Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Sox Therapy > Discussion
Sox Therapy
— Where Thinking Red Sox Fans Obsess about the Sox

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.

Page 1 of 2 pages  1 2 > 
   1. Darren Posted: January 01, 2006 at 01:11 AM (#1801478)
Note: I suppressed TVErik, Joel, and crispy's comments, because they no longer make sense with the stuff above. If you guys want those back, let me know.
   2. The Anthony Kennedy of BBTF (Scott) Posted: January 01, 2006 at 01:16 AM (#1801481)
i still like manny.

*pout*

y'all suck with this "get rid of manny" stuff.

next thing you know Ortiz will have a down year, and he'll be on your #### list too.
   3. dave h Posted: January 01, 2006 at 01:20 AM (#1801486)
Ok, before I actually bother to do the reading on how UZR works (which I'm about to tackle) - I have a question. Does playing in Fenway Park mess things up? There are many factors (where he's standing at the start of the play, how much ground he must cover, how hard a ball is likely to have been hit that falls in a particular zone) that change for a fielder in Fenway, so is Manny just compared to other players at Fenway? That seems to be the only way for this system to work in this case.

I will now go read about whether I'm retarded.
   4. Ivan Grushenko of Hong Kong Posted: January 01, 2006 at 01:28 AM (#1801489)
Is Ortiz -30 at 1B?
   5. dcsmyth1 Posted: January 01, 2006 at 01:30 AM (#1801490)
----"2005, -50, 1225"

I have always sensed that these UZR numbers have too wide a spread. I understand how they are determined, and I understand that UZR seems to have a .50 year to year correlation, so my intuition may be wrong. Still, for a corner OFer to be rated at -50 runs--well, that is an elephant--so bad that no team would leave him in the OF. I would regress all of these numbers 50%. So, maybe Manny really should be expected to post -15 runs in 2006.
   6. tfbg9 Posted: January 01, 2006 at 01:39 AM (#1801498)
Dumb question, but does -50 runs mean that Manny was 40 runs worse than the average LF'er in MLB? That seems really counter intuitive.
   7. The Anthony Kennedy of BBTF (Scott) Posted: January 01, 2006 at 01:41 AM (#1801500)
so if there was an average left fielder the redsox pitchers would have given up 50 fewer runs?

i really can't buy that. that's a ####### HUGE amount of runs, and i'm not sure a left fielder could really even BE that big a component of team defense given the number of opportunities.
   8. The Anthony Kennedy of BBTF (Scott) Posted: January 01, 2006 at 01:43 AM (#1801501)
no, average corner outfielder of his age, tfbg9
   9. dave h Posted: January 01, 2006 at 01:52 AM (#1801507)
Ok, I read as much as I could handle in one sitting, and here's my conclusion:

I'm confused, but suspicious. There are park factors, and Fenway Park, LF, was about .85 over 10 years. So clearly it's special. My understanding is that Manny was then given credit for the number of outs he recorded/.85 in each zone, to determine his uzr.

1) That .85 is regressed (I'm not sure in what way), but we have good reason to believe Fenway really is that different.

2) There is finite granularity in the park factor, so that .85 might still be missing a lot.

I'd like to see uzr determined for all LF's playing in Fenway (as if baseball were not played anywhere else). Would it come up with the same result?
   10. tfbg9 Posted: January 01, 2006 at 01:53 AM (#1801508)
That seems like too many runs, when the difference in overall runs allowed last year between the first and worst club was 294.
   11. The Anthony Kennedy of BBTF (Scott) Posted: January 01, 2006 at 02:04 AM (#1801517)
yes. and when you're not even considering the difference in pitching.

and i think you'd be hard to find too many sox starters that massively underperformed their component ERA in such a way that you could blame defenses for the problem.

i like UZR to an extent, having fielding metrics is a GOOD thing. i just think that because it's the best we have at the moment it gets treated better than it really is.
   12. Buzzards Bay Posted: January 01, 2006 at 02:27 AM (#1801535)
Manny plays a shallow Left....this distorts UZR....it is strategic and it is effective.....there are no metrics for --would of been a basehit-- which in his home park is relevant...
   13.  Hey Gurl Posted: January 01, 2006 at 02:56 AM (#1801555)
Roughly how many "missed plays" results in a negative run?
   14. notonreplay Posted: January 01, 2006 at 03:17 AM (#1801575)
Are there road splits for UZR?
   15. Darren Posted: January 01, 2006 at 03:36 AM (#1801583)
Still, for a corner OFer to be rated at -50 runs--well, that is an elephant--so bad that no team would leave him in the OF. I would regress all of these numbers 50%.

That's about how much MGL reccomended regressing for a single season worth of data. So, if you just had the -50 number and regressing it by 50 percent toward the baseline of -10, you're going to get about -30. Then you take off about 5 for aging and you're going to get -35. Not at all far from -30.

I would regress all of these numbers 50%. So, maybe Manny really should be expected to post -15 runs in 2006.

But the -30 figure already includes a regression of 20 percent. You're now regressing by some 60 percent to get that -15. And why would you do that? Just so you can get a number that matches with your preconceived notions? What's the point of having unbiased metrics if you're just going to change them if they give you a result you don't expect?

Manny plays a shallow Left....this distorts UZR....it is strategic and it is effective.....there are no metrics for --would of been a basehit-- which in his home park is relevant...

Wouldn't UZR credit him with all those extra putouts, thereby making up for his other misses? But let's assume that you're correct that UZR would not reflect this. Wouldn't I notice it when I was watching Manny (which is how I assume you know about it)? When watching Manny, I never see him catching shallow flies that I expect to fall. Have others noticed this?
   16. Darren Posted: January 01, 2006 at 04:01 AM (#1801602)
I'd like to see uzr determined for all LF's playing in Fenway (as if baseball were not played anywhere else). Would it come up with the same result?

One interesting point is that other players who have manned LF for the Red Sox have put up perfectly normal looking UZRs. Troy O'Leary had +32 /162 games in 99 and then +15 /162 in 00, and +9 /162 in 01. When he moved on to Montreal in 02, he put up +6 /162.

Dante Bichette put up a +4 for Boston in LF in 01, after putting up a -18 for Cinci as a RF the previous year.

I don't see any evidence that these two fellas were aversely affected by Fenway. And if you look at the shape of Manny's career, I don't think he was either. He was quite the butcher in RF for the Indians in 00 before ever reaching Fenway. After a flukey 50 LF games in 01 where he posted a +9, he returned to his butcherly ways.
   17. Darren Posted: January 01, 2006 at 04:08 AM (#1801607)
next thing you know Ortiz will have a down year, and he'll be on your #### list too.

I wouldn't want to trade him for having a down year. I'd want to trade him if we had another guy who was also a DH. Manny wanting out gives the Red Sox a good excuse to trade him. They can plausibly say to teams in both the NL and AL that Manny is a superstar that they'd like to keep, but they want to honor his wishes or are tired of his complaints. That excuse can also be used with the fans. It cannot be used to excuse dealing Ortiz. Yet.
   18. dave h Posted: January 01, 2006 at 06:32 AM (#1801641)
It's possible that Fenway isn't the problem, but I'm still unconvinced on that particular point. I'm not saying it necessarily hurts everyone, maybe it just makes things screwy. At the very least, I believe it increases the error in the measurement, since each real out is worth more runs.

I do admit that I'm trying to find reasons why uzr does not match my preconceived notion. That notion is not that Manny is not a bad fielder, I'm fine with that, but that 50 runs is a ridiculously large total for one fielder. As I understand it, that's how many runs he cost in Fenway, where presumably his impact is limited by the small playing area, and possibly includes the impact of his 17 assists (the post I read indicated that uzr does later account for arm and double plays - or is that another metric that starts with uzr?)
   19. dcsmyth1 Posted: January 01, 2006 at 01:33 PM (#1801691)
-----"But the -30 figure already includes a regression of 20 percent. You're now regressing by some 60 percent to get that -15. And why would you do that? Just so you can get a number that matches with your preconceived notions? What's the point of having unbiased metrics if you're just going to change them if they give you a result you don't expect?"

It's called a "sanity check", and there is a legitimate place for it, especially in fielding analysis.
   20. TVerik, the gum-snappin' hairdresser Posted: January 01, 2006 at 02:59 PM (#1801703)
So Manny is a great hitter (although declining) and a brutal defender?

