MGL's numbers are park adjusted, of course, but the consensus among Dial and Chone is that Manny's more in the -10 to -20 range. Dial actually says that he thinks Manny's average, but I really doubt that.

I think it's safe to say that whatever defensive system the Red Sox are using, it doesn't show Manny to be a -30 outfielder (and therefore on par with Reed Johnsons of the world). If it did, they would have traded him already, just to be rid of his contract.

Interesting to hear that MGL's St. Louis work is spoken of in the past tense now. Does that mean we get to see UZR published here again? Please please please!

It's a good thing the Red Sox play in the AL so Manny can DH...oh wait.

Purchase Landsdowne St. from the city of Boston and move the wall back 30 feet.

Huh? Giving Manny more room to cover helps his defense?

Yeah, I'd think they'd want to move the wall in to protect him. To make up for it, just add another 30 feet of height. Imagine a 70 ft wall 250 ft from home plate. Would balls off the monster still be mostly doubles? How many HRs would be lost?

Huh? Giving Manny more room to cover helps his defense?

aren't wall balls are considered "catchable" in uzr? so putting the wall back would make those catchable balls actually catchable.

if this is true, why hasn't someone done a defensive study that subtracts out all the wall balls and considers them home runs (just for the sake of seeing what his defense would look like for actually catchable balls)?

Piehole,

Dial has done just that in in this ST thread. He finds that Manny is -12 runs /150 games, but suspects he'd be even better.

Two problems with this:

1. Dial completely eliminates all of those balls that hit the wall while leaving in all of the other Fenway data. But if it was a normal wall at a normal distance, many of those balls would be in play, and a good outfielder would get them. Therefore, in Dial's study, the Monster is allowed to hide any deficiency that Manny might have going back on the ball.

2. The presence of the shallow, tall wall allows LFs in Fenway to play shallower and get to more balls in front of them, a point brought up in the thread I mentioned above by TH. Dial's study doesn't account for this either.

Overall, I'm guessing that adjustments based on the Monster bring Manny back from -30 to -40 to something in the -10 to -25 range.

(Also, note that MGL's defensive numbers are park adjusted as well and show Manny to be -32.)

This is the NEW YORK Times. They're just jealous....

I saw the thread, Kevin, and it assumes that Manny would get to balls hit, say, 330 feet, if he had more room to field them. This may be true, but I would think if he had 30 more feet behind him, he'd play back further, and he wouldn't get to some balls hit in front of him that he might had he been playing 30 feet closer. So the question of area seems to be a wash.

I didn't mention park adjustments because they are complicated and didn't affect the argument in the story. The deal is basically the following:

1. Yes, MGL told me he stopped working for the Cards in '06.
2. UZR's park adjustment is just a percentage applied to balls hit to a zone (eg, if a neutral-park LF is expected to catch 150 balls, a Fenway LF will only be expected to catch 125). His park adjustment definitely does not pick up year-to-year variation in number of balls hit off the wall, although it does account for the average # of balls hit off the wall per year.
3. MGL says most of Ramirez's poor fielding has been done on the road. John Dewan agrees.
4. David Pinto does account for park on every individual ball, and he had Manny at -20 plays last year. According to Dewan, most of the balls Manny missed went for doubles, suggesting a run value of around 0.95 runs per missed ball.

I watched virtualy every game Manny played last year. He struggled to run, it appeared, but has good hands, and a quick and accurate arm, although not a strong one, maybe average. My eyes tell me he's a pretty bad LFer, as most of the job is being able to get to the ball and therefore catch it.

Also, is he being measured against an AL LF 2006 average, or some other benchmark?

Yes, AL LF 2006 average.

I've seen Manny make too many plus catches, and too many good throws, to anoint him THE uber-suck LF by which all crappy defenders should be judged.

MGL's astronomically poor figures for 05 and 06 may well be due to an above-average number of balls being hit off the wall. MGL's prediction for Manny's D next year is -21 (one end of the range I specified in the piece).

Why doesn't the metric eliminate routine flyballs? These guys are pros...they don't miss too many "routine" flyballs to be statistically relevant, I would think. Why not focus on the tough chances?

Because the routine fly balls that you DO miss cost your team loads. If you catch a ball that you were 95% likely to catch, you'll only be credited with 0.05 marginal plays saved. But if you drop it, you'll be docked for 0.95 marginal plays cost.

How many routine fly balls are dropped in a season, all fielders considered? 1 a week?

As I understand it, Dial doesn't give him credit as if he would have caught balls that hit the wall, he just eliminates them from discussion. This seems to me the correct thing to do. I believe he further thinks that differences in LD% vs FB% off the wall will affect how many runs the other metrics say Manny costs. I agree with Kevin that he also gets an advantage from playing shallow. The general conclusion that he is far from a horrendous LF stands I think. Furthermore, Fenway LF is presumably a good place for him to play since even if he would cost 20 runs playing 150 games in an average park, the reduction in chances at Fenway means a below average fielder will cost fewer runs. I wouldn't be surprised if Sox management is viewing this similarly. His contract was an albatross when it had many years on it (and therefore more uncertainty) and the market was quite different than it is now.

