Graph of the Day II: 2009 All-Star Starting Player's UZR/150 Values

by Jeff Zimmerman (TucsonRoyal) on Jul 13, 2009 1:00 PM EDT Comment 18 comments

Here is a quick look at the UZR/150 values for the elected starters of the All-Star game along with the totals for each team. Since Pedoria (family) and Beltran (injury) will not be starting to the game but their replacements in the starting lineup have not been announced and when they are, I will add the information. The AL looks to be putting a stronger defensive team on the field if you use these numbers.

Raul Ibanez is moved to RF in the following chart and graphs

AL by you.

NL by you.

	National League Starters
Position	Name	Team	UZR/150
1B	Albert Pujols	STL	-2.3
2B	Chase Utley	PHI	1.8
SS	Hanley Ramirez	FLA	-0.4
3B	David Wright	NYM	-10.3
RF	Raul Ibanez	PHI	-0.4
CF	Carlos Beltran	NYM	-7.5
LF	Ryan Braun	MIL	-6.2
		Total =	-25.3
	American League Starters
Position	Name	Team	UZR/150
1B	Mark Teixeira	NYY	-4.3
2B	Dustin Pedroia	BOS	8.6
SS	Derek Jeter	NYY	1.4
3B	Evan Longoria	TB	14.1
RF	Ichiro Suzuki	SEA	13.4
CF	Josh Hamilton	TEX	19.9
LF	Jason Bay	BOS	-9.8
		Total =	43.3

0 recs | Comment 18 comments | Share

Story-email Email Printer Print

Comments

Over/under on the number of times the broadcasters talk about what good fielders Teixeira and Wright are?

SteakGrowsOnDmitri: /crams win into mouth
SteakGrowsOnDmitri: mmm, these s**ts taste like unfair an outmoded pitchin metrics

by natalys26 on Jul 13, 2009 1:02 PM EDT reply actions 0 recs

That's tough to criticize them for

They have been good UZR-wise throughout their careers (Wright pretty good, Tex very). It’s only been half a season of defense for them thus far this year anyway.

by Pat Andriola on Jul 13, 2009 1:42 PM EDT up reply actions 0 recs

I wonder how much this would change

If you did a 3-year weighted average or something. Pujols at -2.3 seems a little odd. From ‘06 through ’08 Pujols leads all 1B with +82 plays by Dewan’s plus/minus.

by Tommy Bennett on Jul 13, 2009 1:09 PM EDT reply actions 0 recs

I had thought of that, but just went with this year.

This is how they are voted on and thought the average would be like putting up their ZIPs projections

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

by Jeff Zimmerman (TucsonRoyal) on Jul 13, 2009 1:29 PM EDT up reply actions 0 recs

Yeah, I could see going either way, depending what you're addressing.

Judging the first half the season? Like this.

Projecting how each team will field in the game? A projection. You can go with Rally’s pre-season defensive projection or use a Fangraphs three-year calendar total for that.

It would be interesting to compare the ideal team each league could put on the field (pitchers for nine innings combined with pinch hitters for pitchers, pinch hitters for good-fielding position players, and the guys who both play the field and hit) to what ends up actually happening. I bet there’s a large WAR difference.

Beyond the Boxscore Not a member? Sign up.

by Sky Kalkman on Jul 13, 2009 3:44 PM EDT up reply actions 0 recs

but uzr isn't a way to measure a half-season at all.

there is so much noise in there the sample doesn’t tell you anything. it’s probably a little better than fielding percentage. why not put a chart together of the winning percentages of the pitchers?

chase utley has UZR/150’s of 10.2, 21.0, and 21.4 over the last three seasons. his 1.8 for this half-season is not representative of his true skill, and probably not of his actual performance over this half-season. pujols has a 4.5, a 16.0, and an 8.5 over the last three seasons; a -2.1 is not an accurate assessment. josh hamilton, on the contrary, has a net -2.8 career UZR/150 in the outfield.

this is a misuse of UZR to say something that it isn’t capable of saying (how someone fielded over a half season). i like UZR, but this article tends to distort its real content.

the truth can't hurt you, it's just like the dark/ it scares you witless, but in time you see things clear and stark -- macmanus

by tom s. on Jul 13, 2009 4:22 PM EDT up reply actions 0 recs

I disagree with the degree to which your implying UZR is useless over a half-season,

especially when you consider how unreliable even hitting stats are.

Again, I’m all for using projections and determining current true talent hitting and fielding levels for players. I think that approach is much more useful.

But if we’re trying to judge the past, we’re going to use straight wOBA or something like that. That contains a lot of flukiness, just like UZR, but it is what happened.

