maybe sabean is right and old omar -is- undervalued...

then why did he still go ahead and pay above market price for him? i think it's unlikely anyone else was rushing to give him a contract as large as this one.

then why did he still go ahead and pay above market price for him? i think it's unlikely anyone else was rushing to give him a contract as large as this one.

The way the market has gone so far, with alot of fring players getting large contracts, it's possible that Vizquel's deal will be bellow market when all is said and done. I don't really believe that, or want to believe it, but it's certainly possible.

That bar graph seems totally sporadic. I'm totally unconvinced that the 37+ behavior is anything other than small sample size wackiness. The salary graph certainly seems to suggest that; why would players who are exactly 40 or exactly 37 have low contracts?

It is, at least in large part, smaple size wackiness. However, I think Studes does raise the interesting point (that I know has been discussed here before) that Sabean does seem to have a good record of getting worthwhile performances out of older players. It was at one time suggested that it was the doing of Baker, but it may be more reasonable to believe that Sabean identifies the right older players, e.g. Grissom and Snow. Does anyone know if anybody has looked extensively at this? It would be difficult to set standards for what constitutes old players, performance above average, and who is responsible, etc.

On a side note, could Baker have developed his love of older players as a result of Sabean's identical obsession? i.e. - Would Baker be more open to young players if he'd been surrounded by them at the beginning of his managerial career, instead of the vets he was given?

That's enough from me, for now.

I'm totally unconvinced that the 37+ behavior is anything other than small sample size wackiness.

I tend to agree, though it does make some logical sense to think that predicting future performance for an old player is more dependant upon scouting than statistics. What I mean to say is that a well conditioned 35 year old could be (I'd even say "probably is") less of a future risk than an obese 30 year old, and there would be no real way to assess that with just offensive stats and age. Alos, old player tend to retire rather than sit on the bench or embarass themselves, young players hang on.

then why did he still go ahead and pay above market price for him?

The question is its own answer. From what I read, a few teams had similar contracts on the table (2 yrs at $4M/yr). When the Giants offered a third year, they had a deal.

I suppose they could've added an extra $1M/yr, maybe that would've done it. We don't know. The point is, you almost by definition have to pay above the "market price" to get a guy.

What I mean to say is that a well conditioned 35 year old could be (I'd even say "probably is") less of a future risk than an obese 30 year old, and there would be no real way to assess that with just offensive stats and age.

This makes a lot of sense... normally, players' careers are ended by injuries from which they can't return effectively. physical fitnes/conditioning probably goes a long way towards "rehabability"... Mo Vaughn at 31 was a terrible signing, to illustrate your point about soft guys in their "peaks" still being a bad investment...

Vaughn was, of course, a bad investment for other reasons -- his non-fenway production just didn't justify a 6 year, 78M contract.

It's actually sample size that makes the graph seem sporadic. If I clustered age categories together into

- 28 to 31
- 32 to 36
- 37 and over

the trend would be very clear. However, grouping things is one of the ways that people manipulate data and I try not to do that.

Plus, I think the logic is self-evident. To repeat the article:

1. Most guys over 37 seem to have shorter term contracts, at today's lower prices, and
2. If a 38 year old stinks, he retires. If a 34 year old stinks, someone somewhere gives him another chance. Case in point: Roberto Alomar.

BTW, someone pointed out at least one error in my age calculations. I may have to recreate some of my math.

It's also worth noting that 3/12 probably overstates the size of the deal. According to the Chronicle, the breakdown of the money is:

2005: 2.5M
2006: 4M
2007: 4M
2008: 1M (deferred payment)
2009: .75M (deferred payment)

...it may be more reasonable to believe that Sabean identifies the right older players, e.g. Grissom and Snow...

How much of the credit has to go to Stan Conte and the rest of the training and medical staff for that matter? Both in terms of keeping players healthy once they arrive in SF and in terms of making evalutaions of which older players are most likely to maintain their health.

I raised this point earlier- if they wanted Vizquel, why didnt they get the Indians to pick up the option and trade a mediocre prospect to them ? I doubt the Indians expected to get a 1st round pick for him- now, they will do just that. That way, the Giants get Vizquel for 1 year and $5 million- far better than 3- $12 mil.

1 for $5 maybe better than 3 for $12, but only $2.5 is being spent on Omar this year in the 3 for $12 deal. I suspect that played a part. If Sabean is giving up draft picks he doesn't exactly have a long term view for the team - he's just playing for today, and this way the budget will go further paying Omar $2.5 instead of $5.

if they wanted Vizquel, why didnt they get the Indians to pick up the option and trade a mediocre prospect to them ?

For two reasons, I suppose. First, because Sabean wanted to burn the first round draft pick, just like last year. And second, because Vizquel might have vetoed the deal as a 10/5 man, so the Giants would have had to bargain with him anyway to get him to approve the trade.

How much of the credit has to go to Stan Conte and the rest of the training and medical staff for that matter?

Another satisfied Will Carroll reader. lol

It does make sense that a player that doesn't decline as markedly from ages 30-36 as say, Vaughn or Shawn Green, wouldn't suddenly begin a rapid decline from 37+, of course barring catostrophic injury.

Speaking of old guys, where is Steve Finley going this winter?

I have actually never read any Will Carroll.

I have actually never read any Will Carroll.

Really? I'm curious are you a Giants fan? I figure the only people who would know Conte's name are Carroll readers or Giants fans.

First, I was a Stan Conte fan well before i ever heard of Will Carroll. The key part of understanding what Stan Conte means to the Giants is that he has veto power over players playing, and can call for a pitching change based on pitcher fatigue. Whether he is the best trainer is another subject but he clearly has more authority than any other trainer. It makes for a much healthier team day by day.