As other posters have pointed out (indirectly), if we assume that Ortiz is the DH, the Sox indeed should move Manny.

But why not put Ortiz at 1B three times a week, inserting Manny as the DH then? It's a sub-optimal platoon, but at this point I can't imagine Manny's trade value meaning that the trade returns would outweigh the value of anyone in LF for only half the games.
   21. IronChef Chris Wok Posted: January 01, 2006 at 03:28 PM (#1801710)
But why not put Ortiz at 1B three times a week

I keep saying this (I said once or twice a week), but I don't think Ortiz's knees are good enough
   22. Matt Clement of Alexandria Posted: January 01, 2006 at 03:43 PM (#1801714)
It's called a "sanity check", and there is a legitimate place for it, especially in fielding analysis.
I'm actually, in some ways, sympathetic to this argument. -30 and -35 are numbers that appear at most once a year in the UZR spreadsheets. Manny could be 10-15 runs better than that and still reside at an extreme point.

I am not going to project Manny anywhere better than -20, but there's a pretty big difference between -20 and -30.

What I do object to in the above argument is the certainty. It's not a "sanity check" becuase, well, you don't know what's sane, either, Duffy. I tend to be worried about extreme projections from a system that is notably more variable than most offensive systems, and I like to cut them off a little.

But, I think something very weird is happening with the -47 UZR. I think Dick Allen once remarked that after a season where he hit 60-some homers in the minors, that it would have been better if he'd hit 40 instead. 40 they could understand. 60 was so high that everyone looked for a way to discount it. The point is that 60 is better than 40, just as -45 is worse than -25. Yes, we should be skeptical of every system, but that doesn't mean we get to ignore the numbers, and in particular you can't use the -47 as evidence to improve Manny's projection.
   23. RB in NYC (Now Semi-Retired from BBTF) Posted: January 01, 2006 at 04:24 PM (#1801726)
Has UZR changed in some way as a system or is my memory playing tricks on me? I seem to recall that for some year--maybe 2004, although it could've been '05, I suppose--Bernie Williams was somewhere around -40 per 150 games. Now this leaves me with three possibilites, (1) UZR has changed (reasonable) (2) My memory is faulty (also reasonable, even more so for being New Year's Day) (3) Manny in LF in Fenway is only ten runs better than Bernie in CF.

I don't want to ignore the moral of Darren's story, but at the same time, I'm having a really hard time wrapping my head around (3)
   24. Rafael Bellylard: Built like a Panda. Posted: January 01, 2006 at 04:29 PM (#1801728)
Scary thought....Manny at 1B
   25. Darren Posted: January 01, 2006 at 04:33 PM (#1801730)
I'm actually, in some ways, sympathetic to this argument. -30 and -35 are numbers that appear at most once a year in the UZR spreadsheets. Manny could be 10-15 runs better than that and still reside at an extreme point.

Manny's actual number in 05 wasn't -35 or -30, though. It was -50. When MGL projects Manny to be at around -30 next year, there's already a 'sanity check' built in--the regression. That's why he's not projected to be at -35 or -40 or something.

It's funny, I started this thread with hopes of putting in a sort of sanity check as well. Although I think mine was slightly different, the fact is that both sides of this argument (pro and anti-Manny) want to somehow tweak the numbers a bit. I think that shows just how shocking that -50 figure is to us.

I think it's important to build a large, wide-ranging consensus that UZR is indeed, the ULTIMATE way to measure defense. That way, we can all agree that Jeter sucks!!!!1111oneoneone
   26. RB in NYC (Now Semi-Retired from BBTF) Posted: January 01, 2006 at 04:34 PM (#1801731)
Apparently, Primer's New Year's Resolution was to continue to deny me the chance to see posts made after mine unless I posted again myself
   27. TVerik, the gum-snappin' hairdresser Posted: January 01, 2006 at 04:38 PM (#1801734)
How much damage could Manny do if he played C when Wakefield pitched?
   28. Darren Posted: January 01, 2006 at 04:40 PM (#1801737)
RB,

The spreadsheets I have have Bernie at -38 /150 in 02 and -51 /150 in 03. I don't know the 04 or 05 numbers, but they're probably not much better. After regressing the numbers, though, I think Manny v. LF and Bernie v. CF would be pretty close.
   29. Darren Posted: January 01, 2006 at 04:42 PM (#1801739)
How much damage could Manny do if he played C when Wakefield pitched?

To who? To me, personally, having to watch that? I'd say a lot. A LOT!
   30. karlmagnus Posted: January 01, 2006 at 04:58 PM (#1801748)
If the main problem is Manny's fileding (which I don't believe -- I think sabermetric fielding numbers are cuckoo) then the solution is to trade Fatso Ortiz and allow Manny to DH. For Ortiz and Arroyo/Clement we could undoubtedly get two top quality OF and markedly improve the team, as well as missing Ortiz' decline/I am underpaid/I want to be traded phase, which is clearly imminent. His value will NEVER be greater than it is right now. Manny on the other hand is a good reliable soldier who's heading for the HOF, which Ortiz isn't (started too late, to begin with.) He's MUCH more likely to be a high quality player in 2010 than is Ortiz, but we're only committed through '08.
   31. IronChef Chris Wok Posted: January 01, 2006 at 05:13 PM (#1801752)
Scary thought....Manny at 1B


I think I"ve read thsi around here... but Manny vs. Ichiro! in a rundown situation.

Oh the pure hilarity

For Ortiz and Arroyo/Clement we could undoubtedly get two top quality OF and markedly improve the team, as well as missing Ortiz' decline/I am underpaid/I want to be traded phase, which is clearly imminent. His value will NEVER be greater than it is right now. Manny on the other hand is a good reliable soldier who's heading for the HOF, which Ortiz isn't (started too late, to begin with.)

You think Manny is pissed NOW?
   32. dave h Posted: January 01, 2006 at 05:43 PM (#1801766)
There's just such a huge uncertainty in these defensive numbers. Yes, uzr is useful, but we know how much Manny provides with the bat and we have a hazy idea of how much he takes away with his glove. You're also neglecting assists with uzr, and baserunning which of course hurts Manny. It's reasonable to make an estimate of what he's worth, but to trade Manny based solely on these defensive numbers seems unwise. No, he's not worth $20 mil per, but you need to get significant value back to be worth moving him.
   33. dcsmyth1 Posted: January 01, 2006 at 06:22 PM (#1801792)
Sometimes it can be helpful (yes, as a sanity check) to reframe something into a batting equivalent, since we are so familiar with it. So, I worked out some numbers. Manny's -50 runs are the equivalent of a batter (in a .260 BAvg lg in which the highest BAvg is only around .310) hitting .420 in 350 ABs. At least, that's what I got.
   34. OlePerfesser Posted: January 01, 2006 at 06:48 PM (#1801822)
This is a neat thread; thanks to Darren (and MGL and MCA) for putting it up.

1) Given the imprecision of fielding metrics, I always argue it's better to view many rather than place exclusive reliance on one. For the record, then, here are Manny's FRAA (per BProsp) from '01-'05:

2, -6, -7, -11, -13

And here are his Zone Ratings as a LF for the same period:

.877, .779, .789, .750, .729

2) Both metrics support the view that Manny's defense is below average and it's in steady decline. That would, in turn, support the idea that it's time to move him. And since Ortiz occupies DH, the move might logically be out of town (sorry, karl).

3) Neither of these other metrics, however, is consistent with the implication of the UZR numbers that Manny's D is historically, catastrophically costly. I share the view of others that the UZR method, whatever its other virtues, might be throwing a rod in this case. My reason:

By my calculations (from Zone Rating data), the average LF sees 367 flyballs and line drives hit into his zone over a season, and converts .867 of them, or 318, into outs. Manny's execrable .729 ZR in '05 means that he converted 268 of these opportunities into outs, or 50 fewer than the average LF. UZR is telling us that each one of these 50 became a run. That's possible, but it seems terribly unlikely--or is a by-product of terrible luck, whereby each time Manny butchered a flyball there were runners in scoring position, or such. My math might be off, but I think this is what other UZR doubters were getting at: a LF just doesn't get that many opportunities to be 50 runs below average.