As I understand it, Dial doesn't give him credit as if he would have caught balls that hit the wall, he just eliminates them from discussion. This seems to me the correct thing to do. I believe he further thinks that differences in LD% vs FB% off the wall will affect how many runs the other metrics say Manny costs. I agree with Darren that he also gets an advantage from playing shallow. The general conclusion that he is far from a horrendous LF stands I think. Furthermore, Fenway LF is presumably a good place for him to play since even if he would cost 20 runs playing 150 games in an average park, the reduction in chances at Fenway means a below average fielder will cost fewer runs. I wouldn't be surprised if Sox management is viewing this similarly. His contract was an albatross when it had many years on it (and therefore more uncertainty) and the market was quite different than it is now.

Sorry about the double post, thought I stopped it so I could correctly identify Darren.

He easily makes 3 times that many acrobatic catches.

Included among them diving stops of relay throws from the centerfielder.

I want to know what Manny kevin is watching. He sounds like a great fielder.

dave h, thanks for crediting me but the observation that a LF in plays shallow and gets to more balls was made by TH--I was only relaying that from the other thread. Your double post was for naught! :)

As I understand it, Dial doesn't give him credit as if he would have caught balls that hit the wall, he just eliminates them from discussion.

Yes, that's an important distinction. If Dial were crediting Manny with getting all those balls, he'd be inflating his value greatly. By simply throwing the data out entirely, I think it's giving Manny a small boost. This is because in a normal park an average fielder would get to some small percentage of those and (IMHO) Manny would get to less of them. I base this on my observation of Manny and how a lot of balls seem to fall behind him, even in the small Fenway LF. But even that's debatable.

BTW, Dan R, I didn't realize you were a primate and I sent you an email complimenting the article. Really nice piece of work.

Why doesn't the metric eliminate routine flyballs? These guys are pros...they don't miss too many "routine" flyballs to be statistically relevant, I would think. Why not focus on the tough chances?

Because positioning and range would make this a very dubious classification. If one OF plays 10 ft to the left of where another plays, different flyballs are "routine" for them. If defensive metrics did this, then suddenly, Buster Olney's critiques of defensive stats would have merit.

I want you to rationalize how Manny's fielding could account of -40 runs.

Think about that, Darren.

I don't think anyone, at least anyone in thread, really think Manny is -40 bad. But I think Darren (and for that matter, me) is questioning Manny's 15 aerobatic catches that he apparently makes in the Red Sox games you watch that no one else does

I'd still play Ortiz at first ~20 times in road games this season, doing at the parks with the
biggest LFs (Metrodome, Yankee, Comerica).

By "making an acrobatic catch", I assume kevin means "tripping on a sprinkler and making a salto mortale"...

how Manny's fielding could account of -40 runs

This has always seemed like nonsense to me, too, but it simply entails an assertion that for several years now the Red Sox pitchers have been a lot better than they've looked. Maybe that's true. Maybe Josh Beckett's ERA should have been 4.50 last year, and it was 5.01 entirely because of Manny.

The conceptual issue comes back to the baseline. What do you compare Manny to? Replacement level? but there are hundreds of better outfielders in the minors. Having no left fielder? that seems insane, but Manny caught 175 fly balls last year, so clearly he contributed something to the cause. Hitters in his class? because Jim Thome or Frank Thomas or David Ortiz sure wouldn't catch 175 fly balls in left.

Or league average, of course, but even then there is always some doubt as to whether theoretical runs allowed by a single defender add directly into a team's runs-allowed column. Short of Jose Canseco letting a ball bounce off his head for a home run, no defender allows a run single-handedly. That's why managers are always tempted to "hide" great hitters somewhere. Yes, he's terrible, but please God nobody will hit the ball to left today ... whereas you cannot "hide" a pitcher or a hitter.

1. Dial completely eliminates all of those balls that hit the wall while leaving in all of the other Fenway data. But if it was a normal wall at a normal distance, many of those balls would be in play, and a good outfielder would get them. Therefore, in Dial's study, the Monster is allowed to hide any deficiency that Manny might have going back on the ball.

2. The presence of the shallow, tall wall allows LFs in Fenway to play shallower and get to more balls in front of them, a point brought up in the thread I mentioned above by TH. Dial's study doesn't account for this either.

These two points are, basically, the same point. You are arguing that Fenway's LF is in fact easier to play than the average LF, because hte fielder has to cover less ground. Talking about balls over the head and playing shallow sort of obscures the point, I think. The question is whether an otherwise average LF would have above-average fielding stats in Fenway (if those stats were computed correctly).