As for the Omar signing, Sabean has made no bones that he is only concerned about next year...and getting the best player with as little of a hit on salary in 2005 as possible. and I think he has done that. I think what this signing indicates more than anything else is that Sabean sees a dramatic increase in salaries over the next few years, and is hedging against inflation.

As for the draft picks...I think, and it may be a bit presumptious, that we are seeing a sort of consipracy among some owners to opt out of the draft and concentrate the early draft picks among a smaller number of teams. The fewer teams with more picks in the first round the harder it is for people like Boros to play hardball.

If you got one first round pick you may be in a bind to sign a guy, but if you have three you can play them against each other.

Really? I'm curious are you a Giants fan? I figure the only people who would know Conte's name are Carroll readers or Giants fans.

Giants fan, though A's fan first.

It's interesting that you mention Will Carroll. In his most recent UTK column, here are his comments regarding Steve Finley:

Age is a misunderstood injury factor. The injuries differ, the healing time extends, but there's also a significant "survivor effect" once a player gets past 35. Most players aren't good enough to make it this long; the decline for the elite athletes who do is often shallow, especially if the player is also a workout fiend. Finley is the type of player who will choose when he leaves the game.

Hope the BPro guys don't mind my pulling the quote. This is sort of what my article is about, too. Players in their late 30's may very well be on a different aging curve, only we haven't discovered it yet because the sample size is too small.

One other thing about the Vizquel contract: if you take the cash flows built into the contract and discount them at 15% a year, you get a present value equal to two years at $4 million a year.

15% may be high, but it also may very well represent the Giants' focus on the short term, given Bonds' situation.

Speaking of Finley, his UZR's for the last 3 years are:

2000 -22
2001 0
2002 -5
2003 -53 (Did he play without a glove)
2004 -19

In prior years, he was bad as well, so it is not just age.

Somehow, he is regarded as a "gold glove CF'er."

If UZR is anywhere close to being correct about his defensive value, his defense pretty much negates his stellar offense. Someone is going to sign him for way too much money.

The Finley example brings up an interesting point. When players get into their 30's and especially their 40's, you have to pay particular attention to their defense. It is likely to drop off the charts such that the player actually has very little value overall. This is especially true in the OF, and even moreso in CF. It is doubltful that any player in his mid or late 30's can play an acceptable CF. Players like Finley, Grissom, Bernie, and Griffey, all have horrendous UZR's. They should all be at the very least moved to a corner outfiled position. Of course, then their offensive value goes way down. Here are those players' 04 UZR's (per 150 games):

Grissom -19
Griffey -52 (sad)
Bernie -44 (also sad)
Finley -19

Bonds was a -19 in left field. He should be a DH. Of course, he still as tremendous value as a LF'er because of his hitting, but that -19 takes a pretty big chunk out of his overall value. These numbers don't even include "arm ratings," which also generally decline a lot as a player gets into his 30's a 40's.

Here are the "arm lwts" for these same players:

Grissom -7
Griffey 1
Bernie -5
Finley -9
Bonds -2

With Vizquel, he was once a truly GG caliber SS and his defense has not declined that much. Here are his UZR's over the last 5 years:

2000 +1
2001 +12
2002 +5
2003 +2
2004 -5

We expect something around the low minuses next year, and he can still hit, so this was indeed not a bad signing.

However, for some of these other older players, teams that do not know how to properly evaluate defense (most teams) are going to pay some of these players far, far more than they are worth...

Awesome insight, as usual, MGL. Thanks.

Question from the peanut gallery, is UZR expressed in runs, or is it expressed in something else?

it's runs.

I second studes' comment about MGL. Studes and MGL are two of the best things going here. Having the two of them together in a thread is like a mayonaise soda, or even better.

Mayonnaise is gross; MGL is tasty.

"2003 -53 (Did he play without a glove)"

WOW thats a lot of negative runs. I really didn't think that was possible.

"mayonaise soda"

You really just have the worst taste ever.

Spider, UZR is expressed in runs saved above average, often pro-rated to 150 or to 162 full games in the field.

Elsewhere, MGL, you posted that you've revised the '04 UZR numbers after finalizing your info from the season, such that numbers you posted in the past few weeks are no longer the most accurate. I think that includes the cool lists of gold glove worthies you'd posted, which I (and I think others) have saved for reference.

Will you be posting a full set of the numbers at some point? (...sounds of everyone pleading and cheering...)

As a big Cardinals fan, I most particularly like to know about Albert, Rolen, and Edmonds. But all the info is enlightening, and I often pull up a spreadsheet of the '00-'03 UZRs to help evaluate trades'n'signings or sportswriter opinions I'm reading about. Now that '04 is over, that spreadsheet is missing the most recent illumination...

Notice this:

"2001 0"

No wonder that was the only year AZ won the WS.

:-)

Finley though has lost a lot in his defense. His first step is slow, his read on lineouts is bad, his arm is weak. I was glad AZ traded him for prospects.

MGL, I'm not sure if you've done the UZR for guys who played in 2004, but I'd love to get the UZR on Luis Terrero in CF for AZ. A reading on his arm strength would be appreciated as well. (If you have them, of course. As always, your comments are much appreciated).

Thanks, Bellhorn-o-copia (nice seasonally-related handle, BTW).

Seriously, -52 or -53. How do you do that? That's one more run every three games. WOW, its mind boggling. Good thing Finley won a Gold Glove.