4) If Manny can't be moved, then the FO and Tito would need to be raked over the coals for not having a Gene Stephens-esque caddy around and using him to minimize damage. I.e., there should be days (flyball pitchers in spacious parks, at the least) when Ortiz sits or tests his knees at 1B, and a rangy guy plays LF and Manny DHs. I suspect he's likely to cost the Sox 1.5 to 2.0 wins per year in LF in full-time duty--but that can be mitigated somewhat if he's used more wisely.
   35. ellsbury my heart at wounded knee Posted: January 01, 2006 at 09:01 PM (#1802003)
I don't really understand fielding metrics, but if those stats somehow correct for the left field wall, which decreases the amount of ground Manny has to cover in left, than in a larger LF Manny would be absolutely abyssmal. I just can't wrap my head around how bad that would be.
   36. RB in NYC (Now Semi-Retired from BBTF) Posted: January 01, 2006 at 09:03 PM (#1802008)
Darren, does UZR correct for park effects? Left in Fenway has always chipped away at a leftfielders opportunities because so many of the doubles hit there are uncatchable.


Post 16 addresses this point, to some extent
   37. Sam M. Posted: January 01, 2006 at 09:15 PM (#1802022)
Manny on the other hand is a good reliable soldier

That has to be a strong early contender for the 2006 Did He Really Write That? Primey.
   38. karlmagnus Posted: January 01, 2006 at 09:59 PM (#1802072)
Manny's performance variance is very low indeed; you know what you're going to get. He also doesn't cause in-season hiccups by prima donnaing. The only hiccups come when these halfwits attempt to trade him every 6 months. I repeat: TRADE FATSO, KEEP MANNY. THAT is what a truly superior trader would do, but then Henry is clearly NOT a truly superior trader. I think a really top-notch GM would keep both, but then he'd have kept Pedro and Damon, too.
   39. sublime Posted: January 01, 2006 at 10:42 PM (#1802121)
Manny's performance variance is very low indeed; you know what you're going to get. He also doesn't cause in-season hiccups by prima donnaing. The only hiccups come when these halfwits attempt to trade him every 6 months. I repeat: TRADE FATSO, KEEP MANNY. THAT is what a truly superior trader would do, but then Henry is clearly NOT a truly superior trader. I think a really top-notch GM would keep both, but then he'd have kept Pedro and Damon, too.


why do you ignore when people pick apart your attacks on the FO, or ideas you present?

again, from a pure HITTING standpoint ortiz has been the better player the last two seasons, is younger (maybe offset by having a smaller SS of success than manny) and costs 1/3 as much. in other words, looking to move him to have a spot for manny is asinine unless you get back a adam dunn/jason bay type.
   40. tfbg9 Posted: January 01, 2006 at 10:57 PM (#1802136)
My eyes tell me that Manny gets to nothing anymore. That said, there is no way I can picture anybody being -50 runs per 150 as a corner outfielder, outside of Chief Ironside in Coors.

ESPN yesterday had the Sox giving up Manny, Clement AND MARTE for Tejada and Gaithwright. This would suck.
   41. Darren Posted: January 01, 2006 at 10:59 PM (#1802137)
OleP--

I'm all for using a variety of stats, but I would not want one of them to be FRAA. There's been no explanation of the methodology (AFAIK) and no testing of its accuracy.

ZR seems to make a fair amount of sense, but remember that UZR is based on ZR, but adjusts it for park/opportunities. That's why, even though Manny only appears to be 50 plays below average, he accumulated -50 runs. In the real world, some combination of park and luck meant that Manny actually missed more than 50 plays that an average LF would have made, and that number of plays (80? 100?) translated into 50 runs. Was that high number a fluke? Probably, and that's why you regress.

Darren, does UZR correct for park effects? Left in Fenway has always chipped away at a leftfielders opportunities because so many of the doubles hit there are uncatchable.

It's park adjusted.
   42. The Keith Law Blog Blah Blah (battlekow) Posted: January 01, 2006 at 11:22 PM (#1802163)
Then you take off about 5 for aging and you're going to get -35. Not at all far from -30.

READ: -30 and -35 are very similar

When MGL projects Manny to be at around -30 next year, there's already a 'sanity check' built in--the regression. That's why he's not projected to be at -35 or -40 or something.

READ: -30 and -35 are not similar.

Darren said both. I don't get it, though the latter sentiment is more agreeable. A 17% difference is not "not at all far."
   43. caprules Posted: January 01, 2006 at 11:25 PM (#1802165)
Darren, can you point me to a study where UZR has shown the full methodology including all the adjustments that are made and has been tested for accuracy?
   44. Darren Posted: January 01, 2006 at 11:49 PM (#1802196)
You're right battlekow, that sounds pretty inconsistent. I'll try to clarify. 5 runs over the course of a season is not likely to have a big impact on the value of a player. At the same time, regressing Manny's stats did in fact move them from somewhere between -35 and -40 to about -30. I don't think my 2nd quote in any way comments on whether -35 to -40 is 'all that far' from -30. It only says that the -30 figure already includes the 'sanity check' that Duffy was looking for.

Darren, can you point me to a study where UZR has shown the full methodology including all the adjustments that are made and has been tested for accuracy?

I thought that MGL's original articles did a good job explaining the methodology and the testing of it. IIRC, it was revised after peer review from some of the defensive masterminds (Emiegh+Co.). I don't have a link to them right now. You might ask him, though.
   45. caprules Posted: January 02, 2006 at 12:06 AM (#1802215)
mgl has made adjustments since his original articles, without any explanation of what changes were made or why. When he posted yearly stats for some players last year, they were different than the stats mgl had originally posted for those players in the given years. Also, the original decision to exclude line drives from UZR was not done based on any study, it was just a choice that mgl made. I have read the UZR articles a couple of times, and I don't remember any test for accuracy in them. Really, it seems the critiques you have for FRAA apply to UZR as well.
   46. OlePerfesser Posted: January 02, 2006 at 12:55 AM (#1802271)
Really, it seems the critiques you have for FRAA apply to UZR as well.

I think MGL keeps less of his methodology proprietary than the BProsp lads, but the point is well-taken that some of his method is "cloaked." I don't necessarily have a problem with that--just that it's one reason for viewing multiple data sources. Also, I've had many fruitful e-mail chats with BProsp folks, and have always found them extremely interested in getting things right, so I tend not to share the generally dim view some BTF posters have of them.

in other words, looking to move [Ortiz] to have a spot for manny is asinine unless you get back a adam dunn/jason bay type

We might have to give karl a little credit for his idea of moving the curent DH rather than the current LF. Once you've signed a guy, then any difference between his contract cost and his perceived value is a sunk cost. Ortiz's lower salary therefore makes him more attractive in trade; the possibility of getting a stllar player or package in return for him might be greater than for Manny, as karl says. But karl: between Manny and Papi, one wants to be in Boston and DH, and the other wants neither. That, more than anything, is driving this situation; no point in ignoring that fact.
   47. The Anthony Kennedy of BBTF (Scott) Posted: January 02, 2006 at 01:00 AM (#1802277)
wait, so your 50% regression is being done just as a WAG?
   48. RobertMachemer Posted: January 02, 2006 at 01:06 AM (#1802286)
A 17% difference is not "not at all far."

I disagree -- a 17% difference can be "not at all far" -- it just depends on the situation.

For instance, a player who doubles his home run output in successive seasons might have done something that feels pretty significant (going from 28 home runs to 56 home runs) or something less significant (going from 1 home runs to 2 home runs). To say that a player has improved by 100% does not necessarily mean that his improvement has been all that significant.

I don't know whether this 17% is "far" or "not at all far," but I do know that the simple percentage is not enough to make that determination.
   49. Joel W Posted: January 02, 2006 at 01:55 AM (#1802345)
Darren,

My complaints are no longer valid. A few points though:

1) I think UZR is probably right on in telling us how bad Manny is in the outfield. I don't think MGL's park factors are off, or his evaluation of Manny's true defensive talent in LF is wrong. I think that he still may have more value to the Red Sox in the same way that Derek Lowe has more value to the Dodger's than to other teams: the park takes away their worst assetts. So while Manny may only be 18 runs above average, he may be significantly mroe runs above average for the Red Sox.

