Hoban - A player ranking (August 8, 2003)
Comments?
--posted by TangoTiger at 11:03 AM EDT
Posted 11:55 a.m.,
August 8, 2003
(#1) -
James Fraser(e-mail)
(homepage)
I'll offer this up too... From Hoban's author comments section at the onlince bookstore selling his product:
"
Bill James on Evaluating Defense and the Author’s Efforts
SABR-L is the official website of SABR and the #1 place where serious baseball research is discussed by serious researchers. (It must be if baseball’s #1 analyst will participate.) During February/March of 2003, a discussion was held by members regarding WIN SHARES, the new book by Bill James. During that discussion, Prof. Hoban expressed some reservations about the way Mr. James assigned offensive and defensive values to some players.
Bill James posted some comments about this discussion. They read in part:
"Mr. Hoban believes (as I understand it) that he has a more compelling logic to assess offensive and defensive value. Well, somebody will, sooner or later; I wish him luck. I think that, as a whole--as a research community--we're finally getting some traction in this area; we are finally beginning to inch toward estimates of defensive value which will stand up to rigorous analysis. It is my impression that we still have a very long way to go, and that there are still some pretty basic questions remaining to be answered." "
Posted 12:08 p.m.,
August 8, 2003
(#2) -
Patriot
That's the kind of inorganic junk that makes sabermetricians look like a bunch of nerdy number crunchers.
Posted 12:15 p.m.,
August 8, 2003
(#3) -
Justin Kubatko(e-mail)
SABR-L is the official website of SABR [...]
SABR-L is a mailing list, not a web site. The Society's official web site is www.SABR.org.
Don't waste your time with this guy. His work has been discussed ad nauseum on SABR-L in the past.
Posted 1:33 p.m.,
August 8, 2003
(#4) -
tangotiger
Hoban was kind enough to reply back to my email, pointing him to UZR, which he is aware of.
I replied:
=========
In order to evaluate Jeter, Nomar, and ARod, you would be forced to use play-by-play data, since that would contain all the data available for the players that you are evaluating.
The less data you decide to use, the larger the margin of error being introduced.
If the intent is to compare Arod to Ozzie to Honus Wagner, it is not clear that using the same data would be the best way. What you can say is that by using UZR, you can say something like Arod is +10 relative to average, +/- 3 runs, and by using the basics, you can say Honus Wagner is +12 relative to average, +/-8 runs (or something along those lines).
While I see the point in using the same metric for all players, you would at least have to use the most advanced metric to baseline the more basic metrics against. The validation of the basic version must be done against the more advanced metrics, with a certain margin of error.
Just my 2 cents.
============
Posted 5:38 p.m.,
August 8, 2003
(#5) -
David Smyth
What a waste of time "Professor" Hoban is.
Posted 9:13 a.m.,
August 12, 2003
(#6) -
tangotiger
I posted this at Clutch:
Just because you do 5*fld% + RF or whatever, does NOT make one thing 5 times more important. This is the same kind of explanation that people give about OPS.
BEcause these things are different scales, the multiplier does not and cannot imply importance.
What that multiplier does is streeeeetch out the range of performance.
If for example the fld% stretches from .950 to .990, that's a .040 swing. If RF stretches from 4.00 to 6.00, that a 2.00 swing.
By multiplying fld% by 5, you are increasing the swing from .040 to .200. The RF swing is still much much larger.
I don't know what Hoban's exact equation is, but, please keep this in mind when you think of "importance" and matching it to the "multiplier".