Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Baseball Newsstand > Baseball Primer Newsblog > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Friday, April 06, 2018

Fangraphs: 2018 UZR and UZR Update!

The methodology has changed a little that allows UZR to account for some of the noise associated with imperfect data. The net result of this change is that extreme UZR’s, which were likely caused by, to some extent at least, noise in the data, rather than extreme performance, will be slightly ‘dampened.’ We think that these new values, while very close to the old ones in most cases, more accurately reflect the actual performance of the players in question.

I’ve always had a problem with trusting defensive metrics. The nature of the beast demands some level of uncertainty, and changing valuations — sometimes dramatically — does nothing to make me trust them more. The ‘dampening’ is particularly notable to players who have been hugely valuable based on these defensive metrics: Andrelton Simmons, Jason Heyward, Kevin Kiermaier, and several others. Adjust your Internet arguments accordingly.

Los Angeles El Hombre of Anaheim Posted: April 06, 2018 at 04:14 PM | 15 comment(s) Login to Bookmark
  Tags: defensive metrics, fangraphs

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. snapper (history's 42nd greatest monster) Posted: April 06, 2018 at 05:06 PM (#5649439)
This seems like quite a big deal.

A bunch of defensive superstars have taken a big value hit in this change, and a bunch a allegedly terrible defenders have improved.

e.g. Jason Heyward is now 48 runs worse for D on his career in UZR vs DRS. Kiermaier is 57 runs worse. Simmons is 81 runs worse.
   2. KJOK Posted: April 06, 2018 at 05:41 PM (#5649443)
Jason Heyward is now 48 runs worse for D on his career in UZR vs DRS. Kiermaier is 57 runs worse. Simmons is 81 runs worse.


MGL may come in and correct me, but I think this is partially due to the 'problem' of making this type of change on a seasonal basis. Because you 'dampen' on a season by season basis, Simmons loses 81 runs in total. BUT if his whole career was treated as one season, he might only get dampened by 20 runs, which would be closer to the 'truth'.



   3. Walt Davis Posted: April 06, 2018 at 07:06 PM (#5649458)
What were the UZR vs DRS differences for those players before the change in UZR? The tables present everybody adjusted by 5+ runs in a season. Heyward does lose 12 runs in 2012 but doesn't appear in any of the other tables and so must have lost less than 5 runs each for 2013-16. Roughly speaking, max loss for Heyward would be 28 runs over 5 years. He is still credited with saving 51 runs vs. an average RF (and occasional CF) for the years 2014-16. Even for 2012, based on these numbers, UZR used to credit Heyward with 7 more runs saved in 2012 than DRS did ... now it credits him with 5 fewer.
   4. Jose is an Absurd Force of Nature Posted: April 06, 2018 at 08:59 PM (#5649495)
Decreasing the numbers at the margins makes intuitive sense to me. Are there any players materially changed in terms of how they are graded defensively? E.g. is Kevin Kiermayer suddenly rated as a mediocre defender?
   5. PreservedFish Posted: April 06, 2018 at 09:08 PM (#5649505)
I think this aligns with everyone's impression that the huge outlier numbers were unrealistic. I am entirely convinced that Kiermaier is a superlative fielder, but the numbers on his BRef page are not credible ... a 5.0 dWAR seems basically impossible to me.
   6. Los Angeles El Hombre of Anaheim Posted: April 06, 2018 at 09:23 PM (#5649514)
I think this aligns with everyone's impression that the huge outlier numbers were unrealistic.
I don't know that everyone's impressions were that these numbers are unrealistic. Over the last decade, we've had a lot of BTF debates as to how historically good Andrelton Simmons or Jayson Heyward are, and now I'm wondering what would happen to the numbers of a guy like Andruw Jones if his numbers got "corrected."
   7. Mefisto Posted: April 06, 2018 at 09:32 PM (#5649525)
Just ask Chris Dial. :)
   8. ptodd Posted: April 07, 2018 at 01:34 AM (#5649641)
MGL has always said 1 year stats need to be regressed 50%. So he is doing it.