2) I still think, as I claimed in another thread, that in evaluation Manny's true talent level, we should dock him for defense only as much as we dock the average DH for playing DH. It isn't Manny's fault that in his Red Sox context that he cannot play DH. Since that appears to be on the order of 2 wins (or so Tango said I think in the Ortiz MVP discussions) that would put Manny right about at his WARP from last year. I think the Red Sox should expect to get that back from a team who can DH him.
   50. Kyle S Posted: January 02, 2006 at 02:23 AM (#1802375)
These threads seem to pop up every few weeks and are always THE EXACT SAME! A group of people here always says, "I have no idea how UZR works, but there is no way that Manny was -50 [before arm runs, btw - he wasn't -50 all told], therefore UZR is bunk/I will add in a 30 run fudge factor/MGL is an idiot/etc. By the way, UZR says Jeter was -30, so let's giggle about how bad he is at defense! Big Papi for MVP!"

There are problems with the -50 number. UZR relies on human judgment calls about which zone a ball lands in, etc, and hence comes with its own error bars. For Manny, I would bet these are from -42 to -58 or so, and it's probably more likely that he was around -42 just because of the outlier nature of the data point.

Additionally, Manny probably was unlucky considering the balls hit to him this year. Just as Kelly Johnson (Braves LF) got a great UZR despite being a mediocre fielder because he had a lot of balls hit right at him, I would bet that Manny had a higher than average number of "difficult" plays, such that a +0 true talent LF would have been perhaps -5 or so given the set of balls hit at Manny this season.

However - those are just guesses. I think it's possible (while not likely) that he was even worse than -50. Also, ALL of the data (Dial ZR, MGL UZR, "FRAA", empirical observation) point to Manny as being a bad defender and getting worse. Finally, as others have stated, no one thinks Manny's true talent was -50. He's probably about -30 at this point.

Don't let sticker shock of the -50 number impede your ability to think analytically. Just because Jose Lima had a 6.99 ERA last season does not mean his true talent is that bad. So too with Manny.
   51. Darren Posted: January 02, 2006 at 02:35 AM (#1802388)
wait, so your 50% regression is being done just as a WAG?

No, it comes from MGL.

On the subject of testing/methodology of UZR, perhaps my memory has failed me. I remember it being pretty well explained and that it had been tested (in its early iterations, anyway), and that it was shown to correlate reasonably well year to year. Sorry if I've overstated that (which I probably have).

I don't mean to in anyway suggest that UZR is infallable, only that I think it's a good way to measure defense. I also didn't mean to cast any aspersions on the BPro guys (in this case), I only meant to say that I didn't trust their fielding metrics because I have no information about how they are created or how well they've worked in the past.

Joel, thanks for your comments on the previous post, which was a disaster from the beginning. In answer to what you're saying here:

1) Do you mean that Manny's defense could be expected to be worse than -30 in a bigger OF, therefore he is more valuable to the Red Sox? It's possible, but it's not an assumption that I'd make without investigating his home/road splits (if available). Alternatively, I might guess that Fenway entices more hitters to hit the ball to LF.

2) I agree with you to an extent that Manny's not fully to blame for not being allowed to DH. But I'm not too interested in his overall value as a player, or as it concerns the Red Sox in 2006-08. One thing that you might also consider, though, is that Ortiz has demonstrated an ability to hit well as an everyday DH. We don't know how Manny would react to that role.
   52. Darren Posted: January 02, 2006 at 02:41 AM (#1802394)
A group of people here always says, "I have no idea how UZR works, but there is no way that Manny was -50 [before arm runs, btw - he wasn't -50 all told],

Really? Did I miss something? What is he all told? Is that the -47 figure that I've seen floating around?
   53. Kyle S Posted: January 02, 2006 at 04:50 AM (#1802537)
I think he was +6 arm runs, so about -44 on defense altogether.
   54. Psychedelic Red Pants Posted: January 02, 2006 at 06:12 AM (#1802648)
By the way, UZR says Jeter was -30, so let's giggle about how bad he is at defense! Big Papi for MVP!

SS's have about twice as many chances as LF's. Manny "as bad as Jeter" would be around -15. Jeter "as bad as Manny" would be around -100. That's why I think something is fishy. Manny's number is huge relative to the number of chances he should get. Jeter's numbers were bad, but not nearly that bad.

It's also possible than speedy no-arm CF's playing LF are moving the average while Manny is turning chances into outs at close to his normal rate. Rather, Chicago playing Podsednik in LF, Cleveland playing Crisp in LF, and Tampa Bay playing Crawford in LF shouldn't effect Manny's value to the Red Sox while it would effect the baseline for UZR. (I have no idea if any of these are actually good by UZR, but they're speedy). In that case, Manny's worsening UZR number wouldn't really be a concern because he's not actually getting worse -- other teams are just fielding true CF's as LF's. If that were the case, though, then Manny's hitting relative to average should have improved.
   55. IronChef Chris Wok Posted: January 02, 2006 at 11:49 AM (#1802766)
Additionally, Manny probably was unlucky considering the balls hit to him this year. Just as Kelly Johnson (Braves LF) got a great UZR despite being a mediocre fielder because he had a lot of balls hit right at him

1) Just because Balls are hit at Manny doesn't mean he's catching them

2) Balls being hit at people mean people were standing at the right place. You can't penalize people for standing at the right place.
   56. richie allen Posted: January 02, 2006 at 12:51 PM (#1802772)
Darren, do you know if UZR incorporates a win-shares style 'top-down' factor, i.e. that individual ratings are constrained by the overall team defense?

Apologies if this is clearly answered somewhere, but in years of following I don't think I've seen this explained.

Rich
   57. dcsmyth1 Posted: January 02, 2006 at 02:38 PM (#1802784)
No, there is no top-down approach. MGL usually does not convert the runs into wins. So, UZR probably "adds-up" on the lg level, but not on the team level.
   58. cercopithecus aethiops Posted: January 02, 2006 at 03:41 PM (#1802817)
Jeter "as bad as Manny" would be around -100. That's why I think something is fishy.

I don't find that fishy at all. I'm absolutely confident that Manny would be at least -100 at SS.

Seriously, here's something that confuses me: every time this subject comes up, some people argue that the numbers must be screwed up by Fenway; then someone replies that UZR is park-adjusted; then the argument turns to what his "real" UZR would be in a neutral park. Well, isn't the point of the park adjustment to give us a neutral context number in the first place? SO isn't it a little bit bogus to adjust a number for context when it's already been context-adjusted?
   59. Chris Dial Posted: January 02, 2006 at 03:44 PM (#1802820)
By my calculations (from Zone Rating data), the average LF sees 367 flyballs and line drives hit into his zone over a season, and converts .867 of them, or 318, into outs. Manny's execrable .729 ZR in '05 means that he converted 268 of these opportunities into outs, or 50 fewer than the average LF.

Nice work, Ole Perf. Always good to see people taking on the lifting for themselves.

See this:
Easy for everyone here
   60. Chris Dial Posted: January 02, 2006 at 03:45 PM (#1802823)
Fenway has some significant park effects.

Anyone want to watch all teh Red Sox home games and see how many FBs hit off the scoreboard?

That'll help see the effect on Manny's UZR. 0.85 seems like a pretty good guess at a PF though.
   61. Mister High Standards Posted: January 02, 2006 at 03:54 PM (#1802838)


Manny's actual number in 05 wasn't -35 or -30, though. It was -50


I've posted this before, specifically in one these threads. It is not remotely possible for a leftfielder with 17 outfield assists to be -50 defensively. It is similar to be player having 200 hits in 198 plate appearances.

Allowing a minus 50 means he has to turn 63 outs into a singles.
or
Allow 200 extra bases
or play 45 outs into doubles.
or some proportional combination.

Now... maybe, just maybe but its extreme that is possible for a leftfielder... but not when you factor in 17 outfield assets. I figure that has to mean manny was really -60 or something in UZR terms before his assets are accounted for... which just can't possibly the case.

Someone counted something wrong, or the monster isn't accounted for correctly. UZR seems like a reasonably strong model, in most cases. In this case it just isn't practical, most likely because IMHO the data that MGL feed into the model was wrong.
   62. Chris Dial Posted: January 02, 2006 at 04:01 PM (#1802842)
MHS,
17 OFA is only about 10 runs.

Allowing a minus 50 means he has to turn 63 outs into a singles.

This is almost *exactly* what Manny does. Nice job.

I have Manny as about -32 for 2005, though.
   63. Mister High Standards Posted: January 02, 2006 at 04:15 PM (#1802854)
Dial - think about what your saying.