"Elsewhere, MGL, you posted that you've revised the '04 UZR numbers after finalizing your info from the season, such that numbers you posted in the past few weeks are no longer the most accurate. I think that includes the cool lists of gold glove worthies you'd posted, which I (and I think others) have saved for reference."

I have tried to commit all of those numbers to memory because I rarely am on the same computer. I will have wasted much of my mental capacity if they are no longer accurate.

"Will you be posting a full set of the numbers at some point?"

OH PLEASE. PLEASE PLEASE PLEASE.

We expect something around the low minuses next year, and he can still hit, so this was indeed not a bad signing.

Yeah, but what can we expect in 2007? I don't think many would have objected to signing Vizquel to a one- or even two-year deal.

By the way, I still stand by my believe that Sabean is a total crock of a GM who gets lucky with his free agent signings and that his failures are far too often overlooked (see Micheal Tucker).

Good thing Finley won a Gold Glove.

Heck, with Jeter winning a gold glove in the AL, someone worse had to win a gold glove in the NL. IMO, Jeter won it with his dive in the seats, Finley won it with his grand slam against the Giants.

"Heck, with Jeter winning a gold glove in the AL, someone worse had to win a gold glove in the NL."

Worse players won it in both leagues jackass (Suzuki in the AL). If Jeter had won with a -19, primates would riot.

Easy there, Mr. "gold glove at SS admirer." Keep your kind words for people who would put up with your crap. Jeter would have won the gold glove this year even if his UZR was -59. It was all about the swan dive.

It's also worth noting that 3/12 probably overstates the size of the deal. According to the Chronicle, the breakdown of the money is:

2005: 2.5M
2006: 4M
2007: 4M
2008: 1M (deferred payment)
2009: .75M (deferred payment)

Not really. That is 12.25 million versus 12. Even including discounts, that is a NPV (at 5%) of 10.87 versus 10.89. It saves a little money this year, but is costs i 08 and 09.

"Keep your kind words for people who would put up with your crap."

How about you worry about facts when you make your ######## posts. ####, you attacked Omar Minaya earlier today for making a trade as the Expos GM years before he took the job. Do you even think about what you're going to post or do you just guess and start typing?

Do you even think about what you're going to post or do you just guess and start typing?

I just stop and wonder "What Would Jeter Do"?

Overlooking the fact that Omar wasn't the one who got shafted in the Irabu trade is one thing; calling people "jackass" is another. I'm sure you are used to calling people "jackass" in your everyday life; I am sure you're miserably and lonely too.

"Overlooking the fact that Omar wasn't the one who got shafted in the Irabu trade is one thing; calling people "jackass" is another."

Right, because your slam at Jeter in post 34 was completely innocent. Get serious.

"I'm sure you are used to calling people "jackass" in your everyday life; I am sure you're miserably and lonely too."

Yes and no. Jackass.

OK, fine. Suit yourself.

Comparing a "slam at Jeter" with calling someone a "jackass"? That's New Jersey mentality for you.

Have a nice day.

"That's New Jersey mentality for you."

####### right.

By the way, I still stand by my believe that Sabean is a total crock of a GM who gets lucky with his free agent signings and that his failures are far too often overlooked (see Micheal Tucke


Tucker had an OPS of .822, and ended up with 15 WS- at 2 million, that's a pretty damn good deal. I cursed about it, and there's plenty to curse at- but calling Tucker bad last year is idiotic.

It's been very nice to see Joe Sheehan, in today's and another recent article, mention UZR in his analysis in addition to BP's historical-stats-based fielding numbers.

"Tucker had an OPS of .822, and ended up with 15 WS- at 2 million, that's a pretty damn good deal. I cursed about it, and there's plenty to curse at- but calling Tucker bad last year is idiotic."

A .266 EQA out of a corner outfielder is my idea of a bad year unless he's a great defensive player, which he's not. So look deeper before you call someone idiotic, there are some sensitive types running around on this board.

All right, we've got a guy who likes to call people "idiots" and another one who likes to call them "jackasses". And they are at it as we speak. I bet two bucks on the New Jersey guy...

By the way, I still stand by my believe that Sabean is a total crock of a GM who gets lucky with his free agent signings and that his failures are far too often overlooked (see Micheal Tucker).

That a pretty bad example, J1F. By RAP, Tucker was 3 runs below the average RF (he played 25 games in CF). By UZR, he's an average RF. So he's basically an average OF that cost ~$2M. That's not a bad deal at all.

Danny, why try to reason with J#1F? You know he'll simply throw some tasty insult at you. Unless you offer to declare that Jeter is the greatest SS that has ever lived...

Man, this thread has gone downhill fast.

It was nice to see Joe Sheehan reference UZR, but does BPro have access to this year's full set of numbers somehow?

Hmmm.

"That a pretty bad example, J1F. By RAP, Tucker was 3 runs below the average RF (he played 25 games in CF). By UZR, he's an average RF. So he's basically an average OF that cost ~$2M. That's not a bad deal at all."

Didn't really think about it that way. There are about 20 or so rf who outperformed him this year is more what I was looking at (20 according to VORP, 22 according to RARP). Plus the fact that they already have Dustin Mohr, who was just as good and probably could have been better. It also cost them draft picks but those are pretty much worthless for Sabean anyway (his inability to draft has clearly affected his decision making in the offseason, so I think it could count against him in this deal). It seems like a wasted two million (1.5 according to B-reference).

"Danny, why try to reason with J#1F? You know he'll simply throw some tasty insult at you. Unless you offer to declare that Jeter is the greatest SS that has ever lived..."

Thanks jackass.

Hehe. Like Jeter, like fan. Totally overrated.

