UZR on Fangraphs

by azruavatar on Dec 7, 2008 2:10 PM EST Comment 18 comments

Use it. Love it. Can't beat it.

One of the best defensive metrics developed along with PMR and Dewan's +/- system, UZR is now available for free at Fangraphs.

Christmas has come early.

4 recs | Comment 18 comments

Story-email Email Printer Print

Comments

Great news! Surprising results?

I just posted this in another thread:

Looks like Berkman and Pujols are tied at 10.4 UZR/150 on the 2008 leaderboard. Huh?

Fielding Bible says Pujols, Teixeira, Kotchman, Pena, Berkman.
RZR says Pujols, Teixeira, Berkman.
Inaz says Teixeira, Pujols, Berkman.
PMR says Pujols, Pena, Berkman, Teixeira.

Orlando Cabrera is also leading SS – an even more stark disagreement with the other cited defensive metrics.

by astrostl on Dec 7, 2008 4:59 PM EST reply actions 0 recs

Some 2008 reference links

UZR/150 (Fangraphs)

RZR (Hardball Times)

Fielding Bible (Bill James Online)

Justin Inaz’s Total Value (jinaz-reds.blogspot.com)

PMR (Baseball Musings) – also partially converted to runs above average (Beyond the Box Score)

by astrostl on Dec 7, 2008 5:14 PM EST up reply actions 0 recs

Fielding Bible: Rollins has 88 voting points (1st), Cabera has 40 (6th)
RZR: Tejada is .879 (1st), Cabrera is .834 (11th)
Inaz: Scutaro is +20.5 field (1st), Cabrera is +2.4
PMR: Scutaro has a 110.16 out ratio (1st), Cabrera 101.14 (19th)

UZR/150: Tied with Rollins for the lead

by astrostl on Dec 7, 2008 6:47 PM EST up reply actions 0 recs

what's the different between

UZR and UZR/150?

this line is dedicated to '09

by Cards Fan in Chitown on Dec 8, 2008 1:43 AM EST up reply actions 0 recs

I think

I think UZR/150 is what your UZR would be if you played 150 games, so it is a rate stat vs a counting in a weird sort of way

"People call me El Hombre," Pujols said. "But only Stan is the Man."

by StLHugo on Dec 8, 2008 8:40 AM EST up reply actions 0 recs

right

UZR/150 – defense over the equivalent of 1 defensive season
UZR – what actually happened that season

Nice Robot

by azruavatar on Dec 8, 2008 8:51 AM EST up reply actions 0 recs

What they said.

It’s like reporting someone’s raw HR total, but also how many homeruns they would have hit in 700 PAs.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Dec 8, 2008 2:09 PM EST up reply actions 0 recs

Saw it this morning

it’s excellent. Defensive metrics are becoming easier and easier to come by.

by chuckb on Dec 7, 2008 6:34 PM EST reply actions 0 recs

Icky

Khalil Greene -6.2 runs. Man he was bad last year, the PTBNL better not be too good.

Not afraid to nitpick

by joker24 on Dec 7, 2008 9:59 PM EST reply actions 0 recs

Especially icky given that

he only played a little over 100 games. His UZR/150 is ~~9.3. +/~~ has him at -12, and PMR wasn’t very kind at 94.64.. At least Justin had him at +9.6.

Those Pilgrims ain't lookin' so proud now...

by giveml on Dec 8, 2008 12:13 AM EST up reply actions 0 recs

OK

so I guess bracketing an expression with minus signs is how you strike through something, huh? That should say his UZR/150 was a negative 9.3 and his +/- was a negative 12.

Those Pilgrims ain't lookin' so proud now...

by giveml on Dec 8, 2008 12:15 AM EST up reply actions 0 recs

RZR, POOZBIZ, UZR/150, PMR and Total Value

I have been building a spreadsheet of multiple defensive metrics trying to see who agrees with whom, whether or not that agreement is consistent, etc. and I think all I have succeeded in doing is confusing myself. Maybe somebody who cares about defensive metrics can help me out.