A leftfielder on average gets something like 2 balls hit to him per game which he has an opportunity to make an out. Thats roughly 300 plays per year. Your telling me manny flubs 1 in 5?

Not remotly possiable.

-32 is the high end of what I think is the max worst case for an left fielder with 17 outfield assists.


17 OFA is only about 10 runs.


Which is exactly what I said:

I figure that has to mean manny was really -60 or something in UZR terms before his assets are accounted for


Maybe if I said it on usenet 10 years ago...
   64. Chris Dial Posted: January 02, 2006 at 04:35 PM (#1802863)
A leftfielder on average gets something like 2 balls hit to him per game which he has an opportunity to make an out. Thats roughly 300 plays per year. Your telling me manny flubs 1 in 5?

Not remotly possiable.


You are simply wrong.

There are 360 FBs hit to LF. Manny catches about 260, while most LFs catch about 320.

He doesn't "flub" them. He has less range, and doesn't get to them (or he jogs after them and watches them land in front of him).

I missed your -60 remark, sorry.
   65. chris p Posted: January 02, 2006 at 04:57 PM (#1802882)
it's just not possible for chris hammond to go through a whole season and have a 0.95 era. it would mean he would have to allow only 8 runs in 76 innings. Not remotly possiable.
   66. Mister High Standards Posted: January 02, 2006 at 05:30 PM (#1802904)

There are 360 FBs hit to LF. Manny catches about 260, while most LFs catch about 320.


Not in Fenway their aren't. The average stadium yes. Fenway no.

Couple that with the extremely unlikely scenario where Manny gets to 20% less balls than average and you have a number that just doesn't reconcile.

Do I think Manny is a good fielder? No. It is mathematically possible for Manny to be as bad as UZR indicates? Yes it is possiable... but not remotely probable. The arguements that have been made just aren't remotly convincing.

Chris - if you have Manny at -35 and MGL has Manny at -50 which number is correct? Of course the answer is neither.
   67. cercopithecus aethiops Posted: January 02, 2006 at 06:18 PM (#1802938)
Not in Fenway their aren't. The average stadium yes. Fenway no.

Again, the -47 UZR is already park-adjusted, right? That means it is not an accounting of runs he actually cost in the parks he actually played in, but an estimate of what he'd cost in a neutral park, no?

So, unless I'm totally misunderstanding the purpose and nature of the park adjustments, isn't it possible to believe that both a) Manny really did have a -47 season and b) it didn't really cost Boston 47 runs?
   68. Chris Dial Posted: January 02, 2006 at 06:34 PM (#1802954)
Chris - if you have Manny at -35 and MGL has Manny at -50 which number is correct? Of course the answer is neither.

This is of course a silly argument, and one that should apply to your desperate attempts to have Papi named as MVP based on clutch hitting.

How many balls were hit to LF in Fenway? Was it less than I stated or more?

Do this - look at Manny's Put outs. Then his ZR. ZR represents (in a nutshell) the percentage of balls Manny caught that were hit to LF (about half in Fenway).

So 0.729 (ZR) times 243 = 333 FBs. That's 90 FBs hit to the zones defined for Manny to catch (you can read about those zones in the Dialed In section). Not just restricted to Fenway either.

In 333 chances, MOST LFs catch 290 of them. That's 50 FBs that most LFs catch *even in Fenway* that Manny doesn't BASED ON MANNY's ACTUAL NUMBER OF FBs.

Now there *is* some PF that I am not including yet, but I want to make sure there is one that is very detrimental first.

Oh, my -32 includes arm, MGL's -50 doesn't.
   69. Mister High Standards Posted: January 02, 2006 at 07:00 PM (#1802971)


Oh, my -32 includes arm, MGL's -50 doesn't.



If that is true, and I have no reason to believe it isn't... then the -50 has no value in retrospect. Its like trying to measure offense without including doubles.


In 333 chances, MOST LFs catch 290 of them. That's 50 FBs that most LFs catch *even in Fenway* that Manny doesn't BASED ON MANNY's ACTUAL NUMBER OF FBs.


We don't know Manny's actual number of flyballs. We know the number of fly balls one person interpreted as catchable in a very tough to determine zone. You know as well as I do, two people watching the same thing see things very differently. I've also asked you specifically a number of times in other threads how a scorer adjusts for the unique dimensions of left field in fenway. Is a ball 1/4 up the wall that would be a routine fly in every other park but fenway scored as a catchable ball? Because on a standard grid it would be... but practically speaking it isn't. Every time I've asked you claim you don't know. Yet you still blindly put faith in these metrics despite the extreme unlikeness that THIS SPECIFIC outfielder would make 20% less plays than an average one. PFIU!

So when you account for ARM MGL has Manny at about -40, and you have him at about -30... which number is correct? You claim its a silly argument, yet I claim it is vital. If you don't know which is right and which is wrong, then your arguing over unknown vagaries that can't be remotely precisely quantified.

btw: I agree Manny is a poor left fielder. Among the worst in the game. I strongly disagree with the run magnitude being assigned as it is not a remotely possible number.

Also I wasn't using clutch hitting as the basis for my correct claim that Ortiz was AS valuable as A-rod. I used context specific statistics which are the CORRECT statistics to use when determining the value one had in an actual setting not a hypothetical average one. Your insistance on using hypothetical run values rather than actual win values is to quote you: “You are simply wrong”
   70. Sam M. Posted: January 02, 2006 at 07:11 PM (#1802982)
So when you account for ARM MGL has Manny at about -40, and you have him at about -30... which number is correct? You claim its a silly argument, yet I claim it is vital. If you don't know which is right and which is wrong, then your arguing over unknown vagaries that can't be remotely precisely quantified.

Tell me which question you're trying to answer, then I'll tell you whether the difference between -40 and -30 matters. If the question is:

What is the accuracy of this metric?

Then it matters. In fact, it matters to all questions related to the metric itself. Can it be improved? If so, how?

But the difference doesn't matter to any important degree if the question is:

Does Manny Ramirez suck as a left fielder?

If we think of these metrics as estimates, based as they are in part on judgment calls of the observers, then they have value even if they differ to some extent with each other. And on the question of Manny Ramirez as a left fielder, their value in telling us that he is horrendous is much greater than their limits in telling us precisely how horrendous he is. He is horrendous enough that he surrenders a ton of his value as a hitter, and he is horrendous enough that he is massively overpaid.

So if you want to talk about fielding metrics, then the issue you take with the results on Manny is important. But if you want to take about Manny's fielding, then your argument really isn't important at all.
   71. cercopithecus aethiops Posted: January 02, 2006 at 08:19 PM (#1803037)
We know the number of fly balls one person interpreted as catchable in a very tough to determine zone. You know as well as I do, two people watching the same thing see things very differently.

Actually, we know the number of fly balls that three different people watching from different parts of the park agreed were in his zone (without comparing notes until after the fact).

I've also asked you specifically a number of times in other threads how a scorer adjusts for the unique dimensions of left field in fenway. Is a ball 1/4 up the wall that would be a routine fly in every other park but fenway scored as a catchable ball?

This, I think, is just silly. I can not claim to know all of the details of the methodology, but I absloutely cannot believe that a ball 1/4 of the way up the wall in Fenway would be judged as catchable. If that were the case, then a ball to the triangle would have to be judged as uncatchable, since it's out of any other park in MLB. Similarly, home runs down the lines in Yankee Stadium would be "catchable" and warning track fly balls to the power alleys in RFK "uncatchable".
   72. Joel W Posted: January 02, 2006 at 08:29 PM (#1803049)
1) Do you mean that Manny's defense could be expected to be worse than -30 in a bigger OF, therefore he is more valuable to the Red Sox? It's possible, but it's not an assumption that I'd make without investigating his home/road splits (if available). Alternatively, I might guess that Fenway entices more hitters to hit the ball to LF.


I think there just aren't enough balls for Manny to get to. His park adjusted UZR prediction is probably -35. Unadjusted I think he can't actually cost them 35 runs, but more like 30 or 25. It's not a ton, but it matters.

We don't know how Manny would respond to DHing, but we do know that a team that could DH him would probably think that he's worth the same with the bat as a DH. So if the Sox were to trade him they should expect back whatever Manny's value as a DH and backup outfielder would be. A team like Oakland, since they could DH him, would have to offer whatever his expected WARP would be as a DH. That's I guess why I care about his true value.
   73. Joel W Posted: January 02, 2006 at 08:56 PM (#1803086)
Oh and I should add, that yes, in a really big outfield I think Manny would be worth even less to that team, since he may have more opportunities to mess up, even if his park adjusted talent level was still that same -35.