I can't really publish full-scale UZR's anymore. Unfortunately, the 2004 ones I have been "leaking" were not very good for various reasons.

Tucker was 0 in 04 in RF. In the past, he has excellent UZR's in RF and negative ones in CF and LF (RF appears to be an easier position for fielders than LF, by around 3 runs per 150).

Terrero had a terrible UZR in CF and in a few games in RF. In CF, it was -11 runs in 51 defensive games.

Edmonds was -24. Don't know what's up with him. He has been negative for several years now. Could be that he plays too shallow. His arm rating is the best in baseball, at +9 (per 150). Terrero's was -4 per 150. Rolen was +33 in UZR, great as usual. He and Beltre are the best. Renteria had another good year, at +12.

OK, Here are best and worst in each league at each position per 150 "games" (min 95 games):

NL

1B

Casey +15
Pujols +14
Wilkerson +12
Helton +7

Thome -21
Hillenbrand -11
Nevin -9
D. Lee -8

(Piazza was -19 for you inquiring minds.)

2B

Kent +20 (who woulda thunk?)
Polanco +11
Womack +9 (another surprise)
Miles +7

Vidro -19
J. Castillo -12
Durham -11
Cora -7

3B

Rolen +33
Beltre +29
Castilla +16
Bell +15

A. Ramirez -15
Lowell -13
Ensberg -12 (92 games)
Batista -6

SS

Renteria +12
Iztruis +10
A-Gon (FLO) +10
K. Greene +8

Kaz -18
Larkin -13 (87 games)
Furcal -12 (hurt?)
Cruz -8

LF

Genkins +25 (should be CF'er)
Bay +11 (what a player!)
Alou +10 (weird)
Conine -2

There are lots of full-time terrible fielders in LF, even though everyhting adds to zero.

Dunn -20 (I think he is a DH-type)
Bonds -19 (what do you want at 41?)
Floyd -19 (more good news for you hapless Met fans)
Burrell -13

(Biggio was -33 in left and -13 in CF. The guy has no value as a player anymore)

CF

Payton +34
Patterson +32
E. Chavez +22 (sleeper)
Redman +20

Finley -28
Edmonds -24 (as I said, weird)
Grissom -19
Burnitz -22 (52 games)

(Griffey was -52 in 78 games)

C (based on errors, PB, WP, and SB/CS allowed)

Schneider +14
Matheny +5

Estrada -5
Barrett -4

(Piazza -19 in 42 games)

AL

1B

Broussard +15
Erstad +14
Delgado +12
Tino +5

Konerko -17
Hatteberg -4
C. Pena -3
Olerud -11 (73 games)

2B

Cairo +20
Kennedy +12
B. Roberts +9
Boone/Hudson +6

Soriano -15
J. Uribe -13 (68 games)
Belliard -11
Scutaro -8

3B

Randa +19
A-Rod +17
Koskie +11
Chavez +9

Crede -16
Munson -13
Hinske -12
Huff/Inge -11 (66/78 games)

SS

Tejada +20
Valentin +20
Lugo +12
Crosby +4

(Jeter 0)

M. Young -32
Berroa -19
Guillen -7
Guzman -6

LF

Crawford +23
Bigbie +12
C. Lee +12
Ibanez +6

M. Ramirez -26
J. Guillen -25
Byrnes -24 (93 games)
Matsui -11

CF

Winn +22
Rowand +16
Hunter +16
Kotsay +9

Bernie -44 (argh!)
Nix -23
Baldelli -15
Anderson -17 (87 games. He is not a CF'er.)

RF

Dye +14
Higginson +13
Jones +3
Gerut +1


Sheffield -11
Surhoff -9
Ichiro -9 (good arm though)
Cruz Jr. -7

C

Wilson +6
Blanco +5

(J. Molina +15 in 59 games)

Posada -4
Pudge -2

I love you, MGL!

Big wows for Tejada and Byrnes.

Dunn -20 (I think he is a DH-type)

Wasn't Cincinatti's Assistant GM saying a few years ago he thought Dunn could handle center if he had to? RDF.

For the players I see regularly, UZR pretty much agrees with what I see (though Jose Guillen's arm rating would have to get some of that UZR back).

It sure looks like defense really does knock Manny Ramirez and Sheffield away from Vlad; those were the two position players closest to him offensively, I think. Miguel Tejada probably has an argument with his defense, but Guerrero was far enough ahead with the bat that I don't really think the glove and position make up it all.

I wonder why Kent was so bad in the NLCS. Was he hurt?

Yankee Outfield: -66. Jesus Christ.

So, uh, am I the only one here who doesn't worship UZR yet? Any system in which Jay Payton is worth 58 more runs on defense than Jim Edmonds, and Tony Womack is solidly above average, is wrong.

Actually if we had an MVP award for lwts+UZR, and adjusted for position, Tejada would beat Vlad hands down. Tejada was +20 UZR plus +23 in batting lwts, or +43. Vlad was -1 UZR (0 in arm) and 43 in batting lwts, or +42.

There is a 17 run difference on the average (for the last 5 years) between SS and RF (the average RF'er hits and runs 17 runs better than the average SS)! That puts Tejada at almost 2 "wins" better than Vlad this year...

Somebody might be getting a real bagain on Lieber, then. If a team with good OF defense signs him, he could be a half-run better than average instead of a third.

There is a 17 run difference on the average (for the last 5 years) between SS and RF (the average RF'er hits and runs 17 runs better than the average SS)! That puts Tejada at almost 2 "wins" better than Vlad this year...

What's the defensive difference? Let's say we had a SS that had +20 UZR and a RF with a +20 UZR; I would assume the SS's defensive contribution is greater, but by how much?

