Questions about Analyzing Defense

by nwroyal on Nov 8, 2009 10:37 AM EST Comment 17 comments

I've been looking at Alberto Callaspo's UZR numbers and I've got some questions about how to use UZR to project for the coming year.

Fangraphs has been reviewing the best fielders from 07-09--pointing out that one season of UZR is equivalent to about 1/3 season of batting (in other words, it takes 3 seasons of UZR to have a good grasp on a players ability. Last year Callaspo was not good (-7.3), in 2008 he was above average (+1.9 although it was in about 1/4 the innings). Should we view Callspo as a -5.4 (by adding 08 and 09 together)? Should we give the 09 numbers more weight?

Or looking at Yuniesky Betancourt. His UZR has gone: 07=-1, 08=-13, 09=-21. For 2010--should we project an average of those 3 numbers or do we follow the progression and project an unfathomable -30?

I know there probably aren't concrete answers to this but I'm curuious to hear people's opinions.

1 recs | Comment 17 comments

Story-email Email Printer Print

Comments

And dont forget the anomaly

that was Jeter’s year defensively.

This space for rent.

by averagegatsby on Nov 8, 2009 11:03 AM EST reply actions 0 recs

I think its good to use as much of the data as possible

while also looking at scouting, to paint the whole picture

but as for trying to project, in terms of getting an actual number, that’s tough to do… guys tend to post defensive numbers all over the place during their careers

by royalsreview on Nov 8, 2009 1:46 PM EST reply actions 0 recs

I can step you through it.

I got this from MGL so it is pretty sound.

You need to do a 5-4-3-2 split regressed to 125 games worth of data. Here is the formula

=(1-(125)/(125+(sum of games over last 4 years))
(
(5UZRYear1*GamesYear1

This value will really regress to the league average of 0. Any value over 5 in either direction is pretty rare.
4*UZRYear2*GamesYear2
+3*UZRYear3*GamesYear3
+2*UZRYear4*GamesYear4)/
(5*GamesYear14*GamesYear2+3*GamesYear3+2*GamesYear4)
) – Aging constant

Aging constant:
for players under 36 subtract 0.2 from the number for aging.
If 36 or older subtract 1.0 for aging.

Where Year1: Most recent year’s stats.

I am actually running league wide UZR projections this year and want to get them down within a week or two.

Jeff Zimmerman - Protecting the world from RBI's and Wins from my mom's guest house.

by Jeff Zimmerman (TucsonRoyal) on Nov 8, 2009 2:13 PM EST reply actions 0 recs

Here are the projected values for some of the recent greats at SS for the Royals:

Name, projected UZR and games at position

Tony Pena 5.7 306
Mike Aviles 4.7 125
Willie Bloomquist -1.0 87
Angel Berroa -1.5 222
Yuniesky Betancourt -9.2 595

Jeff Zimmerman - Protecting the world from RBI's and Wins from my mom's guest house.

by Jeff Zimmerman (TucsonRoyal) on Nov 8, 2009 2:45 PM EST up reply actions 0 recs

Is that UZR/150 or straight UZR

I do something like that, but I use UZR/150, which I assume takes the player’s rate for each “zone” or type of play and adjusts it for the average amount of those sorts of plays the player would be expected to make over 150 games.

I'm not a sabermetrician, but I do play one at Driveline Mechanics.

Can't get enough of me? Check out my Twitter feed.

by devil_fingers on Nov 8, 2009 4:47 PM EST up reply actions 0 recs