It's not fair to Coco

My look at why Coco's Defensive ranking of F- from The Hardball Times was pretty rough on him and what should be done differently.

over 3 years ago Jeff Zimmerman 31 comments 0 recs |

Story-email Email Printer Print

Comments

your way of looking at it is probably better than my way

Founder of the Johnny Giavotella fan club.

by doublestix on Feb 13, 2009 12:09 AM EST reply actions

What would you have done?

by Jeff Zimmerman on Feb 13, 2009 12:21 AM EST up reply actions

nothing

i would have just bitched that defensive metrics are still pretty unreliable.

i don’t know how it’s possible to see Coco as a bad defender.

Founder of the Johnny Giavotella fan club.

by doublestix on Feb 13, 2009 12:24 AM EST up reply actions

the bUZR numbers don’t necessarily say he’s a bad defender, really, they do say he had a bad year this year. In all the other seasons where he had a decent sample size in CF, he’s above average, except for 2006, when he was right around average.

Analogously, we know that Adam Dunn is an outstanding hitter, but just because his wOBA was a bit lower in 2006, we don’t through out wOBA, we simply take it as one data point and weigh is appropriately against others.

Bringing you more-or-less replacement level analysis and commentary since sometime in 2008.

by Matt Klaassen on Feb 13, 2009 12:30 AM EST up reply actions

It is sort of nice to have some kind of metric ...

but single season data is just too variable to be considered. I have thought some more and think I went pretty light on The Hardball Time. A variation of +/- 12 runs would account for 95% of all players. When the metric basically goes from ~~20 to +20, one person could go from +12 (A defender) to -12 (F~~ defender). I think UZR needs to be handled like park factors, average the values over the years. The more data the better.

by Jeff Zimmerman on Feb 13, 2009 12:32 AM EST up reply actions

didn't mean to cross out sentence and don't know how to correct

by Jeff Zimmerman on Feb 13, 2009 12:33 AM EST up reply actions

I think you need to distinguish between true talent and actual performance

and the relation between the two. Yes, there is probably a larger gap between defensive stats and true talent than there is between, say, wOBA and true hitting talent, and that’s why Dewan, Tango, MGL, etc. all say that you need three years of data to start getting reliably at true talent,

Are you saying that one year’s defensive stats have little to no bearing on actual performance, though, “true talent” aside?

A better example on offense (and again, I hesitate to stat this as any more than an analogy) might be J. D. Drew’s 2007 when he wOBAed .346 vs. his career .386. Yes, we know that he was/is probably a better hitter than that, but we don’t say “well, his wOBa was just wrong that year.”

For what it’s worth, I’m pretty sure the THT defensive ratings are based off the RZR/OOZ system. Like a lot of not-quite-PBP systems, from what I’ve seen they tend to be harsher on guys with down years generally. One can just compare, say, it’s F- for Alex Gordon in 2008 with his -3.9 bUZR (which is in the “slightly below average” range).

Bringing you more-or-less replacement level analysis and commentary since sometime in 2008.

by Matt Klaassen on Feb 13, 2009 12:55 AM EST up reply actions

Did THT's grade come from looking at 2008 numbers only?

The immoderate moderator

by Scott McKinney on Feb 13, 2009 12:50 AM EST reply actions

yes, RZR/OOZ

And like I said above, taken for a single season, I don’t think that’s necessarily the end of the world.

Bringing you more-or-less replacement level analysis and commentary since sometime in 2008.

by Matt Klaassen on Feb 13, 2009 12:55 AM EST up reply actions

THT should be smarter than that

Even for a regular playing almost every day, one season is a pretty small sample size for fielding stats. And, of course, Crisp played for less than a full season of regular time. So projecting from that small of a sample size is pointless and likely to often point you in the wrong direction. And RZR/OOZ isn’t exactly the best system to describe or project defense. It’s just what they have, like with BP and FRAA/FRAR. Not that those two systems are equally bad; they aren’t.