For my baseline I am using THT’s RZR stats for all qualifying players in both leagues and then I am adding in columns for POOZBIZ (Plays + OOZ / BIZ), UZR/150, PMR and the positionally adjusted defensive component of Justin’s Total Value numbers. Then I ran correlations by position to see what I could glean. I didn’t use +/- because I don’t have access to the complete data.

I was expecting the various metrics to disagree with each other to an extent, but I was surprised at how much the various correlations fluctuated by position. For example, at 1B (N=23) all the metrics have a .715 correlation or better with each other except for PMR and RZR, but for 2B (N=17) PMR was much more weakly correlated with everyone else.

Similary in the OF, the RF (N=15) numbers were very tightly grouped, but LF (N=12) and CF (N=20) there was significantly less agreement.

Anyone have any idea why this would be? All I am really trying to accomplish is figure out which metric I like best. I had been leaning toward UZR, but the absence of current UZR data is what got me started on this silly little project. Now that fangraphs is going to post the data with a nice, sortable interface and keep it more or less up to date during the season then I will probably just use that. FWIW, UZR seemed to correlate better with the rest of the field than any other metric.

Those Pilgrims ain't lookin' so proud now...

by giveml on Dec 8, 2008 12:08 AM EST reply actions 0 recs

Interesting

Just from comparing eyeball impressions to data, I’d prefer POOZBIZ. I suspect it is because it has an intelligent adjustment for one Achilles heel of zone ratings, player positioning.

by ol Pete on Dec 9, 2008 11:39 AM EST up reply actions 0 recs

It doesn't make any adjustments for player positioning.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Dec 9, 2008 7:09 PM EST up reply actions 0 recs

While I would agree

that POOZBIZ doesn’t make any “intelligent adjustments”, do you honestly believe that the very best any major league shortstop could do would be to fail to make a play on over 12% of the balls hit in his zone? Conversely, do you believe that major league shortstops routinely get to 40-70 balls that are outside their real zone?

I think POOZBIZ is very definitely a blunt instrument, but it does correlate very strongly with both +/- and UZR. Now that UZR is available at fangraphs I guess it doesn’t really matter, but when I came up with the stat I was annoyed that no good data was readily accessible and up to date. Problem solved. Maybe MGL read about POOZBIZ and just couldn’t stand it any more? ;~)

Those Pilgrims ain't lookin' so proud now...

by giveml on Dec 9, 2008 10:25 PM EST up reply actions 0 recs

I'm not knocking POOZBIZ. Just pointing out that it's not doing what ol pete thought it was. It's a good system for the data it takes in.

You are aware that most analysis using the RZR data from THT for the past couple years has used the POOZBIZ methodology, right?

Do you honestly believe that the very best any major league shortstop could do would be to fail to make a play on over 12% of the balls hit in his zone? Conversely, do you believe that major league shortstops routinely get to 40-70 balls that are outside their real zone?

Yes, I do, but you have to realize how STATS and BIS define their “zones”. When they collect data, the field is divided up into lots and lots of little zones. The data companies compute the percentage of balls in each little zone that are turned into outs. Each position’s large zone is defined to be all the little zones where balls are turned into outs at least 50% of the time. So some of those little zones have league-average zone-ratings of 95. Some are 80. Some are 55. When you starting including a few little zones where plays are made 55 of the time (like deep in the SS hole or up the middle), then you’ve got a lot of “missed” plays falling inside each position’s uber-zone. So yes, by that definition of zone, I believe fielders are not converting 12% of them into outs.

As far as OOZ plays go, there are many little zones that don’t have 50% conversion rates, say 25%. If a player is cheating to one side or another of his uber-zone, he’s bound to make a fair number of those over the course of a season.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Dec 10, 2008 8:31 PM EST up reply actions 0 recs

Yeah,

the only reason I have been talking about POOZBIZ was because most of the analysis you are talking about, at least to my limited knowledge, doesn’t seem to be freely available and/or up to date. Now that UZR is available I will be happy.

BTW, thanks for the excellent explanation of the zone computation methodology. I had seen the “little zones”, but not the >50% rule.

Those Pilgrims ain't lookin' so proud now...

by giveml on Dec 10, 2008 9:22 PM EST up reply actions 0 recs