Baseball Musings: Probabilistic Model of Range, Second Basemen, 2006

November 26, 2006

Probabilistic Model of Range, Second Basemen, 2006

As with the left fielders, I'll present both the velocity model and the velocity/distance model. We'll start with the old velocity model:

Probabilistic Model of Range, Second Basemen. Model is Based on 2006 Data Only. Minimum 1000 Balls in Play. Uses Velocity for Fly Balls.
Player	In Play	Actual Outs	Predicted Outs	DER	Predicted DER	Difference
Tony Graffanino	1702	186	161.05	0.109	0.095	0.01466
Neifi Perez	1374	166	152.40	0.121	0.111	0.00990
Jamey Carroll	2806	396	372.04	0.141	0.133	0.00854
Joe S Inglett	1349	157	145.78	0.116	0.108	0.00832
Orlando Hudson	4128	552	522.65	0.134	0.127	0.00711
Aaron W Hill	2777	358	340.66	0.129	0.123	0.00625
Jose Valentin	2367	316	301.96	0.134	0.128	0.00593
Mark Ellis	3407	390	370.15	0.114	0.109	0.00583
Jose C Lopez	4045	473	451.94	0.117	0.112	0.00521
Placido Polanco	2838	373	359.11	0.131	0.127	0.00489
Luis Castillo	3663	416	401.24	0.114	0.110	0.00403
Chase Utley	4151	476	460.10	0.115	0.111	0.00383
Robinson Cano	3160	385	372.91	0.122	0.118	0.00383
Chris A Burke	1012	128	124.23	0.126	0.123	0.00373
Dan C Uggla	3935	485	473.20	0.123	0.120	0.00300
Tadahito Iguchi	3782	428	418.52	0.113	0.111	0.00251
Josh L Barfield	3755	442	435.89	0.118	0.116	0.00163
Jose Castillo	3832	387	381.40	0.101	0.100	0.00146
Brandon Phillips	3791	404	399.50	0.107	0.105	0.00119
Marcus Giles	3589	412	413.47	0.115	0.115	-0.00041
Mark Grudzielanek	3595	367	370.61	0.102	0.103	-0.00100
Mark Loretta	3578	401	410.00	0.112	0.115	-0.00251
Brian Roberts	3634	398	407.17	0.110	0.112	-0.00252
Ray Durham	3525	393	402.11	0.111	0.114	-0.00258
Adam Kennedy	3386	406	415.75	0.120	0.123	-0.00288
Ian M Kinsler	3288	424	433.63	0.129	0.132	-0.00293
Aaron Miles	2016	238	245.21	0.118	0.122	-0.00358
Ronnie Belliard	3860	448	464.43	0.116	0.120	-0.00426
Jeff Kent	2811	325	338.95	0.116	0.121	-0.00496
Craig Biggio	3162	360	376.73	0.114	0.119	-0.00529
Hector Luna	1487	151	159.76	0.102	0.107	-0.00589
Rickie Weeks	2402	263	278.24	0.109	0.116	-0.00634
Kaz Matsui	1403	169	178.92	0.120	0.128	-0.00707
Jose Vidro	2905	305	327.93	0.105	0.113	-0.00789
Jorge L Cantu	2859	283	311.98	0.099	0.109	-0.01014
Ty Wigginton	1075	105	117.30	0.098	0.109	-0.01144
Todd Walker	1279	128	147.72	0.100	0.115	-0.01542

And here's the mixed velocity/distance model:

