Posted by MikeT23 on 12/6/2011 8:34:00 PM (view original):
Checking one team in a partial season is a little less of a sample size than 20+ seasons spread out over 4 worlds.

Just sayin'.

Posted by deanod on 12/6/2011 8:37:00 PM (view original):

Posted by MikeT23 on 12/6/2011 8:34:00 PM (view original):
Checking one team in a partial season is a little less of a sample size than 20+ seasons spread out over 4 worlds.

Just sayin'.

it's also probably just about as statistically relevant as the way you went about it.

Posted by tecwrg on 12/6/2011 1:02:00 PM (view original):

Posted by tecwrg on 12/6/2011 11:05:00 AM (view original):

Posted by MikeT23 on 12/6/2011 9:55:00 AM (view original):
I was looking at it on more of a percentage and just using my teams as an example.   Nonetheless, SS is the most important and 2B/CF can be debated for #2.  3B/CF could be debated for #4 and followed by LF, RF and 1B.    I think #1 and 6, 7, 8 are clear.

Using your formula against the most recent snapshot on the all-world statistical snapshot page, we get the following:

Pos	Inn	PO	A	E	+	-	Pctg	Weight
SS	17.915333	0.000	6.648	0.391	0.113	0.077	0.403509	4.92
2B	17.915333	0.000	5.606	0.186	0.080	0.058	0.331001	4.03
3B	17.915333	0.000	4.747	0.224	0.113	0.050	0.286570	3.49
CF	17.915333	4.526	0.116	0.115	0.071	0.058	0.272727	3.32
LF	17.915333	3.753	0.094	0.073	0.078	0.056	0.226287	2.76
RF	17.915333	2.750	0.111	0.051	0.053	0.068	0.169296	2.06
1B	17.915333	0.000	1.255	0.033	0.127	0.055	0.082053	1.00

The last column (weight) is the relative number compared to the lowest, i,e, first base.

Found an error in my formula, below are the correct numbers.  Relative order is unchanged, though the relative weightings compared to 1B move a little bit:

Pos	Inn	PO	A	E	+	-	Pctg	Weight
SS	17.915333	0.000	6.648	0.391	0.113	0.077	0.351	4.86
2B	17.915333	0.000	5.606	0.186	0.080	0.058	0.304	4.21
3B	17.915333	0.000	4.747	0.224	0.113	0.050	0.256	3.54
CF	17.915333	4.526	0.116	0.115	0.071	0.058	0.253	3.51
LF	17.915333	3.753	0.094	0.073	0.078	0.056	0.212	2.93
RF	17.915333	2.750	0.111	0.051	0.053	0.068	0.156	2.16
1B	17.915333	0.000	1.255	0.033	0.127	0.055	0.072	1.00

I was thinking about this while in the car earlier this evening.  These numbers put 3B and CF on near equal footing, which doesn't really feel right.  The I realized that it excludes the number of times the outfielders "touch" the ball by cleanly fielding base hits.  A poor fielding OF may bobble a base hit and that could be reflected as an error, but a clean handle of a base hit doesn't show up in any statistic.  This might have a net effect in undervaluing the outfielders a bit when compared to the infielders (whose clean "touches" usually show up as assists).

Also, first basemen are getting undervalued because, while you're eliminating IF putouts (under the assumption that most of them are catching balls throw to them), you're also eliminating the 3-U putouts.  Grounders to 2B, SS or 3B get captured because they are usually credited as assists, but grounders to 1B for the most part are not (other than 3-1 PO's or 3-6 FO's).  So this may skew the relative weights of everybody to 1B.

"SHOW YOUR WORK!!!!"

Posted by tecwrg on 12/6/2011 10:26:00 PM (view original):
This is interesting.  I just came across this by accident while looking something else up.  Bill James came up with the following "position value" for defensive positions that he uses in his "Similarity Score" calculation:

Catcher               20
Shortstop            14
Second Base     11
Third Base            7
Center Field         5
Right Field            4
Left Field               3
First Base             1
DH                         0  (duh!)

(found on page 91 of his book "Whatever Happened to the Hall of Fame?"

For HBD purposes, I'd flip 3B and CF.   Simply because of the "drop down" positioning I mentioned earlier.   Way more overqualified, in the field, 3B than CF.   And LF/RF.    RF needs a larger skill set(arm) but gets fewer opportunities.   Of course, at 4 to 3, there's not much difference anyway.

Posted by deathinahole on 12/7/2011 7:10:00 AM (view original):
deano would say "WTF does Bill James know. Hack!"