I'm quite skeptical that the effect here is particularly large. Is Fenway's LF really easier to play by a factor of 5-10 runs per year? I don't see it. I'd be interested to see statistics on this effect, but it's a bit hard to see for me.

Dial, Arthur, and Chone have been pretty clear that by the numbers, Manny has been about a -10 to -15 fielder for the Red Sox. This should be the starting point for future stat-oriented discussion of his defense.

Incidentally, I don't know how the hell I managed to come up with "aerobatic" rather than "acrobatic" but I actually kind of like it

You were thinking of your typical brain activity when you wrote it.

Ah, the insult by non sequitur. A devastating weapon in any debate arsenal

These two points are, basically, the same point. You are arguing that Fenway's LF is in fact easier to play than the average LF, because hte fielder has to cover less ground.

I guess in the way they are the same point. But two separate things are happening. Thing 1: Manny (and other Fenway LF) can play shallow, knowing that balls hit over their heads are not going to turn into triples and inside the park HR. This allows Manny (and co) to catch more shallow balls than he normally would.

Thing 2: Dial is wiping out the 30 balls that hit the wall as unplayable. I trust his judgment that these are unplayable balls for Manny or anyone else. But in another park, those balls are all in play. Manny gets to less in-play balls than other LFs, so he would likely get to less of those as well. If you're trying to determine how many real runs Manny is costing the Red Sox, this is irrelevant. If you're trying to determine Manny's skill as a LF, it's part of the equation.

I'm quite skeptical that the effect here is particularly large.

I agree it's not large. I'd be surprised if it made up for 1/2 of the 30 plays which have been recredited to Manny by Dial's observation.

Is Fenway's LF really easier to play by a factor of 5-10 runs per year? I don't see it. I'd be interested to see statistics on this effect, but it's a bit hard to see for me.

I don't have the stats or observation to prove this, but I would not be at all surprised if it is in that range. Consider, though, that if we take Dial's latest work at face value, Dante Bichette and Troy O'Leary become great, great, LFers.

Dial, Arthur, and Chone have been pretty clear that by the numbers, Manny has been about a -10 to -15 fielder for the Red Sox. This should be the starting point for future stat-oriented discussion of his defense.

I don't think Dial's work here is complete. He uses observation to adjust for one quirk of the park, but has not accounted for the two issues I mention above. He also compares Manny playing in Fenway with special observational adjustments to other players playing outside Fenway with no such adjustments.

Chone's numbers are in this ST thread, and after adjusting for park, he got -20/400 chances in 05, -35/400 chances in 06. (Incidentally, using the same park adjustment on Dial's numbers, I got -23 /150g in 05, -32/150 in 06.)

Arthur's work is based on one injury-shortened year of data. It's far from complete.

One conclusion that you left out was MGL's, which put Manny as -32 in 06 and worse in 05 IIRC.

I'd hardly say that consensus is -10 to -15.

To summarize, here's what we have from each for 06 for runs cost per full season of play:

Arthur: -12 (using an average of his and AROM's park adjustment)
Dial: -12
Chone: -35
MGL: -32

Even if we take Arthur's and Dial's numbers at face value (which I don't), the average is -23.

I'm not at all sure that I've got those numbers right for Arthur. I think that -12 may be using ONLY Arthur's numbers, not averaging in Chone's.

Kevin,

When are you going to learn to view wild-ass guesses with some skepticism? When are you going to learn anything for that matter?

Per Dial, a play in LF is worth about .831 runs, so if Manny were 30 runs below average per 150 games, he'd have to not make about 36 plays that other LF make over a 150 game period. That means that about once every 5 games, Manny would miss a play that an average fielder would make. I find that quite believable. Of course you don't because you're seeing him make all those acrobatic plays.

Having seen Ramirez quite a bit, intuitively, -10 runs--a game a year--seems right to me. He isn't good, but I don't think he is killing the Red Sox out there, relative to what he does at the plate.

It also seems to me that play by play analysis of this would need to be situational, although it would be hell to do. A ball Ramirez fails to get to with the bases loaded in the 7th in a 4-4 game is different than one he fails to get to in a 12-2 game with the bases empty in the 9th. Such a distinction might not be related to skill in isolation but is related to runs.

Such a distinction might not be related to skill in isolation but is related to runs.

But the skill in isolation is far more important, IMHO. It tells what Manny alone contributed to the play, as opposed to what the hitters and baserunners on the opposing team, the pitcher, and other fielders, contributed to make the play meaningful. This tells us more about Manny's contribution looking backward and more about what he's likely to contribute going forward.

All that said, a fielder's positioning and tactics (you dive for a ball in a 4-4 game, you don't in 12-2 game) are probably affected considerably by game situations.

All that said, a fielder's positioning and tactics (you dive for a ball in a 4-4 game, you don't in 12-2 game) are probably affected considerably by game situations.