The more we discuss Manny's defense, it really sinks in how much -50 in one season is worth. Especially because Damon was less above average last year, Trot was at best average, Mueller at best average, Renteria at best average, second base at best average, first base below average (well, if we take the right side of the infield it was below average, Millar's crappy decisions are factored in that). Eyeballing PADE I am guessing the Red Sox were -40 or so runs, but I'm not very good at eyeballing that. Could somebody help out here? Regardless, it doesn't make me more optimistic about next year, except that I think we'll have better defense around the infield and maybe in left?
   74. chris p Posted: January 02, 2006 at 09:07 PM (#1803099)
Oh and I should add, that yes, in a really big outfield I think Manny would be worth even less to that team, since he may have more opportunities to mess up, even if his park adjusted talent level was still that same -35.

the way you put it i think is a little misleading. i believe the end result is not "park adjusted" but rather the average outfielder that we are comparing manny to is adjusted for the parks manny plays in.
   75. OlePerfesser Posted: January 02, 2006 at 09:13 PM (#1803109)
Chris D.: Thanks for noticing my calcs in #34 and for your link in #62 to your handy-dandy Zone-Rating-based defensive metric. It's interesting that we both, independently, judge Manny to have converted about 50 fewer opportunities into outs than an average LF last year (before considering his arm).

The question is how that number should be translated into a run value. Where did your runs/play (0.831 for LF) for each position come from? Do you use the same ones for each park/player? (Apologies if you explained all that in the thread--I didn't wade through all the posts/replies.) But it seems to me that this is where Fenway park effects might be unusually strong and hard to measure--and the effects may even differ from player to player. A risk-averse LF who plays with his heels on the warning track may have a low ZR, but give up more low-cost singles in front of him; another guy might play shallow 'cause he thinks it's easy to get back to the all, and give up fewer but more costly doubles.

In any case, this has been a thought-provoking thread. Thanks to all who have chimed in.
   76. Chris Dial Posted: January 02, 2006 at 09:51 PM (#1803161)
We don't know Manny's actual number of flyballs. We know the number of fly balls one person interpreted as catchable in a very tough to determine zone.

That's so ignorant, you should just stop posting on the matter.

You know as well as I do, two people watching the same thing see things very differently. I've also asked you specifically a number of times in other threads how a scorer adjusts for the unique dimensions of left field in fenway. Is a ball 1/4 up the wall that would be a routine fly in every other park but fenway scored as a catchable ball? Because on a standard grid it would be... but practically speaking it isn't. Every time I've asked you claim you don't know.

Well, I don't "claim", if I said I don't know I don't know.

However, since this Manny thing came about, I have found out - and the answer is yes that Manny gets hosed on *some* balls hit off the wall.

I have found that out this offseason. But how many of his 150 FBs to LF is that? Is it significant? What percentage?

Yet you still blindly put faith in these metrics despite the extreme unlikeness that THIS SPECIFIC outfielder would make 20% less plays than an average one. PFIU!

I don't "blindly put faith" in anything. I put faith in things I have studied and worked myself. That you personally don't understand them is fine, but don't project your ignorance to me.

You want to prove the FB theory wrong, drag your arse over to MLB and count the FBs to LF that hit off the wall.

I'll do it sooner or later and see how much it affected Manny.

If you have bothered to research my position, you'ld learn that I generally do not include the Fenway LF because it's clear the wall is troublesome. However, in Manny's case, his marks get worse and worse.

He's atrocious, and I know that other LFs or CFs without the Wall can post numbers this poor (see Bernie Williams).

Do you think Williams is that bad in CF? Do you think he could possibly be that bad? What is the excuse for Bernie's suckitude? He's not that bad and the metrics must be wrong because no CF could be 20% worse?
   77. Robert S. Posted: January 02, 2006 at 09:52 PM (#1803163)
Was anyone saying that Manny seemed exceptionally awful during the course of the season?
   78. Mister High Standards Posted: January 02, 2006 at 09:53 PM (#1803166)

So if you want to talk about fielding metrics, then the issue you take with the results on Manny is important. But if you want to take about Manny's fielding, then your argument really isn't important at all.


Sam your wrong. If Manny's overpaid because we are using the assumption that horrendous defense is worth -50 thats one thing. If we are assuming its worth -30 its another thing, as is -20 or what have you.

His horrendous defense does restrict his value but the question is magnitude. My position from the very beginning is that the numbers being thrown around are far too large for leftfielder, who has 17 assists, and is playing in a park that will "hide" his defensive short comings.

For Manny to be a minus 50 or even minus 40 or 35 as a leftfielder he would have to all time bad. Biblically bad. Unbelievably bad. The worst aspects of Lonnie Smith and Kong combined. He isn't that bad... he is jut run of the mill bad. Which in my estimation puts him somewhere between -20 to -30. Which hurts his value and likely makes him overpaid, but not remarkably overpaid when his offensive constancy is included in that valuation.
   79. Chris Dial Posted: January 02, 2006 at 09:53 PM (#1803167)
Olperfesser,
read the other D-## articles just before the one I linked.

Basically that is the average run value of a ball hit to the position.

There's a ton of research behind it.
   80. Psychedelic Red Pants Posted: January 02, 2006 at 09:56 PM (#1803177)
Well, isn't the point of the park adjustment to give us a neutral context number in the first place? SO isn't it a little bit bogus to adjust a number for context when it's already been context-adjusted?

It'd be an argument against the accuracy of the context adjustment. Most context adjustments are rough and round off "real" ability under the assumption that every park effects every player in exactly the same way. They are very good in determining how many wins a player's run contributions created but not very good in translating actual player performance. BBRef would tell you that Fenway was a better hitter's park than US Cellular, for example, but Konerko would hit more HR's at US Cellular. Theoretically you could remedy this with "micro-PFs" for different components of hitting and/or by handedness, though you run into sample size problems there. I don't know how UZR is "park adjusted" for Fenway's LF, but if it uses a scalar then it's likely to miss some information in its translation. It's one place where a flaw leading to a seemingly nonsensical result could hide, and it's an appealing explanation because of Fenway's unusual LF and its effect on our valuation of UZR. I have no idea if it's really relevant. I'm not really arguing that there is a fenway-related problem, I'm just trying to wrap my head around a -50 from a corner OF.

I suspect a factor is other teams fielding better fielders as LFers, though. That would move the average higher relative to replacement and explain why Manny got so much worse relative to average without appearing to be that much worse than his prior "performance" to those who watched him. By ZR, Manny went from .750 last year to .729 this year while Podsednik (.898) and Crisp (.906) replaced Byrnes (.837) and Lawton (.798) as qualifying LF's. Holliday also improved from .798 to .890. Qualifying LF's on average went from .865 to .850 (though that's a flat AVG -- not weighted by ZR opportunities) so it looks like the league got better by about the same margin that Manny got worse. Manny gets a little worse + league average gets better makes the -50 make more sense.

And on the question of Manny Ramirez as a left fielder, their value in telling us that he is horrendous is much greater than their limits in telling us precisely how horrendous he is. He is horrendous enough that he surrenders a ton of his value as a hitter, and he is horrendous enough that he is massively overpaid.

I think "how bad" is more important in Manny's case than in other cases, though. For most players, their projected defense isn't that different from their past defense. For Manny, the swing could be 2-3 wins. In theory, we'd be having a similar discussion if a player were +50 in LF.

This, I think, is just silly. I can not claim to know all of the details of the methodology, but I absloutely cannot believe that a ball 1/4 of the way up the wall in Fenway would be judged as catchable.

Are there a fixed number of zones in UZR? (I thought there were, but I could be confusing it with another PBP method) If so, are those zones "sqeezed" by the wall at fenway and "stretched" towards the gaps and in the triangle? Or does Manny just forego credit on balls in zones covered by the wall? (I'd imagine a squeeze/stretch method would make Manny look better rather than worse, though.)


If that is true, and I have no reason to believe it isn't... then the -50 has no value in retrospect. Its like trying to measure offense without including doubles.