(I am not a fan of putting positional adjustments on hitting stats, for the most part).

So, uh, am I the only one here who doesn't worship UZR yet? Any system in which Jay Payton is worth 58 more runs on defense than Jim Edmonds, and Tony Womack is solidly above average, is wrong.

I am the messenger. The actual data is the message. These numbers are exactly what "happened" in the field. Like any other sample metric (stat), it doesn't mean that it represents the true defensive value of the player. If you want to estimate the true defensive value of a player, again, like anything else (e.g., offensive value), you have to look at as much data as possible (multi-years generally) and then make some inferences.

Womack was a bad SS. He should pick up some runs at 2B, but probably had a "fluke" season. As I said, Edmonds numbers are surprising. Perhaps he is not as good as people think his is, especially now that he is in his mid-30's. Perhaps he plays so shallow (which he does) that he looks like a good fielder, but he is out of position. Payton has compiled great UZR's for the last 5 years. Obviously this season is a little flukey in terms of his actual defensive value.

Bj (and others) have correctly said that if you are not surprised (I say shocked) at some of a mettic's numbers, it is not a good metric.

What is the point of an advanced metric, if you only validate it if it agrees with what you think you know? An advanced metric, especially a defensive one, is supposed to tell you what you don't know. And defensive value is near impossible to gauge by observation, for several reasons. Certainly scouting an observation can help to nail a player's true defensive value.

If you (Ankiel!) understand anything about sampling statistics, which you probably don't, "by definition" a certain percentage of the values in a sample measurement will automatically be off by more than a certain amount, depending on the standard error or standard deviation of the measurement. That is true even if the measurement is "perfect"...

BHW, positional adjustments MUST be done when comparing players. The easiest way of doing them is to adjust for the average hitting in each of the defensive positions.

Presumably this will equal the difference in true defensive ability across the positions as well. IOW, since there is a 17 run (per 150 games) difference in hitting between the average SS and average RF'er, there should be around the same difference (17 runs) if you moved an average SS to RF or vice versa.

Here are approximate UZR (per 150) adjustments across defensive positions, based on players who have played more than one position over the last 6 years. Obviously there are different skills necessary for different positions, so you can't just put a player at any position and expect these numbers to hold up. However, they are a good guide:

3 -9
4 +4
5 +2
6 +7
7 -1
8 +4
9 -4

So an average SS would be around +11 in RF, rather than the +17 based on the hitting. Maybe that is a better way to do the positional adjustments. I don't know. The reason it is a lot lower than the 17 runs is that managers sacrifice lots of hitting at SS on players who they think are good defensively, but are not (Guzman, Perez, etc.).

Tango has a nice article on his website where he looks at positional adjustments. Unfortunately he never followed up the article with more and better research, which he had planned to do...

The big shocker to me is Jose Guillen, who did very well in RF in previous years. Manny is no surprise and I'm really hoping that the Red Sox are able to unload that contract on Minaya before he pulls the trigger on Sosa.

mgl, considering defense, how close did Beltre and Rolen get to Bonds in overall value? btw, thanks for posting the #'s.

Presumably this will equal the difference in true defensive ability across the positions as well. IOW, since there is a 17 run (per 150 games) difference in hitting between the average SS and average RF'er, there should be around the same difference (17 runs) if you moved an average SS to RF or vice versa.

I understand that this is traditionally done, and why, but I think it's a large presumption that the offensive difference is equal to the defensive difference.

Cross, well, Bonds was about 70 runs better offensively than Beltre or Rolen. He was, however, around 50 runs worse defensively. There is also a big difference between the positions. The positional adjustments I use have a 14 run difference between LF and third base. So yes, the gap is closed quite a bit in overall 04 value between Bonds and Beltre or Bonds and Rolen. That's not even including baserunning. I haven't done the 04 baserunning yet, but I suspect that Bonds is going to be pretty bad in that department as well. So you certainly could make an argument for Rolen or Beltre as MVP over Bonds, if that's what you were implying...

Amazing stuff, mgl. Thank you so much!

Kevin, I don't think you can ever "predict" that someone will be a +20, even though there probably are some "true +20's" at SS. That's kind of a philosophical/mathematical thing though. I suppose that if a player posted several years of great numbers and you were convinced from observation, scouting, etc., that he were an absolute whiz with the glove. I don't know that you can say that about Crosby.

+20 is probably the absloute top tier of true defensive ability at any position (of course lots of payers will have sample UZR's higher than that), the Ozzie Smith's, Willy Mays', Maseroski's, Brooks Ropbonson's, etc., in their prime...

thanks, mgl. I think it's interesting that even the stat-savy here at primer took it for granted that Bonds was worth more runs than anyone else and really underestimated how big a difference defense can make.

That is true even if the measurement is "perfect"...

Which there's no way to know whether your measurement is perfect or not. It sounds like you have good data (although there are problems with zone data), your general methodology sounds reasonable, but until you tell people precisely how you get your ratings (all the adjustments you make, etc.), there's no way to know.

I'm not saying you're wrong to be secretive; I would be too if it meant a job in baseball. I'm just saying that it's not necessarily ignorance of statistics that causes people to object.

And I don't see what would be worthless about a statistic that generally agrees with our visual assessment of a players' defense. The important aspect of defensive evaluation isn't revealing bad defenders to be good, it's quantifying defense that's important. It just so happens that in quantifying defense we often find out surprising things about players.

I agree that observation is inadequate. Still, how do we know that visual assessments are wrong? Because advanced metrics say so? Or because there's often disagreement among them? If it's the latter, then I don't see why defensive metrics are any different than observation.