I realize UZR is only an educated guess as to what actually happened. But it’s not useless. Yes, it’ll miss completely sometimes. But compare it to hitting in terms of predictability:

http://www.hardballtimes.com/main/blog_article/how-reliable-is-uzr/

Over similar playing time, you’d regress wOBA 47%, infield UZR 64% and outfield UZR 74%. That’s in the ballpark of hitting.

Beyond the Boxscore Not a member? Sign up.

by Sky Kalkman on Jul 13, 2009 6:17 PM EDT up reply actions 0 recs

i had read the tht article. i'm not sure that i would wrap up the 47% and 64% and 74%

regressions in an “oh, close enough” ballpark. Requiring a 74% regression to the mean is a pretty weak correlation.

and there was no discussion of the need for regression in the post, nor of the pretty glaring inconsistency of the defensive metrics with the history of the individual players. josh hamilton is 27 runs better than carlos beltran in CF this year? i would think that would prompt a poster to at least provide a caveat.

last, the metric for whether the regression “worked” in the THT article was the “correlation between one year and the next.” if the original proposition was that accuracy is proving correlation between one year and the next, why regress to the league mean at all? why not use the large supply of prior data points in earlier years? i have to imagine that regressing a player to his own career average would more accurately depict what he’s doing than regressing him to the league average. i thought the THT article was more about how reliable SSS UZRs were for players who have short track records, or limited exposure at one position than in trying to describe an experienced players performance over a short period.

the truth can't hurt you, it's just like the dark/ it scares you witless, but in time you see things clear and stark -- macmanus

by tom s. on Jul 13, 2009 7:55 PM EDT up reply actions 1 recs

I agree that 1/2 of a year defensive stats are not the best measurement, but I have no problems posting them for a fake game that player might play 4 innings.

Maybe in the future I will post “projected” stats along with the yearly stats.

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

by Jeff Zimmerman (TucsonRoyal) on Jul 13, 2009 11:54 PM EDT up reply actions 0 recs

My initial thoughts on reading this article were along the same lines as tom s. – and I don’t think the argument is over projected vs. reality. The argument is the accuracy that UZR measures how a player has performed. With something like wOBA – we recorded exactly what happened (singles, 2b’s, 3b’s, etc), so regardless of how well it predicts future performances, we know exactly what the player has done in a given time period. The difference with UZR is there is a lot of noise in the measurement of what actually happened. So rather than the need to post “projections”, I would like to see a 3 year weighted average for the sake of accurately measuring what actually happened.

by Missing Barry on Jul 14, 2009 2:28 PM EDT up reply actions 0 recs

3 year weighted average for the sake of accurately measuring what actually happened.

That is exactly what projection systems are

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

by Jeff Zimmerman (TucsonRoyal) on Jul 14, 2009 2:51 PM EDT up reply actions 0 recs

With either lineup you could win the world series easily

Great Graph

I Have Spoken.

by The_Fan on Jul 13, 2009 7:02 PM EDT reply actions 0 recs

I hope to God that Bay doesn't pull an Uggla on us.

@bs_uf15bosox9be:OverTheMonster-ALLERGEN WARNING:May contain PB.

by bs.uf15bosox9bears23 on Jul 13, 2009 7:14 PM EDT reply actions 0 recs

3 years

Pujols 3306 innings 8.6 UZR/150
Utley 3292 innings 16.2 UZR/150
Ramirez 3285 innings -9.0 UZR/150
Wright 3610 innings 1.0 UZR/150
Ibanez 3018 innings -12.8 UZR/150 (In LF)
Beltran 3170 innings 2.5 UZR/150
Braun 2051 innings -3.2 UZR/150

+3.3 total.

Mark Teixera 3154 innings 1.8 UZR/150
Pedroia 3263 innings 7.2 UZR/150
Jeter 3263.2 innings -6.5 UZR/150
Longoria 1715 innings 17.0 UZR/150
Suzuki 3437 innings 8.5 UZR/150
Hamilton 1768 innings -5.7 UZR/150 (CF)
Bay 3344 innings -13.4 UZR/150

+8.9 total.

5.5 difference, not exactly the same as a 68.6 point difference.

Decrease runs scored?
Maybe.

Decrease winning? Never seen that proven.
-SFTU

by hazel on Jul 13, 2009 7:54 PM EDT reply actions 0 recs

I love when people take the effort to crunch the numbers themselves in addition to pointing out how things should be done.

Beyond the Boxscore Not a member? Sign up.

by Sky Kalkman on Jul 14, 2009 3:03 PM EDT up reply actions 0 recs

Accurate?

Aren’t the NY players’ stats off because of the new ballparks? I thought I read that UZR was still using the data from their old ballparks, not the new ones, It really wouldn’t make a difference for an IF like Wright, but certainly would affect Beltran.

by tarheelcoach on Jul 13, 2009 8:31 PM EDT reply actions 0 recs