Agreed, and I think that is a "skill" as well.

But the skill in isolation is far more important, IMHO.

You may be right; if so, this is why I prefer the Dewan plays metric as opposed to a runs metric.

That's just it. It isn't a wild ass guess. Dial reviewed the game tapes. Mgl didn't.

1. You didn't ask me to give Dial's numbers a chance. You asked me to view statistics skeptically. In order to view stats skeptically, you have to insert your own observation. That's where the wild ass guessing comes in.

2. The reason you have been skeptical of MGL's work in the past is that it didn't fit with your WAG about certain players' defense.

The only numbers which eliminate actual numbers of balls off hte wall are Dial's and Arthur's. Chone and MGL use a generalized adjustment. I see no reason to use statistics which do not eliminate from the start actual doubles off the monster.

I grant that Dial's and Arthur's numbers may need adjustments, but we should be adjusting those numbers rather than averaging them with statistics built ultimately off a data set so flawed as to be nearly unusable.

Ramirez does make acrobatic plays, but they're of the lunging-catch-after-taking-a-circuitous-route variety. Kevin's apparently of the Web Gem School of defensive evaluation.

The only numbers which eliminate actual numbers of balls off hte wall are Dial's and Arthur's.
Okay, but don't lump Chone in with them. And learn how to spell "the." At least use the standard "teh."

I grant that Dial's and Arthur's numbers may need adjustments, but we should be adjusting those numbers rather than averaging them with statistics built ultimately off a data set so flawed as to be nearly unusable.

Chone's and MGL's park adjusted numbers based on years of data are nearly unusable, but Arthur's 1 season of data of 1 player is a gold standard that can only be tweaked at the margins?

kevin--

have you ever disagreed with someone without calling them a complete idiot?

kevin--

have you ever disagreed with someone without calling them a complete idiot?


I can answer that one, its rhetorical quality aside. He has disagreed with me and has done so in a respectful way. Plus, has he actually called you a name in this thread?

shut up robin, you're an idiot.

Fenway differs from other parks in that it has the Wall, sure. And it's good to look carefully at that. But Fenway also differs from other parks in that it has essentially zero foul territory in left field. That's a factor here, too. There are a lot of balls hit foul down the line that are catchable in other parks but not catchable at Fenway.

Anyway, I suspect that Manny is a pretty terrible fielder in the abstract, possibly as bad as the -30ish numbers would suggest, but Fenway hides a good chunk of his terribleness, in at least three ways:

1. Of the balls that are unplayable by anyone at Fenway because of the Monster, some of those would be caught by a good outfielder in a more typical park, but not by Manny in that park.

2. Of the balls that are unplayable by anyone at Fenway because of the seats along the left field foul line, some of those would be caught by a good outfielder in a more typical park, but not by Manny in that park.

3. Manny's arm is quick and accurate but not strong. At Fenway he can play shallower than in a more typical park, and does not need to make as many long throws as he would have to make in that park.

I think it could very well be that Manny is a -30 fielder in the abstract but only -15 or so when he gets to play half his games at Fenway.

shut up robin, you're an idiot

.

Well, that is probably something most Red Sox fans here can agree on. After all, I root for the ####### Cincinnati Reds and also dislike Larry Lucchino, although I think he is a smart man.

I enjoyed the "Jane you ignorant slut" point/counterpoint of posts #43/44.

Still funny after all these years.

I do. You were thinking of your typical brain activity when you wrote it.

JFCOAPS, kevin, even when I agree with just about everything you're saying in a given thread*, you make me wish I didn't because you're so often obnoxious.

* Maybe not *15* acrobatic catches, but he does make some fine plays.

Chone's and MGL's park adjusted numbers based on years of data are nearly unusable, but Arthur's 1 season of data of 1 player is a gold standard that can only be tweaked at the margins?

I'm open to arguments for tweaks that go beyond the margins, but yes, that's exactly what I think. Unless we remove doubles off the wall from the data set, we're just not going to be able to get good information.

Certainly, if someone applies Arthur's method to pre-2004, and finds that Manny was total crap then, that'll be useful information. But previous to such study, all we've got that's particularly useful are the numbers that only include balls that a human being can catch.

But it's one season of data. It's practically useless. Similarly, Dial's new -13 number is one season of data. The two are even from the same season, so it's not like they complement each other. At this point they are completely useless.

Just because they are based on first person observation doesn't mean that they account for balls off the monster better. For example, I could watch one game and see 3 balls off the monster and draw conclusions from that, but it would be completely useless.

I'd like to clarify one thing - I haven't seen all the video, for some reason, some games at mlb don't come up. But the scoresheet are good and indiate ball walls for FBs as well. I've seen alot of the video.

Also, I don't think Many *is* average; but that our data (prior to me looking at the bip) indicated he *could be*.

