<$BlogRSDURL$>

Monday, November 22, 2004

PROJECTION REVISIONS
(I have a comparison to ZiPS at the end of this post.)

I have that row for projections when I do my player reviews. It's kind of there as a junk thing, but also just to see if it works. It never had any age adjustments, but finding this data on Tangotiger's site gave me an idea of how to work it. So here are revised projections for everyone I've done so far, except for Alfredo Amezaga because no one gives a damn:

           G    AB   H   2B  3B  HR  SO  BB  SB  CS  AVG  OBP  SLG  

Anderson 140 563 168 37 2 22 82 30 5 3 298 331 489
DaVanon 87 228 61 10 2 7 44 33 12 3 268 357 420
Eckstein 139 542 146 23 2 4 48 42 17 7 269 339 338
Erstad 116 464 127 22 2 7 67 32 15 2 275 325 373
Glaus 98 351 87 18 1 22 86 56 6 3 248 354 499
Everyone loses a little bit once age is accounted for, which makes sense (though Glaus moving from 27 to 28 is no biggie).

Again, this is pretty much in fun, to see if it works and because I'm not afraid of looking like a moron.

You can find the original explanation of what I'm doing here. If you care about this sort of stuff, read on; if not, feel free to check out (I won't be offended).

To begin with, I take a weighted average for each player of the last four years in each category. So that works out as 2004 x 4, 2003 x 3, etc. This established the base level.

Next, I take Tango's age adjustments. I have to modify them to make them work ... for instance, BB/AB goes from .660 at age 21 to .800 at age 24 (what that means is that, at age 24, the average player has reached 80% of the high he will reach in his career for BB/AB; BB/AB actually peaks at age 37, per this data). So I take the four-year weighted average of that, as I described above ... so the four-year weighted average for BB/AB after age 24 is .754.

So it's .754 for age 24; it's .794 for age 25. That's an increase of 5%. So, let's say we had a player who had a four-year weighted average of .100 BB/AB through 24 years of age. I would expect that to improve by 5% in the age 25 year, so his BB/AB would be projected to be .105.

So you can run this through all the categories, and use some algebra, some double-checking, and eventually you come up with a projected line. All you're doing is predicting the predictable, by finding a player's baseline performance and adjusting for age. Garret Anderson's baseline AVG/OBP/SLG for 2001-2004 is 305/338/503. He's going to be 33 for 2005, so you would have to expect some decline there, so we come up with a line of 298/331/489.

I don't have the means to figure out park factors on this, but since I'm only doing Angel players, that should come out pretty fine.

UPDATE: I had looked at them when they were posted, but Richard's link today reminded me to check in with Dan Szymborski's ZiPS predictions over at BTF.

Just looking at rates:

                 Me            Szymborski

Player AVG OBP SLG AVG OBP SLG
Anderson 298 331 489 299 333 488
DaVanon 268 357 420 280 367 436
Eckstein 269 339 338 276 343 339
Erstad 275 325 373 271 325 373
Glaus 248 354 499 258 364 471
Wow, the Anderson and Erstad predictions are kinda scary, and Eckstein is damn close as well. There is a bit of a difference on DaVanon, though it's mostly batting average (adding 12 points to my projections gets 280/369/432, very close to Dan's). The big difference we have is on Glaus' power, and my projection is a lot closer to his career mark than Dan's is.

Comments: Post a Comment

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