THE BOOK--Playing The Percentages In Baseball

Tuesday, March 23, 2010

Some basic zone-type defensive numbers for Jacoby Ellsbury

I am sending this data to Peter Abraham of the Boston Globe.  He phoned me the other day with some questions about defense.  He is doing a pre-season piece on the Red Sox and he wanted some fielding data on Jacoby that he could put into a graphic in the paper.

All I did was compare Ellsbury last year (2009) to all other center fielders (NL and AL) from 06-09 in terms of how many balls (FB, PU, and LD) he and they (all other center fielders) caught versus how many fell for hits or errors in 6 sections of the OF.  Those 6 sections are short RC, C, and LC, and medium/deep RC, C, and LC.  I used the retrosheet zones, 78S, 78M, 78D, 78XD, 8S, 8M, 8D, 8XD, 89S, 89M, 89D, and 89XD.  You can go to the retrosheet.org web site to see a diagram of those zones. I made no adjustments for park, batters, etc.  (Tango will like this.)

Short LC

All others: 920/15242/6.0%
JE: 3/105/2.9%

Medium/Deep LC

All others: 2409/4399/54.8%
JE: 16/113/14.2%

Short CF

All others: 14491/31308/46.3%
JE: 87/211/41.2%

Medium/Deep CF

All others: 28937/35932/80.5%
JE: 235/297/79.1%

Short RF

All others: 907/15519/5.8%
JE: 3/112/2.7%

Medium/Deep RF

All others: 2151/14285/15.1%
JE: 13/71/18.3%

Total actual outs minus total expected outs (based on all other CF’ers in all parks in the NL and AL combined, 06-09) is -23 outs.

Obviously comparing him to all other fielders in the NL and AL while he plays half his games in Fenway is not fair, as you can see from his out conversion rate in medium/deep LC compared to all other CF’ers in all other parks (NL and AL).

Here are the same numbers for Fenway Park only.  Again these are JE versus all others (not including himself). Keep in mind that the home player should outperform the road player (although “the others” includes other home CF’ers at Fenway besides JE), especially at Fenway.  By how much, I don’t know off the top of my head.

Fenway only

Short LC

All others: 16/539/3.0%
JE: 0/44/0%

Medium/Deep LC

All others: 42/495/8.5%
JE: 6/60/10.0%

Short CF

All others: 574/1090/52.7%
JE: 38/95/40.0%

Medium/Deep CF

All others: 808/1130/71.5%
JE: 130/167/77.8%

Short RF

All others: 69/505/11.4%
JE: 2/43/4.7%

Medium/Deep RF

All others: 67/348/19.3%
JE: 2/32/6.2%

Total actual outs minus total expected outs (based on all other CF’ers in Fenway Park only, 06-09) is -9 outs.

Finally, here is Jacoby in all other parks (not Fenway) in the AL versus all other CF’ers in all other AL parks, not Fenway.  Keep in mind that as a visiting fielder he is expected to do a little worse than all other fielders combined, of which around half are the home fielders and half are the visiting fielders.

Other AL parks only

Short LC

All others: 449/6796/6.6%
JE: 3/53/5.7%

Medium/Deep LC

All others: 1129/6243/18.1%
JE: 8/40/20.0%

Short CF

All others: 6505/13825/47.0%
JE: 47/112/42.0%

Medium/Deep CF

All others: 12489/15142/82.5%
JE: 95/118/80.5%

Short RF

All others: 376/6819/5.5%
JE: 1/57/1.8%

Medium/Deep RF

All others: 907/6142/14.8%
JE: 11/34/32.4%

Total actual outs minus total expected outs (based on all other CF’ers in the AL, not in Fenway Park, 06-09) is -3 outs.

In general, he appears to be above average at going back on balls and not very good coming in, at least in 2009.  Of course with such small samples, we don’t know the exact trajectory (hang time) of each ball, we don’t know the exact location on the field to which each ball was hit, we don’t know where Jacoby was playing, and we don’t know the weather conditions.  Without doing any adjustments, this “UZR-lite” (simple plus-minus ZR) assumes that all line drives and fly balls hit to a certain zone (regular UZR does not use such large zones of course) are created exactly equal, and that Jacoby and all other center fielders are playing in the same location in the OF at all times.

IOW, with such a small sample (one year of data), the UZR numbers may not represent what actually happened in 2009, and they may not represent his true defensive talent, regardless of how much of the numbers represented what actually happened.  Certainly his multi-year numbers are more likely to reflect his true defensive talent (as well as reflect what actually happened) than any one-year numbers.