Until I see play logs and game logs for these stats I cant accept them. No way to validate them . If all you see are a cumulative number how do you know someone is not pulling them out of there arse. Not accusing anyone and I know MGL offers them for free, but there is a noticeable lacknof transparency. I dont know why WAR does not just average all the available stats for defense
   9. KB JBAR (trhn) Posted: April 07, 2018 at 02:30 PM (#5649723)
RE #5 and #6: Eyeballing by interested amateurs or casual data consumers probably isn't good enough to determine whether or not outliers are inaccurate or unrealistic. Back when UZR came out, I thought it was unrealistic because it had Erstad as a borderline MVP and, like, +40 in CF. Back then, sabermetric orthodox (or neo-sabermetric? or maybe uncritical pseudo-sabermetric?) was that defense didn't move the needle much. After a while people accepted that Simmons or Adam Everett were worth tons of runs on defense. It might be over-correction on my part, but this shift made me realize I'm not really capable of intuiting the run value of defense.

So I can agree with the sorting of defensive metrics, but be unsure of the magnitude in either direction. If I'm unsure of the magnitude, I don't know if trimming the extremes (or noise?) is right or not. I guess there might be a few ways to judge the realism of defensive numbers? First, verify that all positions roughly add up to about zero (at least over a few years). Maybe defense + pitching WAR should roughly equal hitting WAR (once you ensure your baselines are properly aligned)? I think Walt has used the Inside Edge stuff along with the value of offensive events to estimate roughly what a reasonable range might be. Maybe that'd be a worthwhile project for some outliers?

#8 How much does a three year average need to be regressed? Is the average of three individual years regressed 50% more accurate than the average of three years of raw data cumulatively regressed less than 50%?

   10. The Duke Posted: April 07, 2018 at 08:02 PM (#5649810)
Does this mean heyward can’t be a defensive specialist now?
   11. bfan Posted: April 07, 2018 at 08:17 PM (#5649812)
Says Theo Epstein "Now you tell me". Oh, well, what is $28,000,000.00 a year to a large market team.
   12. snapper (history's 42nd greatest monster) Posted: April 08, 2018 at 09:20 AM (#5649853)
I think this aligns with everyone's impression that the huge outlier numbers were unrealistic. I am entirely convinced that Kiermaier is a superlative fielder, but the numbers on his BRef page are not credible ... a 5.0 dWAR seems basically impossible to me.

Yup. I've been skeptical of any of the >+20, <-20 numbers for a long time. Chris Dial is probably the one who peaked my skepticism initially.
   13. puck Posted: April 08, 2018 at 12:44 PM (#5649875)
I don't remember what Arenado's UZR totals were before. He's not on any of the "changed by 5+ runs lists," but they're all 6.7 runs or lower since his 2nd year. BB-ref has him 57 runs higher 2014-2017.
   14. Walt Davis Posted: April 08, 2018 at 09:02 PM (#5650038)
Says Theo Epstein "Now you tell me". Oh, well, what is $28,000,000.00 a year to a large market team.

He's paid $21-22 M a year, not $28, on a back-loaded contract. (If you include the signing bonus, he was guaranteed 3/$78 before the first opt-out, which is still just $26 per year.)

As I noted, Heyward is, at worst, reduced by 28 runs over 5 years so, by new-UZR, was half-a-win per year worse than previously thought -- hardly a sea change that suggests he's a vastly worse player than we thought. For the last two years, Statcast rates him as 27 outs above an average OF ... an average RF is probably about an average OF so that probably puts him 27 outs ahead of an average RF as well. An extra out in the OF is probably worth about 1 run so, by Statcast, let's just call it 27 runs. UZR put him at +22, mainly due to far less credit for 2017; Rfield puts him at +36, mainly due to far more credit for 2017.