The immoderate moderator

by Scott McKinney on Feb 13, 2009 1:03 AM EST up reply actions

it's not a projection

it’s an evaluation of the previous single season’s performance, in the same way that “.319” describes his 2008 wOBA, whether that reflects his true offensive ability or not.

Or better yet, in the case of RZR/OOZ, his .283 batting average.

Bringing you more-or-less replacement level analysis and commentary since sometime in 2008.

by Matt Klaassen on Feb 13, 2009 1:06 AM EST up reply actions

Ok, that makes more sense

The immoderate moderator

by Scott McKinney on Feb 13, 2009 1:11 AM EST up reply actions

Excited about the crazy WAR fantasy league?

And are coming up with 2009 WAR projections for hundreds of players, based on PECOTA/CHONE? I don’t think I’m up to that level of work.

The immoderate moderator

by Scott McKinney on Feb 13, 2009 1:13 AM EST up reply actions

I think I'm in the same boat as you

I was excited for about 10 seconds after I realized I got in, then suddenly realized “oh sh-t, this is going to be about 3,000x more complicated than any other fantasy league I’ve ever been in,” and not sure it’s worth the effort.

Maybe I’ll just draft all Royals. Should Farnsworth or Jacobs be my No. 1?

At least we’re in difference divisions (and how is that going to work? different draft pools?) Maybe we’ll meet in the WAR World Series.

Bringing you more-or-less replacement level analysis and commentary since sometime in 2008.

by Matt Klaassen on Feb 13, 2009 1:16 AM EST up reply actions

At least we’re in difference divisions (and how is that going to work? different draft pools?) Maybe we’ll meet in the WAR World Series.

Yeah, I think we draft separately. But I’m glad there isn’t inter-league play, because it would ruin the specialness of the WAR World Series.

If I have the time, I think I’m going to use Fangraph’s downloadable WAR leaderboard for 2008, plug in as many 2009 salaries I have time for and then let Excel calculate a WAR/dollar metric for me. That’s a good starting place. And then since we have so much time to make our draft picks, I can see who is left and tinker with the WAR numbers for a handful of likely candidates to get something which is a more likely 2009 projection…and then make my pick off of that.

The immoderate moderator

by Scott McKinney on Feb 13, 2009 1:22 AM EST up reply actions

"Yeah, I think we draft separately. But I’m glad there isn’t inter-league play, because it would ruin the specialness of the WAR World Series."

Awesome.

If I were participting, I would definitely have a list of projected WAR, then go down the list and cross of guys I KNOW have bad contracts. Then highlight really good contracts. Then put players into positions and try to find tiers within each position of good $$$/WAR deals.

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

by Sky Kalkman on Feb 13, 2009 8:39 AM EST up reply actions

Yeah, but a .283 describes actual results

A UZR doesn’t describe exactly what happened, just a decent guess. There’s plenty of room for plays of moderately different difficulties to be counted as the exact same pay.

observed variation = talent variation + metric noise variation + random variaton

AVG doesn’t stuggle with that middle one, UZR does, and something like ERA really does.

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

by Sky Kalkman on Feb 13, 2009 8:35 AM EST up reply actions

Agreed

that’s why I said it was analogous, not “just like”

Bringing you more-or-less replacement level analysis and commentary since sometime in 2008.

by Matt Klaassen on Feb 13, 2009 12:08 PM EST up reply actions

also, in the case of Crisp

I think it says something that although the exact numbers differ, almost all the defensive metrics I’ve seen say he was great in 2007, and had a bad 2008. Almost all also agree that he wasn’t as good in 2006 as he was in 2007 (except PMR, which thought 2006 was great, it’s been a while since the trade, when I went through all this)

Bringing you more-or-less replacement level analysis and commentary since sometime in 2008.

by Matt Klaassen on Feb 13, 2009 12:10 PM EST up reply actions

I really don't get

Why they only used one year. Wasn’t Crisp excellent in 2006 and 2007 in UZR?

Relive Royals History at royalsretro.blogspot.com

by RoyalsRetro on Feb 13, 2009 9:52 AM EST reply actions

Can someone clarify the context in which these rankings occurred?