Then again, I've become a bit skeptical about the possibility of adequately quantifying defense at all.

Yay Kaz and Piazza!

Has anyone been having a problem lately with thread lag? For example, I can see that Jeremy H. has posted here, but until I post this, I can't see his comment

How did Manny rate - overall LWTS - compared to the average LF? How far is he from being worth $20M per year?

It looks from the last few years UZR that he may be a true -20 in the outfield, and his baserunning LWTs have been pretty weak, too. I'm definitely reconsidering the merits of the Crazy Dump Manny Plan.

From my observation in the playoffs, Edmonds plays way too shallow. I thought the doubles over his head in Game 2, well, not that they would've definitely been caught if he'd been playing more traditionally, but he wouldn't have been so far away. But that's only a few games, on tv, so it's not worth much.

And this:

I can't really publish full-scale UZR's anymore.

is the saddest sentence of the offseason.

Has anyone been having a problem lately with thread lag?

I actually have too... especially on that thread that had the fake Beltran to the Mets rumor...

And RB, I'm just kidding with you by bringing it up, but in all seriousness, I did have problems with lag on that thread.

Posting again, to see what Jeremy has to say...

Beltran to Mets rumor? What kind of ass would suggest that? What is this, an ESPN board?

Fancy, actually the entire UZR methodolgy, inlcuding all the adjustments, are described in my UZR articles. I don't know the web address offhand. It is in the old Primer archives, which I think are still available on BTF. Do a seach for UZR. It's nothing earth-shattering. Just a basic way of taking the PBP data (batted ball speed, type, and location) and using it to quantify a player's affect on his opponents' hit and out rate as compared to an average player at his position. Not rocket science. Lots of people can and do duplicate the basic methodology...

Yup, Manny's peripherals are terrible, probably -6 to -8 a year. He is one of the best hitters in baseball, but with his defense and baserunning (and GDP's), he is worth nowhere near 20 mil a year. In fact, he is probably worth around 10. Although hitting indeed comprises the plurality, maybe even the majority, of a player's value, once you include all of a player's skills, assuming you know how to estimate their true value, you quite often get quite a different picture of a player.

If a team ignores or even uses observation/scouting/reputation to assess a player's defensive value (and other peripherals), they are going to make some horrendous mistakes in valuing players for trades, salaries, etc.

FWIW, I agree that Sabean is one of the poorer GM's in the business, at least in terms of evaluating player value. He appears to be ignorant of even the most basic of evaluation tools. In this day and age, there is no excuse for that. He may have many other good qualities, however. A poor GM (or manager) can be very successful (in terms of his team winning) and a good GM can be unsuccessful in the same way that a lousy blackjack or poker player can win any number of dollars in any finite amount of time, and vice versa (the best card counter in the world, playing a liberal BJ game, can lose...). Especially when one adds payroll to the mix. Even a poor GM is likely to have a successful team if he spends a lot of money, and vice versa...

mgl, what was Cameron's 2004?

Cameron = +9 per 150 in 140 games....

Okay, we can now eliminate the Giants as the team that MGL works for.

I am the messenger. The actual data is the message. These numbers are exactly what "happened" in the field.

Well, not exactly. You translate the data to a less accurate form, and based on your numbers, you do something that gets silly high (low) negative numbers.

In addition, you use the actual value of a missed batted ball, rather than the league average (or some equivalent). this is clearly wrong because at that point, you are adding in not just the fielder in question's skill, but the quality of the player running down the ball he missed.

The 56 hole in Project Scoresheet is a poor reflection to represent the range of a SS or 3B. For one, it weighs it far too heavily as a "difficult" play under the guise of a play being really far away.

For example, (assuming you get STATS data), you assign equal value to a ball hit 30 feet to the right as a ball hit 15 feet to his right. that's flat out wrong analysis.

In addition, doing every player, and extrapolating them to 150 games when they played.

Terrero played 488 inings in CF and posted a poor ZR of 0.840. That's a worse rate than Finley. The average CF converts at a rate of about (what?) 0.86-.87. That's just not 30 runs difference over 150 games.

While Bill James has said your metric should surprise you, if it gives you really funky numbers, you may want to re-check your math.

Maybe I've misunderstood UZR in the past--a couple questions:

--Do you buy the raw STATS data? How much does it cost? How is it delivered?

--When you're trying to assing a run value on a missed ball, how do you arrive at that value? Is it, as Chris suggests, based on the actual results of a given play or is based on the average results of a ball hit at speed X to spot Y?

I'd look it up on your old articles, but I've never been able to use the search or archives on this site.

Wow, I can't believe Byrnes is that bad, and in LF, where he's being compared to some awful defenders! I always thought Byrnes was not as good a defender as he should be given his speed, but -24!!! That's just bizarre.

I agree that observation is inadequate. Still, how do we know that visual assessments are wrong? Because advanced metrics say so? Or because there's often disagreement among them? If it's the latter, then I don't see why defensive metrics are any different than observation.

Then again, I've become a bit skeptical about the possibility of adequately quantifying defense at all.

Very well put. I think this pretty much sums up my assessment of the present situation.

Not meant in any way to disparage the terrific work of mgl and so many others. This is a very, very complicated and difficult task. But it is just a very, very long way from being anywhere close to being accomplished.

Darren,

When a batter misses a ball, the value of the hit is the average value of a hit (and error) in that location given the type and speed of the ball, as reported by STATS.

Well, not exactly. You translate the data to a less accurate form, and based on your numbers, you do something that gets silly high (low) negative numbers.

