Friday, January 07, 2005

Back on the Ledge

Baseball Prospectus announced today that they're entering the crisis negotiation business. Their first target is Mark Shapiro, who reportedly is offering Kevin Millwood $7 million:

Maybe they think they can afford to chuck $7 million on a look-see. A much smarter look-see is the Red Sox signing of Wade Miller for the same length of time but at a fraction of the price.

Miller and Millwood were fairly comparable in 2002-03 (combined VORP of 67.4 for Millwood and 60.7 for Miller). Both ran into arm trouble last year and missed a considerable number of starts. When they did pitch, though, Miller was superior, boasting a VORP of 21.8 to Millwood's 9.3. Both have comparable strikeout rates (within a quarter-K per nine). Miller is two years younger.

Millwood, of course, has more "experience." . . . To our way of thinking, the Red Sox established what the market value is for moderately successful starting pitchers coming off injuries is when they signed Miller to one year at $1.5 mil. Because of that, anything beyond $2 million for Millwood is excessive on the Indians' part.


Now, there's a lot of fine analysis in there. However, a couple things stand out to me that I think are valuable lessons.

First, don't look at dollar figures too abstractly. Miller gets $1.5 million guaranteed, yes. But he also gets $3 million in incentives. Millwood is a pitcher who has, apparently, already recovered from his injury. The Indians deal has been held up for a while, reportedly, because the Indians are requiring Millwood to submit to an extensive battery of medical tests. Miller, on the other hand, still has a ways to go, and there are reasons to doubt whether he'll be able to pitch this season at all. So the cost of Miller's "look-see" is $1.5 million, and the cost of his performance is between $0 and $3 million. On the other hand, the Indians look like they'll only make this deal if they're fairly certain that Millwood can pitch a full season.

Second, all in one stats are what they are. VORP is value over replacement player, not talent or ability. For pitchers, VORP is runs above replacement per innings pitched times innings pitched. It's based on runs allowed and park factors. It doesn't make any allowances for defense or luck. Miller had a big edge in 2004, yes, but was that due more to his performance or to noise like sample size and defense? Let's take a basic look. I'm not gonna go all out on this, so here is each player's raw ERA, DIPS (defense independent pitching), ERC (component ERA), and BIPA (balls in play average) for the past five years. I've also added ERC% (ERC/ERA), DIP% (DIPS/ERA), and D/E (DIPS/ERC). First Millwood:

ERC DIPS ERA ERC% DIP% BIPA D/E
2000 3.84 4.03 4.66 0.824 0.865 0.285 1.049
2001 4.28 4.72 4.31 0.993 1.095 0.266 1.103
2002 2.89 3.45 3.24 0.892 1.065 0.264 1.194
2003 3.39 3.69 4.01 0.845 0.92 0.278 1.088
2004 4.63 3.79 4.85 0.955 0.781 0.321 0.819


So it is pretty evident that the point about each player's age is relevant, as Millwood appears to be in gradual decline by DIPS. However, last year was the only season besides his injury-ruined 2001 where his ERC was over four, and until last season his ERC had outperformed his DIPS, indicating that the apparent decline in his balls in play skills last season was probably more a factor of sample size and luck. He has routinely had an ERA higher than his component ERA, so perhaps a projection for him should figure he's worse with runners on. However, the magnitude of that going forward probably won't be as large, so maybe DIPS/.95 is a decent baseline. That baseline would yield a 3.63, 3.88, and 3.99 ERA in each of the past three seasons, not park adjusted.

Now, let's look at Wade Miller:

ERC DIPS ERA ERC% DIP% BIPA D/E
2000 4.39 4.4 5.14 0.854 0.856 0.285 1.002
2001 3.6 4.39 3.4 1.059 1.291 0.251 1.219
2002 3.63 3.63 3.28 1.107 1.107 0.281 1
2003 3.71 4.02 4.13 0.898 0.973 0.274 1.084
2004 3.78 4.58 3.35 1.128 1.367 0.253 1.212


He sure outperformed his DIPS last season, didn't he? There's really not any pattern here that I can discern which would indicate that he has balls in play skills or pitching with runners on base skills that make DIPS a poor baseline of his performance. His BIP skills have never been below the DIPS baseline, so let's go ahead and be optimistic by using DIPS/1.05 as his baseline: his ERA's the past 3 seasons would be 3.56, 3.83, and 4.36.

Millwood pitched dozens more innings in each of the past three seasons, and his performance indicators aren't really worse than Miller's. Is Miller really the pitcher that you want? Certainly, there's an argument to be made. But if the Indians have reason to expect a full season from Millwood, then the Miller price to compare him to is $4.5 million.

Is Millwood worth $7 mil? Would that be a better deal than the Miller deal? The answers to each are probably not quite, but I'd have to do a lot more work to increase my confidence level on that. My point is, the answer to those questions lies in in-depth performance analysis, not VORP. So if you try to talk Shapiro down from the ledge, you better know why he's out there in the first place instead of inferring the answer is simply "experience."

Comments:
I would also have to add that Millwood's elbow ligament injury is at worst likely to result in a Tommy John surgery, the results of and recovery from which are well known. Miller's frayed labrum, on the other hand, has a much more problematic history of recovery. The Sox paid less for a reason, if only based on the type of injury.
 
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?