Joe Arthur's road data indicates Manny may not struggle so much with a bigger park. His play in Cleveland wasn't bad.

There are other *smaller* issues with the work I have done. there's appears to be a shift in scoring BIP of the wall as LDs or FBs.

There is more work to be done than I have done. The 2004 eason (and others) need to be reviewed.

From Retrosheet 2003-2005 data. Manny at Fenway, 439 fly balls, 283 outs, 64.4% conversion rate. All visiting left fielders at Fenway, 623 fly balls, 417 outs, 66.9% conversion rate. Line drives: Manny 214, 26 outs, 12.1% rate; visitors 386, 40 outs, 12.1% rate. That means the average visiting left fielder would have caught 11 more of Manny's flyballs but 4 fewer line drives. Because Retrosheet assigns a hit ball according to the fielder that picks the ball up, this raw data needs to be adjusted for the balls that the center fielder hustled and got to while Manny was, uh, "conserving his energy". Even after adjusting for this, the total for Manny is only -18 PLAYS not made over THREE YEARS for both home and away. Because Manny has a great arm, the actual runs that Manny is costing the Red Sox over having an average fielder in left is only 3.3 runs per year. I have capitalized PLAYS and THREE YEARS because the way other metrics present fielding data it often leads to confusion between runs and plays and three year totals instead of single year numbers.

Manny is not a great outfielder, but his play in Fenway is only slighly worse than average and surprisingly his play on the road is not much worse. Although the fans (and some of his pitchers) will be angry that he doesn't hustle on every play, he does miss very little time during the season. The Red sox would be crazy to even think of trading him.

I fully admit I don't have a specific metric other than my eyes and the number of Red Sox games I watch.

Manny's biggest problem is when he makes a bad play, it's often a spectacularly bad play on a ball a decent Pony League player would routinely make. Of course, a play like that makes the lowlights about 300 times on ESPN and is remembered. He doesn't do it often, but he does this more often than any other player in the majors. He seems to have these "Manny Moments" (aka "Brain Farts") just often enough to get a rep as a horrible fielder.

I'd wager, if the perfect fielding metric could be formulated (and I haven't seen it yet), #66 would have it about right: A little less than average at home, and less than that on the road. Is he the worst LF in baseball? Probably, but he's not THAT much worse than whoever is directly ahead of him.

This article wouldn't have been written if the NY Times weren't owned by the Boston Herald, Jack Nicholson, and the Jimmy Fund. Can you believe how the time is their own personal PR machine?

Please set sarcasm meters.

From Retrosheet 2003-2005 data. Manny at Fenway, 439 fly balls, 283 outs, 64.4% conversion rate. All visiting left fielders at Fenway, 623 fly balls, 417 outs, 66.9% conversion rate. Line drives: Manny 214, 26 outs, 12.1% rate; visitors 386, 40 outs, 12.1% rate. That means the average visiting left fielder would have caught 11 more of Manny's flyballs but 4 fewer line drives. Because Retrosheet assigns a hit ball according to the fielder that picks the ball up, this raw data needs to be adjusted for the balls that the center fielder hustled and got to while Manny was, uh, "conserving his energy". Even after adjusting for this, the total for Manny is only -18 PLAYS not made over THREE YEARS for both home and away. Because Manny has a great arm, the actual runs that Manny is costing the Red Sox over having an average fielder in left is only 3.3 runs per year.

Those are some odd numbers and conclusions.

A. Home fielders are expected to catch more balls than visiting ones, especially LF in Fenway. So you can't compare Manny's numbers to visiting teams' numbers unless you make an adjustment for HFA.

B. You give the same percentage for LD's and then say that Manny caught 4 LD more than he should have. ??

C. You say that Manny has a "great arm," give no data to support that assertion and then credit him 2.7 runs per year in "arm" (I think). While 2.7 is not inconsistent with a "great arm," Manny has in fact contributed 1.5 runs a year in "arm" over the last 4 years, using my methodology for computing "arm runs" which is essentially baserunner advances/outs per opportunity as compared to a league average LF'er and park adjusted as well (which is necessary for Fenway LF'ers of course). Since the 1.5 is based on less than perfect data and methodologies, I suppose it could be 2.7, although I have never heard before that Manny has a "great" arm (a very good one maybe).

If someone is not comfortable with a park adjustment for a player, especially one with many years of data, simply use his non-Fenway (or whatever park you are not comfortable with) numbers (road, and while playing for CLE). That should still give you a pretty good idea of a player's fielding value/talent. However, if nothing else, you can certainly infer SOMETHING more (using a larger sample) about a player's fielding by comparing his stats in a quirky park (like Fenway) to other fielders at that same park, again, adjusting for HFA.