You are right in that I translate the STATS data, which is more granular into retrosheet format, where the zones are larger. Before I got STATS data, I used the retrosheet data and my programs are set up for that. I could easily modify the programs to accomodate the STATS more granular data, but I haven't yet. I don't know if it's going to make that much of a difference. I doubt it. And I think my numbers will be less extreme (less variance) and not more extreme, but I, not sure OTTOMH.

In addition, you use the actual value of a missed batted ball, rather than the league average (or some equivalent). this is clearly wrong because at that point, you are adding in not just the fielder in question's skill, but the quality of the player running down the ball he missed.

If I understnd you correctly, I don't. As I explained above, I use the league average value of a hit to that sector, given the speed and type of ball. If I said otherwise in the articles it was a mistake.

The 56 hole in Project Scoresheet is a poor reflection to represent the range of a SS or 3B. For one, it weighs it far too heavily as a "difficult" play under the guise of a play being really far away.

I don't understand what you mean. The 56 zone is a combination of 3 sectors of STATS data. As I said, it would probably be better to use the 3 zones separately, but I don't and I don't think it matters all that much. Besides, for some strange reason, the STATS data I get records the area in the outfield where ground ball hits are fielded, so you have to interpolate where they left the IF anyway.


For example, (assuming you get STATS data), you assign equal value to a ball hit 30 feet to the right as a ball hit 15 feet to his right. that's flat out wrong analysis.

Well, 5 feet would be better than 15 feet as well, but that's the way it is. It is not "flat out wrong." Each sector I end up using is around 25 feet wide. I could use the 8 foot wide sectors (or so) in the IF that STATS uses, but I don't. I use the 25 feet wide sectors that retrosheet uses. Using the STATS sectors would be better than using the RS sectors of course, but using the retrosheet ones is not "flat out wrong." It is just not as good as using the smaller ones. And the retrosheet sector down the lines in the IF, is the same as the STATS sector (normal 8 foot wide one), for what that is worth.

You have an "interesting" way with words Chris. One reason I have referred to you as a "prick" on occasion. I do appreciate the critical comments though. I may redo the programs to see how much the numbers would change if I use the STATS sectors, if that's what you are suggesting. As I said, I don't think it will change all that much. In any case, it will just make all the numbers more accurate. Sort of like the difference between OPS and lwts. OPS is not "wrong". It is just not as good as lwts because the data is not being used in the most granular fashion.

In addition, doing every player, and extrapolating them to 150 games when they played.

Terrero played 488 inings in CF and posted a poor ZR of 0.840. That's a worse rate than Finley. The average CF converts at a rate of about (what?) 0.86-.87. That's just not 30 runs difference over 150 games.

Again, I don't know what you mean. If you mean that if a player has 15 games and he has saved his team 3 runs, then he is not a "+30" fielder, that is correct of course. I use a fielder's UZR rate and the number of games he played to establish that rate. The "user" can do anything he wants with that. I am just giving the raw data.

While Bill James has said your metric should surprise you, if it gives you really funky numbers, you may want to re-check your math.

Have no idea what you mean. Did I make a math error somewhere?

Did I make a math error somewhere?

It's certainly possible, although there's no way anyone but you could find it. I recognize that this is a limitation imposed upon you by the data you use, but peer review doesn't just catch theoretical or methodological flaws.

Although I can't comment on the validity of Chris's specific criticisms--I'd be willing to guess I'm not alone there--I think they are illustrative of a basic problem that UZR poses: The methodology is intimately tied with the data. Very few are equipped to make judgments about the limitations of the data or how you handle it. And even Chris can only surmise based on the articles, which are concerned more with the methodology than the nitty gritty of STATS zones. (Not a criticism)

The greatest virtue of UZR, the data, prohibits it from being fully peer reviewed. Which is fine. I keep UZR in my analytical toolbox. I'm just not convinced that I should throw away my hammer or my screwdriver.

For the curious:
UZR part 1

UZR Part 2

I understand that this is traditionally done, and why, but I think it's a large presumption that the offensive difference is equal to the defensive difference.

I don't think that's necessarily what people are presuming. There's a different pool of players eligible to play LF than SS. The comparisons that MGL posted (and that Tango has made in the past) are based on players who have played multiple positions and comparing their performances.

But there's more to it than that. For instance, left-handed hitters can't play the throwing infield positions. So the pool of outfielders and 1B contains far more hitters. That may explain why the difference in batting performance is larger than the difference in observed fielding ability. It's debatable whether to credit players for that additional difference.

Of course, it depends on what your question is. If you're looking at actual contribution to winning games, a RF that is +20 offensively has contributed just as much with the bat as a SS that is +20 offensively. However, a RF that is +20 at his position defensively has likely contributed less with the glove than has a SS that is +20 at his position defensively.

Just looking at the offensive side of it, maybe that +20 RF is only +5 above average for his position, where the +20 SS is +25, so the SS is rarer, and deserves to be paid more and is less replaceable. But in terms of their actual contribution to winning games offensively, they are the same.

I think that what Tango figures is that every player comes to the table with X hitting talent and Y fielding talent, and his value and hence salary (experience, status, and age not withstanding) is simply X+Y regardless of what defensive position he actually plays.

If that is the case (which I think it roughly is), then once you establish a player's fielding talent independent of his actual defensive position, you know his value. In reality, we don't know how to quantify a player's absolute (indpendent of position) fielding talent, even if we had perfect defensive metrics. We can only estimate their fielding talent relative to other fielders at the same position, using metrics like UZR. So if we had some way of establishing each positions worth, relative to the other positions, we can solve the overall problem (being able to compare players across positions). That is what Tango attempted to do with his study, and that is what I have attempted to present with my chart above - the relative values of the various positions...

