adjustments I introduced in 2007. Since EWSL works from past performance, it's necessarily less stable when dealing with guys who have a limited track record in the majors, especially second-year players. Second-year players, of which very young players are a large proportion, tend to be split between guys with a full season or a partial season of experience. Given that I haven't had the time and methods to split the two, the age adjustment for very young players tends therefore to over-project non-pitchers with precisely one full season in their early 20s under their belts, while perhaps under-projecting high-quality players with a half-season of experience. Last season I tried to fix that on an ad hoc basis with subejctive tweaks to players whose numbers looked out of whack. Note that I use Win Shares numbers from the Bill James Handbook. Let's see how I did. First, downward adjustments on non-pitchers:

1. Ryan Zimmerman, age 22. Adjusted down from 54 EWSL to 30. Actual 2007 WS: 20.

Zimmerman was a compelling case for a downward adjustment - while I thought and think that his short- and long-term future is excellent, 54 Win Shares would be a career year by a large margin for Albert Pujols. Zimmerman was over-projected because so few players are quality full-season regulars at 21. As it turned out, I still think 30 EWSL was at least a realistic expectation, but one he failed to meet with a sophmore jinx that dropped him to 20.

2. Hanley Ramirez, age 23. Adjusted down from 36 EWSL to 27. Actual 2007 WS: 27.

Right on the nose. Ramirez took a big leap forward from a great rookie year, but 36 was too far for him to reach.

3. Dan Uggla, age 27. Adjusted down from 25 EWSL to 22. Actual 2007 WS: 16.

With Uggla, by contrast, I was dealing with the fact that the typical 27-year-old is not a second year player, so I was really hedging him down for the fact that his rookie year looked like a bit of a fluke. Dropping his EWSL to 16 would have been too big an adjustment to make on that subjective assessment.

4. Melky Cabrera, age 22. Adjusted down from 29 EWSL to 15. Actual 2007 WS: 12.

Like Zimmerman, Melky's numbers were way out of whack - EWSL had him as the best player in the AL East - and doubly so because Melky was very unlikely to get everyday playing time. Also, he was unusual in that most 21-year-old regulars get a job due to their exceptional talent rather than due to a battery of team injuries.

He still projects as a quality player but the downward adjustment looks fine in retrospect.

Conclusion: these were, as a group, necessary and wise adjustments.

Upward adjustments for non-pitchers:

1. Stephen Drew, age 24. Adjusted upward from 8 EWSL to 11. Actual 2007 WS: 16.

Drew looked like a quality player who would play everyday and should not be projected to repeat at less than half a season's at bats. As it turned out he had a poor 2007 with the bat, but his raw WS total went up anyway from a full season's glovework.

2. Ryan Shealy, age 27. Adjusted upward from 6 EWSL to 9. Actual 2007 WS: 1.

Shealy had injuries but he also represents the reason to be cautious about upward adjustments: sometimes the guy who hasn't proven himself over a full season, doesn't.

3. Chris Duffy, age 27. Adjusted upward from 6 EWSL to 8. Actual 2007 WS: 5.

Same problem as Shealy, and again with a 27-year-old.

4. Scott Thorman, age 25. Adjusted upward from 3 EWSL to 8. Actual 2007 WS: 2.

Same problem again, although in Thorman's case it was clear from the outset that he wasn't that good.

5. Kelly Johnson, age 25. Adjusted upward from 5 EWSL to 8. Actual 2007 WS: 19.

Johnson actually did have an excellent year after missing 2006 with an elbow injury and switching to second base. An odd case, and one that's hard to generalize from.

6. Gerald Laird, age 27. Adjusted upward from 3 EWSL to 8. Actual 2007 WS: 10.

Laird did somewhat better with a sudden promotion to full-timer than some of the others, but then he was a backup being promoted rather than a late arrival from the minors.

Conclusion: I may skip the upward adjustments this season, or at least will be much more stingy with them. They were largely unreliable. Better simply to rest on the general fact that EWSL is a system that measures established performance.

Upward adjustments for pitchers:

1. Daisuke Matsuzaka, age 26. Adjusted up from 5 EWSL to 10. Actual 2007 WS: 12.

I'd do this one again; Dice-K was known to be a high-quality Japanese import in his prime, and should not have been rated with the 5 EWSL I hand out to rookie pitchers.

2. Matt Garza, age 23. Adjusted upward from 1 EWSL to 5. Actual 2007 WS: 4.

I suppose this worked out OK. My theory was that Garza after 50 major league innings should not rate lower than a straight rookie. He ended up close to the rookie average. That said, he didn't pitch that well and the rookie average includes guys who have not yet struggled in the bigs. I'll be careful with this in the future.

Downward adjustment - pitcher

Josh Johnson, age 22. Adjusted downward from 12 EWSL to 9. Actual 2007 WS: 0.

It may be that I should not have rated Johnson at all, given that he began the season injured with an uncertain timetable for return. Next time in that situation I may apply a steeper discount to an injured pitcher, maybe 50%.

Posted by Baseball Crank at 1:03 PM | Baseball 2008 • | Baseball Studies | Comments (1) | TrackBack (0)
Comments

While I do have huge reservations about the validity of this particular endeavor I do find the resulting numbers interesting. At least it gives us some baseball to think and talk about. My only suggestion would be to maybe include a column that shows the 2007 Actual Win Shares for each player. Even better would be an additional 2 columns for your 2007 Raw EWSL and the Age Adjusted so there would be some sort of comparison of the prediction of the year before and the actuality of the player's statistics. Perhaps that is a) asking a bit more more work than the amount already required to do this and b) asking the weatherman to keep a percentage score of his forecasts available.

Posted by: jim at February 11, 2008 5:41 PM
Site Meter 250wde_2004WeblogAwards_BestSports.jpg