When all is said and done, I think you will find that Manny appears to be (fron the data and from a consensus on scouting/observation) a poor LF'er. Whether his true talent is -10 or -20, who knows. After crunching my data in my imperfect way, it appears to be closer to -20 (-21 is my best estimate going into this year), but as always, that could be wrong. All estimates drawn from sample data have an uncertainty level and have an X percent chance of being "wrong," where X varies with the magnitude of "wrongness." Also keep in mind that on average we see a rather steep decline in fielding in the OF with age. When looking at Manny's historical data (his UZR's were good while in CLE) and making future projections, keep that in mind. For a typical OF'er, from age 25 to 35, you are probably looking at a 15-20 run decline. In order for Manny to be around average now, he would have had to be a superstar defender early in his career. Given his speed and body type, that is close to impossible. Again, this is all given a typical aging pattern.

Using scouting reports and observation (video, etc.) can reduce the uncertainty level of a "conclusion" (best estimate) drawn from sample data, and give us a more accurate number, if done and used properly (the observation and scouting reports) of course. I use a "lazy approach" (to quote some other poster) for my final defensive numbers for several reasons. One, it is not particularly important (to me) how accurate I am. Two, I don't generate defensive numbers for a living. Never did. This relates to one. Three, and most importantly, I have 10,000 other things that I do in my baseball research. I would have to put some or all of them aside if I took the time to use a "non-lazy approach." I have to balance streamlining all of my methodologies with an effort to use what I consider a quality one (methodology).

Finally, this thread is about half interesting and useful and about half assinine. The assinine part is why I rarely post here anymore.

Actually I think you (misterdirt) are crediting Manny with 2 runs a year in "arm" which is probably about right.

Sooo....

How about posting new UZR numbers?

Who would want crappy numbers using a "lazy man's" methodology? Seriously, I'll talk to Tango about putting them up on his site.

Finally, this thread is about half interesting and useful and about half assinine. The assinine part is why I rarely post here anymore.

And Primer is worse off for it. Certain posters' biases aside, your contributions are always a net positive here.

assinine should be asinine of course, lest I be accused of using a "lazy man's" spelling methodology...

But it's one season of data. It's practically useless. Similarly, Dial's new -13 number is one season of data. The two are even from the same season, so it's not like they complement each other. At this point they are completely useless.

Well, then I'd say that all we have is completely useless information.

If Stats and BIS scorers haven't been marking when a ball hits 20 feet off the wall, so as not to count it against the Fenway left fielder, then their data for Fenway's left field is basically unsalvageable. (I'm following your use of terms here - if one year of good fielding data is "completely useless", then the Stats/BIS numbers are at least equally useless.)

I'm fine with saying that Manny may well be worse than -13 becuase only one year of data has been compiled. That's totally fair. But I think that mixing good data with bad data is not going to add value to the discussion.

(Should be noted, since mgl's here, that I also would love to the see the UZR numbers, and I'm excited to see them published. Thanks for your work! My stance here is wholly separable from my evaluation of UZR - the Stats data was compiled wrong for Fenway's LF, that's got nothing to do with MGL or his analysis of said data.)

You give the same percentage for LD's and then say that Manny caught 4 LD more than he should have. ??

I should proofread my posts better. The conversion rate (of LDs to outs) for visiting fielders is 10.4%. The +4 plays for Manny is still correct.

Great arm is too strong for Manny. I stand corrected. It is a better than average arm.

Home fielders are expected to catch more balls than visiting ones, especially LF in Fenway.

I am not sure why logically a home fielder should catch more balls than a visiting fielder and I have not seen any data that says that they do. Are they sopposed to be faster at home? Have better reaction times? See the ball better? I do agree that an outfielder should play the ball off the wall better, and Manny does.

You don't give any details about how you park adjust outfielders plays so it is hard to compare but adjusting for only the BIP hit to that particular field would seem to me to be more accurate than any generic adjustment.

For a typical OF'er, from age 25 to 35, you are probably looking at a 15-20 run decline. In order for Manny to be around average now, he would have had to be a superstar defender early in his career. Given his speed and body type, that is close to impossible. Again, this is all given a typical aging pattern.

A fast outfielder will decline a lot as he ages because his speed will decline a lot as his body changes. An outfielder like Manny, who didn't start out as a fast outfielder, will not decline as much since the other skills that make a good outfielder do not decline as much with age until the eyesight and reaction times begin to fail, which happens later. Good eyesight and reaction time are also necessary for good hitting so when Manny starts to get really bad as a fielder he will have lost most of his value as a hitter as well.

I am not sure why logically a home fielder should catch more balls than a visiting fielder

Logically, I can think of several reasons:

Better knowledge of the effect of wind conditions, park angles, nearness of wall and other impediments.

Better knowledge of how the playing surface affects ball movement, bounce, etc.

More experience seeing the ball come off the background, and thus being better able to judge flight.

Better positioning and spacing among outfielders as a result of better knowledge of conditions.


This is all in addition to the overall better physical condition many players feel from not being on the road, sleeping in their own bed, having the at-home routine, etc.