I think Walt has used the Inside Edge stuff along with the value of offensive events to estimate roughly what a reasonable range might be.

I've certainly used it at times -- it's very annoying to work with. Whether it is "reasonable" or not is another question. It produces very few extreme values -- I'm not sure I've ever seen anything bigger than +9. Even statcast produces more extreme values than that. The main justification that it almost has to be reasonable is that it codes a huge proportion of FBs as "routine" and these are converted 98-99% of the time by pretty much every OF. For 2017, IE says that Heyward faced only 36 non-routine BIP, converting 20 of them which is probably no more than 6-7 extra compared with an average RF/CF.

So by IE, almost no OF contributes anything important defensively. That doesn't seem "reasonable" but it does seem to do a good job of identifying "routine" and it records the same number of plays made so it's hard to see what it could be missing. The only way I can reconcile this with something like statcast is that, for very good OF like Heyward, too many balls are getting classed as routine ... but I'm pretty sure I've looked at a few examples and there's no obvious evidence that's happening.



   15. bfan Posted: April 09, 2018 at 07:42 AM (#5650122)
(If you include the signing bonus, he was guaranteed 3/$78 before the first opt-out, which is still just $26 per year.)


Humble apologies for over-stating Heyward's contract which I valued currently at $28 million per year. At $26 million per year, he is a bargain.

You must be Registered and Logged In to post comments.

 

 

<< Back to main

News

All News | Prime News

Old-School Newsstand


BBTF Partner

Dynasty League Baseball

Support BBTF

donate

Thanks to
Darren
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

NewsblogYelich, Betts Win MVPs
(61 - 12:23am, Nov 18)
Last: What did Billy Ripken have against ElRoy Face?

NewsblogOT - November* 2018 College Football thread
(283 - 12:17am, Nov 18)
Last: Lance Reddick! Lance him!

NewsblogIndians' Trevor Bauer pleads his own Cy Young case using a spreadsheet on Twitter
(27 - 11:48pm, Nov 17)
Last: ERROR---Jolly Old St. Nick

NewsblogMichael Wilbon Weighs In On Jacob deGrom With Worst Baseball Take Of Year | MLB | NESN.com
(15 - 11:44pm, Nov 17)
Last: ERROR---Jolly Old St. Nick

NewsblogMarlins get rid of orange, cite South Florida cultures with new look
(36 - 11:39pm, Nov 17)
Last: phredbird

NewsblogOT - NBA Thread (2018-19 season kickoff edition)
(2518 - 11:09pm, Nov 17)
Last: Booey

NewsblogSale of Baseball Prospectus
(341 - 10:41pm, Nov 17)
Last: You Know Nothing JT Snow (YR)

NewsblogOT - Catch-All Pop Culture Extravaganza (November 2018)
(440 - 10:06pm, Nov 17)
Last: RJ in TO

NewsblogOT: Soccer Thread (2018-19 season begins!)
(1220 - 10:05pm, Nov 17)
Last: Count Vorror Rairol Mencoon (CoB)

NewsblogZack Greinke trade makes sense for these teams
(36 - 9:32pm, Nov 17)
Last: McCoy

Gonfalon CubsNow what?
(233 - 7:44pm, Nov 17)
Last: Walt Davis

NewsblogCarson Cistulli From Fangraphs Joins the Blue Jays
(9 - 7:24pm, Nov 17)
Last: GGC

NewsblogBen Zobrist, start designing your customized cleats. MLB grants players more shoe freedom.
(2 - 3:23pm, Nov 17)
Last: Sunday silence

NewsblogMike Elias Orioles general manager | MLB.com
(20 - 2:47pm, Nov 17)
Last: QLE

NewsblogHow Kevin Brown Became Baseball's First $100 Million Man
(10 - 1:36pm, Nov 17)
Last: Adam Starblind

Page rendered in 0.3077 seconds
46 querie(s) executed