What sort of batting data was presented? Were there forecasts, too?

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

by Sky Kalkman on Feb 13, 2009 10:02 AM EST up reply actions

Let me repeat this, in ALL CAPS, so everyone can hear it

THIS ISN’T A PROJECTION OR “TRUE TALENT” EVALUATION, JUST PART OF THEIR 2008 YEAR IN REVIEW STATS SHEET FOR THE RED SOX.

It’s just like looking at FanGraphs bUZR and seeing that he was -15.7/150 in 2008, and giving that an “F,” but not thinking that’s his “final grade.” It’s their season-in-review for one season. Just like when Justin Inaz had him sucking on defense, or when MGL said that sUZR had Crisp in double digit negatives for 2008…

Bringing you more-or-less replacement level analysis and commentary since sometime in 2008.

by Matt Klaassen on Feb 13, 2009 12:16 PM EST up reply actions

Studes (Dave Studeman, the guy who puts the annual together)

jumped in on the thread at BtBS and confirmed my point here — it’s a single season evaluation and is explained as such in the annual

And UZR did not have Crisp as “excellent” in 2006 they have him at 0.0 — exactly average.

My earlier survey at the time of the trade came up with this (I think) — all the systems have him around average or a bit better in 2006 (except for PMR, which loved him), awesome in 2007, and terrible in 2008 (except for PMR, I think, which had him about average). I didn’t look at pre-2006 in that survey, but the numbers are good for both LF and CF (in smaller sample sizes). The Fans Scouting Report loves him.

Bringing you more-or-less replacement level analysis and commentary since sometime in 2008.

by Matt Klaassen on Feb 13, 2009 12:25 PM EST up reply actions

d_f

I stated in the article that the F- was from single season data a couple of times in the article, but that single season data is not reliable, is that not what you got?

by Jeff Zimmerman on Feb 13, 2009 12:44 PM EST up reply actions

I was more addressing the questions in this thread about the grade

I thought that’s what you were saying in the article as well, then I got confused when the discussions in the comments here started. I think that going from that to UZR/150, and then using several seasons of data to make your point might have been confusing as well.

Bringing you more-or-less replacement level analysis and commentary since sometime in 2008.

by Matt Klaassen on Feb 13, 2009 12:54 PM EST up reply actions

Maybe I misunderstood the article

I think there’s some good stuff in there, but maybe if you’d just made clear whether you were writing about about 1) Crisp’s likely defensive performance going forward/true talent was better than F minus/minus 15 or 2) whether you through there was reason to think that pretty every defensive metric was simply wrong about Crisp’s performance in 2008 thing would have been more clear.

Bringing you more-or-less replacement level analysis and commentary since sometime in 2008.

by Matt Klaassen on Feb 13, 2009 12:57 PM EST up reply actions

Thanks

I need another editor/proof reader, but my wife no longer will read any of work to read it first. I just need to hope my 3 month year old takes up writing and baseball quick.

by Jeff Zimmerman on Feb 13, 2009 1:08 PM EST up reply actions

no problem

don’t get me wrong, I liked it. My own stuff has much worse clarity problems. It’s hard to remember that most people reading blogs probably don’t read as slowly as they would a book or something. I know I forget that when I’m making minute distinctions at a rapid rate.

Bringing you more-or-less replacement level analysis and commentary since sometime in 2008.

by Matt Klaassen on Feb 13, 2009 1:12 PM EST up reply actions

I just wish Studes wasn't so mad

No had ever looked into the season to season data. I thought it would be useful. I am probably going to get in trouble for this, but the THT is starting to turn into BP. They used to give out criticism to everyone else (Bill James, BP, etc.), but now that they are one of the big boys out there they can’t take it like they used to dish out. I am going to give Dave (studes) a pass on this one. I just don’t see the Tango gang freaking out over the differences in Aging curve.

by Jeff Zimmerman on Feb 13, 2009 1:24 PM EST up reply actions