Just because they are based on first person observation doesn't mean that they account for balls off the monster better. For example, I could watch one game and see 3 balls off the monster and draw conclusions from that, but it would be completely useless.

Um, I have also done 2005, so it is at least two years of data for Manny.

Better knowledge of how the playing surface affects ball movement, bounce, etc.

We were talking about outfielders and catching the ball in the air for an out. I still was. I guess I should have made that clear.

There is not much difference in background for outfielders that you can get used to. It's people and sky; pretty much the same everywhere but differing from day to day. Lighting is a possibility but is probably pretty good everywhere at the major league level. Dome background might be something that a player could adjust to and gain an advantage but obviously wouldn't apply to Fenway.

Wind is a problem for all outfielders when it is blowing. Everybody looks at the flags. There might be some advantage for the home outfielder where the park is unusually windy with strange currents like Candlestick was but I would like to see numbers that prove it.

So you are pretty much left with a better nights sleep as an advantage.

I believe in a home field advantage for pitchers and batters. But I think the majority of the advantage will be in home runs, walks, strike outs, and the distribution of hit ball types rather than the conversion rate for any specific hit ball type. But I haven't studied it, so I can be convinced by some numbers if someone has studied it.

there are plays where it is obvious that Manny takes the safe or conservative approach to the ball and then executes the safe relay---a question may be--what is the price of a sure thing,baseout states included--#1 should of,#2 could of ,#3 didn't,#3 #2 blur to the pop mythology of Manny--and might miss what is really going on--

We were talking about outfielders and catching the ball in the air for an out.

Cutting off balls that are hit on the ground can be just as important. Knowing the bounce off the turf can be a pretty big advantage for guys playing in the metrodome.

Wind is a problem for all outfielders when it is blowing. Everybody looks at the flags. There might be some advantage for the home outfielder where the park is unusually windy with strange currents like Candlestick was but I would like to see numbers that prove it.

I don't agree. Winds swirl. The hit corners, they change directions. They swirl at Shea. They do different things in different parts of stadiums, depending on how much they are enclosed, etc. Just looking at a flag is not going to give you the kind of information that playing in park regularly does.

There is not much difference in background for outfielders that you can get used to. It's people and sky

It is not just people where the ball meets the bat. There are different backgrounds everywhere, some signage, some stands close up, some far back. Picking up the ball coming off the bat is a big deal for outfielders. I have heard outfielders say it is easier in some parks than in others.

In addition, there is the configuration of the park, and how used to it you are. Some guys can get very cautious about approaching walls, juts and angles that they are not used to.

From 03 to 06, using STATS data, looking at all fair, non-HR FB, LD, and PF greater than 150 feet from HP, the home fielders catch 60.52% and the visiting fielders catch 60.59%. So you are right about that. For left field only (slices C-K, where A and B and Y and Z are foul and the whole field is A-Z), the numbers are 57.8 at home and 57.49 on road, the difference not being statistically significant. For left field at Fenway, it is 47.02 at home versus 47.78 on the road, not close to a statistically significant difference.

Aging curves in LF and RF:

From age 24 to age 30, players (min 50 games per year) lose an average of 11 runs per 150 games in UZR. That is around 1.8 runs a year.

From 30 to 40, it is another 19.1 runs, or 1.9 runs a year.

This is using the "delta method" of course.

Interestingly the average UZR for all players at any age is around zero (slighly better at younger ages, 23-27), which suggests that only good defensive players (at the corners at least) stay at their position for very long.

Now let's break that down into fast and slow players. I used park adjusted triples rate for that.

For fast players only:

Overall, they lost 2.3 in UZR per 150 games per year. From 25 to 30, they lost 11.5 runs in 5 years, or 2.3 per year. From 30 to 39, it was 3 runs per year. So from 24 to 39, it was 2.75 runs per year.

The slow ones:

Overall loss per year: 1.57 runs per 150 games. There are no slow players less than 27 in my group. From 27 to 32, they lost 1.9 runs a year. From 32 to 40, they lost 2.3 a year.

So indeed it appears that fast player lose defensive skills more rapidly than slow players, but the difference is not all that great.

Looking at the very slowest players, to which Manny belongs, the average yearly decline is 2.3 runs. For the very fastest players, it is only 1.3 per year.

If I use BMI (body mass index) rather than speed, I get an average yearly decline of 1.58 for those with a BMI less than 27. For more than 26, it is 1.43. For the real low BMI (25-) players, we get almost no yearly decline, .195. For the high BMI guys (29+), we get 1.52 per year decline.

I am not sure that we have much evidence of a clear relationship between speed and defensive decline in LF and RF.

Regular HFA is mostly manifested in walks, K's, errors, and triples. I would have thought it would be significant among fielders, but apparently not, at least as far as catching air balls.