I'd disagree with this. Arm runs may have some predictive value, but depend far more on a small number of highly variable opportunities. The ~-50 base number is over a larger number of chances and, as a result, is more likely to be reflective of a "true talent" level in the player. It is also what has caused the hubbub over Manny's defense, and understanding how a LF could lose 50 runs on range would seem to me to be key in determining whether Manny really is "that bad" or if there's some problem in the tails/minutae of an otherwise accurate fielding metric.

Oh, my -32 includes arm, MGL's -50 doesn't.

What was Manny in your system without his arm?

Also, Darren, why regress to the mean for LF's his age? I thought the purpose of the regression toward positional mean in fielding metrics was supposed to account for the fact that managers/front offices still believed a player capable of playing the position. If that's the case, then why adjust for age? Is this the sunk cost fallacy rearing its ugly head?
   81. Mister High Standards Posted: January 02, 2006 at 09:57 PM (#1803179)

Do you think Williams is that bad in CF?


It is much more plausable for a CF to be minus 50 than a leftfielder. Which is why the defensive spectrum is set up the way it is.
   82. Chris Dial Posted: January 02, 2006 at 10:08 PM (#1803197)
MHS,
that's not what I asked you.
   83. Chris Dial Posted: January 02, 2006 at 10:12 PM (#1803203)
By ZR, Manny went from .750 last year to .729 this year while Podsednik (.898) and Crisp (.906) replaced Byrnes (.837) and Lawton (.798) as qualifying LF's. Holliday also improved from .798 to .890. Qualifying LF's on average went from .865 to .850 (though that's a flat AVG -- not weighted by ZR opportunities) so it looks like the league got better by about the same margin that Manny got worse. Manny gets a little worse + league average gets better makes the -50 make more sense

All my results are in the thread I linked too and I use a weighted ZR (league total)
   84. Mister High Standards Posted: January 02, 2006 at 10:24 PM (#1803216)
I don't know if Williams is that bad or that good. I watch less than 20 yankee games a year. I believe it is fundementally probable for centerfielder to be that bad not common, but something that may show itself to happed occasionaly over a number of years. I don't believe the same is true for a leftfielder.
   85. karlmagnus Posted: January 02, 2006 at 10:41 PM (#1803226)
-50 not using his arm would make sense; that would be Manny's fielding value if he had to catch the ball with his teeth or between his toes :-)

Otherwise it just proves the futility of the wilder sabermetric numbers. From observation, Manny drops a few, but he also catches some pretty good ones -- he's below average but much better than Bernie. In any case, if it keeps you awake at night, the problem can be solved by trading Ortiz.
   86. Darren Posted: January 02, 2006 at 11:15 PM (#1803298)
But Karl, didn't Duquette also try to sign Bernie Williams? Doesn't that given him the same infallability as Manny has?
   87. richie allen Posted: January 02, 2006 at 11:36 PM (#1803341)
We know that Johnny Damon was above average in UZR in 2005, right? MGL has said as much, tellingly adding that Boston doesn't like his defense as much as the numbers do.

David Pinto said this:

"What's clear is the Damon got to more balls than any other AL center fielder. My calculations say he should have gotten to even more."

So here's my question: has Damon been getting a lot of balls in Manny's zones? The Red Sox allowed an awful lot of fly balls, and Damon got to most of them. How does UZR handle fly balls in other outfielders' zones? Could this explain how so many fly-balls are not ending up in Manny's glove that UZR thinks should? It gels with MGL's point about the Sox thinking that Damon isn't as good as his numbers, which might imply that they don't think Manny's as bad as his numbers. Again, nobody's saying that Manny isn't awful, simply that it really does look like something fishy is throwing people off here.

Or is this too simplistic? The Red Sox are not idiots: if Manny was costing them 50 runs we have to think they would've done something about it (their scouts would have been beating down doors with blunt objects until they were hearD). BP has Manny at -16 for fielding. That seems a lot more reasonable.
   88. OlePerfesser Posted: January 03, 2006 at 12:11 AM (#1803394)
How does UZR handle fly balls in other outfielders' zones?

"Poaching" enhances the poacher's ZR, but does not directly penalize the poachee's, AFAIK.

That is, when the poacher goes into a neighbor's zone to record a PO, that is added to both the numerator and denominator of the poacher's ZR. It does NOT get added to the poachee's denominator as an opportunity not converted into an out, however, from what I've read. So poaching can make you look good, because you only do it when it's an easy play, while you never get penalized for failing to catch a ball outside your normal zone of responsibility (unless, perhaps, it gets recorded as an error).

The indirect effect of poaching, however, is that the poachee's number of opportunities to make plays is reduced, and if these are disproportionately easy plays (i.e., lazy flyballs where the CF has no problem ranging over), it's certainly conceivable it could bias the ZR results slightly. It may also be an issue in comparisons like this, where we're trying to score Manny relative to some average rate of plays made, and his number of "easy opps" is lower than average as a result of this phenomenon.

And to Chris D.: I'll check out the other articles on the runs/play research; thanks.
   89. dave h Posted: January 03, 2006 at 12:33 AM (#1803446)
Trying to get at a few separate issues here:

1) Does anyone know for a fact whether hits off the wall count as a LF's zone? I would think they do; unless there's a zone for "wall" (which would be nearly useless in most parks) they're going to count as a ball in play (speaking of which, does this explain one reason - other than the usual problems - why Sox pitchers might have lower DIPS ERAs than actual). Admittedly, I don't know the magnitude of this effect. And of course, this would be included in the "park factor", but no park factor is perfect.

2) If there is a variability in the measure, and of course there's a large one, that's important. Even if we know he's horrendous, it's important to know how much. We don't just say he's a fantastic hitter, and leave it at that. Half the point of these metrics is to be objective and half is to quantify.

3) I still haven't gotten clear as to whether the -50 is how many runs Manny cost relative to an average LF would given the same opportunity, or how many runs Manny would cost in an average park compared to an average LF.

I think there are potentially some other issues with uzr, not devastating, but enough so that if we would make a trade if he's expected to be -35 runs next year but not if he's -25, well, we have a problem.
   90. Darren Posted: January 03, 2006 at 01:11 AM (#1803511)
I'm not trying to be dense, but I don't understand how uncatchable balls of the wall would hurt Manny. Those are going to be uncatchable for everyone. When the park adjustment is made, shouldn't it wipe out all of those balls on both sides (Red Sox LF and visiting LF)? Maybe I'm misunderstanding the park adjustment.

I looked at the Chris's thread for the first time. Apparently, his, Gassko's, and Rallymonkey's systems all have Manny in the -30s somewhere for 2005. Based on that 1 year of data, and assuming all would be regressed at 50 percent, and subtracting ~5 runs for aging, they'd all project Manny to somewhere in the low -20s next year.

It's hard to say that Gassko and Dial's methods in anyway confirm what UZR is telling us because Gassko's was modeled in an attempt to recreate UZR and Dial's using the same source data. I haven't read anything about Rallymonkey's system, so maybe that's another, separate source.

Are there a fixed number of zones in UZR? (I thought there were, but I could be confusing it with another PBP method) If so, are those zones "sqeezed" by the wall at fenway and "stretched" towards the gaps and in the triangle?

According to the original article on UZR (found it here), the zones are fixed according to this chart.


Also, Darren, why regress to the mean for LF's his age? I thought the purpose of the regression toward positional mean in fielding metrics was supposed to account for the fact that managers/front offices still believed a player capable of playing the position.

You'd have to ask MGL about that. Remember, I started this thread by completely misunderstanding the point/method of the regression, so I'm probably not much help here. As far as I could tell from what MGL was saying, he used the age group because it's another factor, outside of the observed sample of data, that you know about the player.
   91. Sam M. Posted: January 03, 2006 at 01:20 AM (#1803537)
Sam your wrong. If Manny's overpaid because we are using the assumption that horrendous defense is worth -50 thats one thing. If we are assuming its worth -30 its another thing, as is -20 or what have you.

His horrendous defense does restrict his value but the question is magnitude. My position from the very beginning is that the numbers being thrown around are far too large for leftfielder, who has 17 assists, and is playing in a park that will "hide" his defensive short comings.