Probabilistic Model of Range, Second Basemen. Model is Based on 2006 Data Only. Minimum 1000 Balls in Play. Uses Distance for Fly Balls.
Player	In Play	Actual Outs	Predicted Outs	DER	Predicted DER	Difference
Tony Graffanino	1702	186	165.06	0.109	0.097	0.01230
Jamey Carroll	2806	396	370.20	0.141	0.132	0.00920
Aaron W Hill	2777	358	334.22	0.129	0.120	0.00856
Orlando Hudson	4128	552	520.38	0.134	0.126	0.00766
Jose Valentin	2367	316	301.32	0.134	0.127	0.00620
Mark Grudzielanek	3595	367	344.87	0.102	0.096	0.00616
Chase Utley	4151	476	451.34	0.115	0.109	0.00594
Joe S Inglett	1349	157	151.57	0.116	0.112	0.00403
Mark Ellis	3407	390	376.45	0.114	0.110	0.00398
Neifi Perez	1374	166	160.56	0.121	0.117	0.00396
Jose C Lopez	4045	473	457.26	0.117	0.113	0.00389
Robinson Cano	3160	385	373.02	0.122	0.118	0.00379
Luis Castillo	3663	416	403.31	0.114	0.110	0.00346
Placido Polanco	2838	373	363.35	0.131	0.128	0.00340
Chris A Burke	1012	128	125.30	0.126	0.124	0.00267
Brandon Phillips	3791	404	397.11	0.107	0.105	0.00182
Jose Castillo	3832	387	384.20	0.101	0.100	0.00073
Tadahito Iguchi	3782	428	425.35	0.113	0.112	0.00070
Dan C Uggla	3935	485	482.86	0.123	0.123	0.00054
Brian Roberts	3634	398	396.92	0.110	0.109	0.00030
Josh L Barfield	3755	442	441.53	0.118	0.118	0.00013
Ian M Kinsler	3288	424	426.80	0.129	0.130	-0.00085
Adam Kennedy	3386	406	409.98	0.120	0.121	-0.00118
Marcus Giles	3589	412	417.71	0.115	0.116	-0.00159
Ray Durham	3525	393	401.46	0.111	0.114	-0.00240
Craig Biggio	3162	360	369.43	0.114	0.117	-0.00298
Jeff Kent	2811	325	335.51	0.116	0.119	-0.00374
Mark Loretta	3578	401	415.16	0.112	0.116	-0.00396
Aaron Miles	2016	238	246.01	0.118	0.122	-0.00397
Hector Luna	1487	151	157.64	0.102	0.106	-0.00447
Rickie Weeks	2402	263	275.27	0.109	0.115	-0.00511
Ronnie Belliard	3860	448	472.37	0.116	0.122	-0.00631
Kaz Matsui	1403	169	179.09	0.120	0.128	-0.00719
Jose Vidro	2905	305	327.48	0.105	0.113	-0.00774
Jorge L Cantu	2859	283	307.27	0.099	0.107	-0.00849
Ty Wigginton	1075	105	116.62	0.098	0.108	-0.01081
Todd Walker	1279	128	144.79	0.100	0.113	-0.01313

I'm starting to like the mixed model more. Grudzielanek does better in the second chart. He won the gold glove and did very well in John Dewan's +/- system. Both system agree on the bottom five. Among the second basemen who play every day, Orlando Hudson comes out on top in both systems. I believe he came out near the top last year as well.

Posted by David Pinto at 05:50 PM | Probabilistic Model of Range | TrackBack (0)

Comments

As a Jays fan it's friggin' GREAT to see Aaron Hill just a notch behind Hudson. Shame they're going to move him to SS.

Posted by: Jurgen at November 26, 2006 07:21 PM

i'm not surprised to see mark loretta so low. he seemed to have pretty poor range, despite his good hands. so, i'm keeping score at home, and i noticed that the red sox came out as just slightly above neutral in groundball defense, and so far, without 3rd base, they're at about neutral. so does that mean that lowell is going to be about average? every other statistic i've seen has shown that he had a phenomenal season in the field. i've seen figures like +15 to 20 runs. i guess this is kind of jumping the gun.

Posted by: amos at November 26, 2006 09:54 PM

Dave, I like the site. However, I am consistently confused what this information means. Many times when I come to your site, I skip over all this hard work because it makes absolutely no sense to me. Could you post an explanation of these models.

Posted by: Bob at November 27, 2006 08:13 AM

Sorry Pinto but your system is WHACKED this year. Perhaps it is the data you are receiving, I don't know. But you have some players WAYYY too high for their production last year.

Just an opinion

Posted by: d-man at November 27, 2006 10:23 AM

Some D'Backs fans were apparently less than enamoured of Hudson's defence this year, and Chris Dial's system has him as below league average this year. Could you please let me know what the results were for groundballs only. Some have suggested that Hudson makes a significant part of his numbers by pop-fly hogging.

Posted by: Mike Green at November 27, 2006 10:50 AM

I think the presentation of this data would be greatly improved by dropping the last 3 columns -- DER, predicted DER, and difference -- and replacing them with two columns that are more helpful at the player level:
1) Out difference (actual minus predicted outs)
2) Out percentage (actual outs divided by predicted outs)

DER is great at the team level, both overall and separately for GBs and airballs. We know what it means. But it has no intuitive meaning at the player level: no one "expects" Hudson to make outs on 12.6% of all balls in play when he's in the field. However, saying that a player makes 110% or 92% as many outs as the average player at his position has clear meaning. So does saying "he made an extra 8 outs," or "he cost his team 12 outs."

Posted by: Guy at November 27, 2006 11:19 AM

I completely agree with Guy, and said as much on my blog.

Posted by: tangotiger at November 27, 2006 02:19 PM

re: chase utley

nice to see a guy with a great glove and a great bat.

if only the phils had held onto scott rolen, what an infield they'd have of Howard, Utley, Rollins & Rolen. It's be better than Connie Mack's of the 1910s.

--art kyriazis, philly

Posted by: art kyriazis at December 1, 2006 05:14 PM