From UZR Part 1:

After discovering that Shorty had indeed found an error in the UZR methodology, mgl wrote:
Well, all those months and sleepless nights for naught! Anyway, I'm glad you caught the error and I thank you (and the sabermetric community should thank you as well). This is one of the reasons why new and important metrics should not be proprietary. Peer review and open source is a must!

I completely agree with that last statement. Unfortunately, that doesn't appear to be the case anymore with UZR.

mgl,
I've asked that set of questions at least two other times. If there is some "tone" to them, it is due to having to ask it repeatedly.

Well, 5 feet would be better than 15 feet as well, but that's the way it is. It is not "flat out wrong." Each sector I end up using is around 25 feet wide. I could use the 8 foot wide sectors (or so) in the IF that STATS uses, but I don't. I use the 25 feet wide sectors that retrosheet uses. Using the STATS sectors would be better than using the RS sectors of course, but using the retrosheet ones is not "flat out wrong." It is just not as good as using the smaller ones. And the retrosheet sector down the lines in the IF, is the same as the STATS sector (normal 8 foot wide one), for what that is worth.

Okay, it is NOT flat out wrong. I overstated that. I am wrong on that comment. It is definitely better than WS or Clay's stuff (no offense to Clay) or Michael Humphries stuff that I have seen (or Charlie Saeger, again, no offense). I hope I didn't leave anyone out.

The OF "down the lines" as I am looking at the grids appears to be (assuming you mean the 5 set set) is about two ZR zones (C and D). Not 25 feet wide, but 16 feet. I am certain that on the Project Scorsheet (PS) grid the 5/5S zone cuts right past the 3B cutout where as on the STATS grid, both zones C and D are within the cutout.

In addition, the treatment of your data would "work" better - significantly - if you attempted to do what Mike Emeigh did.

How much does positioning matter? Did Jeter make more plays this year in the F zone or did he make more plays in the L/M zone? Or did he just make more plays in the 6 zone?

There is tons of speculation done on why Jeter saw such a rise in ZR. My personal belief is that he worked a lot on bench=pressing and armstrengthening so balls in the hole (the 56 zone) were outs more. Heck, all of his plays turned into a few more outs.

You have the answer to that question *if you don't translate the data*. You know whether Jeter increased his out rate in the L zone (or you *could* know). The L zone *is* the SS responsibility (defined as MLB SS turn more than 50% of balls hit into that zone into outs), but PS does not.

I'm not sure either how much this matters, but after my previuous work with the data, I don't believe that the range of numbers you generate can fit logically with the ZR data.

Russ Steele (I think) and I had a similar discussion when STATS UZR first came out, but I didn't get it then, so maybe I don't "get it" now.

The first step in all of this (IMO) is recognizing and using zones as defined. For starters, it will open up the biases of each fielder. You can tell that Omar Vizquel makes more plays to his left. That Jeter makes no plays more than 16 feet away. That Andruw Jones is catching far more short balls (LDs) than other CFs. This mining of the data tells us a good deal about positioning and then allows us to assign the proper value of that positioning.

Tris Speaker used to play a brutally shallow CF. He has a silly number of unassisted DPs. His motto was "I can cut off a ton more singles so the occassional triple is no big deal."

Is that true? IS what Andruw Jones does in the OF valuable? I run AJ down a lot for taking discretionary outs, but I recognize it is more than just possible that he catches a different subset of BIP and that new set of BIP is enough more valuable to justify his positioning if BIP aren't going over his head very often (into an area where another CF would have caught them).

Is Bret Boone's UZR much higher than his ZR because he positions himself in a manner to get a different subset of balls? Ones that have higher value in UZR than in ZR?

I'll have to site-search to find my other questions for this.

Chris, all good points. As I said, one of the problems with the STATS data is that they only give the location in the OF where the ball is fielded, on hits thru the IF. That is crazy. I need to talk to them about that.

Other than that, some day when I have the time, I will try and refine UZR to incorporate your and others concerns. Right now, it is what it is.

Thanks for the insight and suggestions...

mgl,

Thanks for the UZR updates, it's much appreciated. I had a feeling Tejada would be an easy #1 when you combined lwts and UZR, but it's nice to see it confirmed.

I was also (pleasantly) surprised to see Mora not make the bottom 4 for AL 3B. I remember seeing an early season UZR that pegged him as by far the worst 3B. He looked much better for the last 4 months of the year, so he must've been above average once he got used to the position. Do you do splits for UZR?

Mora ended up with a respectible -4 per 150. I don't do "splits," but I could as the data I use is broken up into games of course.

Mora is interesting. He played mostly outfield and SS before this year, and a little 2B. His prior UZR numbers were stellar in the OF and around average in the IF. He should by all rights be above average at 3B, using "positional translations." So next year, we should not be surprised if he posts a plus UZR at third. Who knows though?

A player who now hits as well as he does and can play so many positions, is quite the valuable player! He is probably one of the more underrated players in baseball - completely under the radar...

DKDC,

I am writing an article for THT about the MVP award (how including defense and other peripherals, as well as using positional adjustments, can and should change the landscape of the MVP voting)...

mgl,
see that article here

mgl,
see that article here

Chris, I don't know anything about your CPI metric, but it seems like that ranking would be more valuable if it included full-season stats. For example, I have a hard time believing that Sheffield was 50% more valuable on offense than Tejada.

You said in the article that you expected to reach results similar to MGL's, yet many of the numbers don't match up. Do you think CPI is better?

Chris, I am reading the article now. Will comment later. Also, I am not going to write the one for THT...