Except you are asking for precision when the very point you're making presupposes a subjective evaluation. I mean, trying to determine the extent to which Ramirez is "overpaid" is inherently almost impossible to do with any precision. When people talk about paying so many $$$ per marginal win, I just shake my head, as if there's a one-size-fits-all ratio for that assessment. What is "worth it" for the Yankees isn't the same as for the Reds. The 75th win isn't worth as much as the 92nd. The marginal wins a player provides to Team A may not be the same as for Team B, because the playing time opportunity (and the position he will play) may be different. Etc., etc., etc.

So if the whole idea of calculating the degree to which Ramirez is overpaid must be rough anyway, why should we expect a precise measure of his defensive badness? It really won't get us any certainty in assessing his value relative to his contract anyway.

IMO, here's what the defensive metrics tell us: anyone who would acquire Manny Ramirez to be anything but a DH, with his contract, is crazy. Unless they've got Ralph Kiner (age 83) playing left field.
   92. karlmagnus Posted: January 03, 2006 at 01:24 AM (#1803549)
Darren, you don't understand. Williams would have been Joe DiMaggio in center field if he'd ACCEPTED Duquette's offer, but since the ungrateful SOB turned it down he's well below average :-)
   93. Chris Dial Posted: January 03, 2006 at 01:33 AM (#1803571)
Does anyone know for a fact whether hits off the wall count as a LF's zone?

I know for a fact that some of them do. Not all of them. Some of them. Depends on how far the ball was hit.
   94. Darren Posted: January 03, 2006 at 03:23 AM (#1803692)
I mean, trying to determine the extent to which Ramirez is "overpaid" is inherently almost impossible to do with any precision.

I don't think the difference between a -15 defender and a -45 defender (both considered possibilities) is such a fine point of precision. In batting stats, you'd laugh if someone said player X had the same value as player Y, if one of them was 3 wins better than the other.

I know for a fact that some of them do. Not all of them. Some of them. Depends on how far the ball was hit.

But wouldn't they also count as being in the LF zone for all other LF playing in Fenway? And wouldn't the park adjustments catch that?

What's interesting, to me at least, is that UZR sees Manny has 15 runs or so worse than your numbers, despite UZR having a park adjustments and your system not having them. Why do you think that is? (The granularity? The adjustments?) On a related note, why do you think your system differs from all the others so much on Jeter's defense?
   95. dave h Posted: January 03, 2006 at 03:46 AM (#1803716)
I'm not trying to be dense, but I don't understand how uncatchable balls of the wall would hurt Manny. Those are going to be uncatchable for everyone. When the park adjustment is made, shouldn't it wipe out all of those balls on both sides (Red Sox LF and visiting LF)? Maybe I'm misunderstanding the park adjustment.

Ok, for the park adjustment, all LF in Fenway are compared to the same LF everywhere else over some significant period of time. This gives one number (.85 last I knew) which is divided into the number of outs actually made to make how many outs would have been made if the park played neutrally. The normal calculation is then made comparing how many adjusted outs were made compared to how many outs an average fielder would have made given the same number of chances in each zone. This is then converted to runs above average. At least that's how I understand it, I could certainly be wrong.

So the number is clearly park adjusted. But it's not perfectly park adjusted. First of all, to increase sample sizes, there is one park factor for the entire LF, fly balls and line drives. Last I knew, this factor was determined using 10 years of data. Also, it was regressed in some way. All of this means that the park factor, in my view, is somewhat suspect. I'm sure that .85 is better than 1.0, but for the time Manny's been in there, I don't know if it should actually be .75 or .8 or .9 or what. I believe this introduces a huge amount of uncertainty, on top of what is already present in uzr.

The end result, for me, is not that Manny should not be traded. It's that when determining whether a particular trade is worthwhile, it doesn't just matter that Manny is overpaid, it matters how much.
   96. Chris Dial Posted: January 03, 2006 at 04:19 AM (#1803764)
But wouldn't they also count as being in the LF zone for all other LF playing in Fenway? And wouldn't the park adjustments catch that?

As I note, 0.85 sounds pretty good. Just thinking about how I have seen it impact BIP.

What's interesting, to me at least, is that UZR sees Manny has 15 runs or so worse than your numbers, despite UZR having a park adjustments and your system not having them. Why do you think that is? (The granularity? The adjustments?) On a related note, why do you think your system differs from all the others so much on Jeter's defense?

I think the granularity *of my system* is better than UZR in that respect. Does mine differ that much on Jeter?

My system differs from UZR on Jeter due to the changing of zones for the SS, because MGL fuses some , and with the Yankees particular BIP distribution, this is bad.

My system differs from DSG's and Clay's because they try to estimate GBs from BIP/staff handedness, and we already know that this method is incorrect for the Yankees.

SO I differ for two different reasons.
   97. Sam M. Posted: January 03, 2006 at 04:51 AM (#1803822)
I don't think the difference between a -15 defender and a -45 defender (both considered possibilities) is such a fine point of precision. In batting stats, you'd laugh if someone said player X had the same value as player Y, if one of them was 3 wins better than the other.

Well, sure. But I have seen nothing on this or any other thread that leads me to believe that -15 is anywhere close to an accurate representation of Manny's defense in 2005 or his real ability level. You have to take the most optimistic (IMO unrealistic) number, then assume you need to make an additional park adjustment, then also regress towards the mean. Too many assumptions, all in Manny's favor, that are just not realistic.

I would be quite confident advising Omar Minaya that Manny Ramirez would be at least 25 runs worse in left field than Cliff Floyd, and even that would be assuming Floyd is significantly worse than he was in 2005.
   98. Matt Clement of Alexandria Posted: January 03, 2006 at 05:27 AM (#1803861)
Well, sure. But I have seen nothing on this or any other thread that leads me to believe that -15 is anywhere close to an accurate representation of Manny's defense in 2005 or his real ability level. You have to take the most optimistic (IMO unrealistic) number, then assume you need to make an additional park adjustment, then also regress towards the mean. Too many assumptions, all in Manny's favor, that are just not realistic.
We have three pbp-defense estimates of Manny's 2005 defense. They are -47 (UZR), -32 (Chris Dial), -26 (Chone Smith). We're looking at a difference of more than two wins. That creates a difference in future projection in the range of a full win. To me, that's a difference more than large enough to deserve some debate.
   99. RobertMachemer Posted: January 03, 2006 at 05:38 AM (#1803876)
For Manny to be a minus 50 or even minus 40 or 35 as a leftfielder he would have to all time bad. Biblically bad. Unbelievably bad. The worst aspects of Lonnie Smith and Kong combined. He isn't that bad...

How on earth can you know this? These sorts of play-by-play estimates of defense have only just started being used to estimate defense. How do you know that -35 is Biblically bad? How do you know that Greg Luzinski wasn't -90?

You make tons of assertions, but it's basically coming down to your gut feeling that -35 or -40 or -50 are too large (on an absolute scale). If you have anything other than a gut feeling, then include it. These "it's not true because it just feels wrong" arguments are not terribly compelling.

I'm not saying I am convinced that Manny Ramirez is -50. I have no idea (if I had to guess, I'd guess he was a lot closer to -30), but I don't just dismiss the -50 because it feels wrong. That's not how the scientific method works. I certainly would never claim that a -50 <u>has</u> to be wrong without having something a little more substantial than "it feels wrong" fuelling my assertion.
   100. dave h Posted: January 03, 2006 at 06:23 AM (#1803918)
It's not that -50 has to be wrong. It's just that we have some idea of what the likely values are, and it doesn't fit, and therefore deserves further thought. If, before some of the advanced metrics came around, someone told me that a batter had produced 200 runs more than the average player, I'd find that hard to believe, given the spread between low and high scoring teams. While our intuition is shakier for defensive contributions, that's partly because defense is more poorly measured, and therefore defensive metrics are more suspect.

In possibly the greatest season ever by a pitcher, Pedro prevented 64 runs more than a replacement pitcher in 217 innings. Manny played 1251 innings in LF (if I have my math right). So either Manny's performance was beyond historically bad, or the LF is no less than 1/6 as important as the pitcher in preventing runs (I realize that really we're talking about the spread since it's runs above average). Maybe defense is much much more important than I previously thought (I'd like that to be true since I used to be a good fielder and couldn't hit to save my life) or maybe Manny really did have a far worse season with the glove than Pedro had with his arm that year. But I'm going to need significantly more evidence to believe either of those conclusions. "Extraordinary claims require extraordinary proof."
Page 1 of 2 pages  1 2 > 

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
BarrysLazyBoy
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 1.0780 seconds
60 querie(s) executed