The biggest "evidence" that Manny is a terrible left fielder is simply that there is virtually no such thing as a 35 yo average or better LF'er who was not a star defender when young.

I am not sure why logically a home fielder should catch more balls than a visiting fielder

In addition to the reasons listed by Srul, the average home team is going to have to record around 120+more outs (that's strictly a guess, based on a slightly above-average home record minus walkoff wins) than the visiting team. It may not affect UZR rates, but it will impact total putouts.

Thanks for doing the work to study the question of fielding HFA, MGL. Its always better to have the numbers than guessing. The aging numbers were interesting too and seemed entirely reasonable.

Those are "quick and dirty" numbers I am generating, so take them with a grain of salt...

The biggest "evidence" that Manny is a terrible left fielder is simply that there is virtually no such thing as a 35 yo average or better LF'er who was not a star defender when young.

Well he certainly was not a star defender in his younger days, but from your above post:

When looking at Manny's historical data (his UZR's were good while in CLE) and making future projections,

I'm not sure what years you're refferring to, but let's say Manny was a good UZR RF in his last two years in Cleveland (1999 and 2000) at ages 27 and 28.

You have slow players losing 1.57 runs per year (I think). Let's just say it's -2 per year. So during the course of Manny's time in Boston we might expect him to lose about 12 runs. and yet he has acctually gone from a good UZR RF (what's good +5? +7?) to an historically bad LF to the tune of a -32 in terms of performance and a -20 in terms of true talent.

Do you have a UZR difference in the move from RF to LF? I don't know. At least traditionally he has moved down the defensive spectrum from a harder position to an easier one as he's aged which might offset some of his age related decline. But at least in terms of UZR you have his performance absolutely cratering. Where we might expect an average slow player to lose 12 runs or so you have Manny actually losing close to 40 runs (that's the -32 plus whatever positive UZR he had in Cle).

It seems to me that most of what you've posted in this thread is more suggestive of the -10-15 range than what the UZR methodology is actually spitting out.

The data I have (which may be the old UZR) has Manny at -22 /150 in 2000. However, he was +9 in 01.

Philly, you are trying to make way too much "sense" of noisy data then is possible. All we can do is look at a player's historical record and make inferences about his true talent/future performance using models which are based on lots of real data and "work" (conform to the data) That is what I do. Based on his historical UZR (batted ball) data I have, whether 3 years or 10 years, using a basic "Marcel" model adjusted for a typical aging pattern, I have Manny at around -20 runs per 150 games. Based on observation/concensus scouting and his age, that does not appear to me to be unreasonable. If it is really (and no one will ever know) -15, so be it - that would not surprise me. If it were really -10, that would not shock me. It is what it is based on the data and models we have. Most of the talk in this thread about what Manny really is or is not is nonsense. Playing in a home park like Fenway makes the process a little more difficult and the results a little more innaccurate, but not THAT much. If we look back at the late 90's we would expect to see any fielder who is now -20 to be close to zero, based upon the aging patterns that I think work (around 2 runs a year). In fact, that is what we see. If we see a +5 back then, so what? Maybe he was really worse than that back then. Maybe he is better than a -20 now. Who knows.

As I see it, mgl used a quick and dirty approach to adjust for home field. Since the data he was using counted wall balls as in zone, Dial went back and removed those from the data initially. It seems clear to me that this is better, and it's ridiculous that STATS or whoever didn't do this themselves. If mgl's quick and dirty approach gave the same result as Dial's, that's great. But it didn't - and therefore in this case I think we need to pay much more attention to the thorough method. At least we've gotten away from the -50 runs nonsense even just with mgl's calculations.

I'd also point out that as far as the Sox are concerned, what's important is how many runs Manny costs playing 1/2 his games in Fenway, not how many he would cost given average chances in an average park. If he's a below average fielder then it's to their advantage to be able to play him in Fenway's left field where he has fewer plays to make.

If you read the original article in the NYT, you will see that at least half of Manny's damage is on the road, so regardless of how rigorous you are with the data at Fenway, he is still going to be an horrendously poor fielder.

As far as the STATS data, they give the "slice" and the distance. In whatever slices are the "wall," (C-G or whatever) beyond a certain distance, no balls will be caught. It would be nice if they told you when any ball was "off the outfield wall," but it is not really necessary and can easily be inferred from the data at Fenway. I am not really sure what the argument is at Fenway. I can easily remove all balls in a certain slice and beyond a certain distance at Fenway and get a more accurate number for Fenway. I am reasonably certain that it won't change things a whole lot, and to tell you the truth, it is not worth the trouble. I really don't care whether Manny is a -17 or a -22.

Keep in mind that park adjustments for defense in quirky parks are not that easy to do. Somtimes you might be better off just using road data. Even if you remove all "wall balls" from the data at Fenway whether you know from video that they were off the wall or whether you infer it from the location in the data, you still need some serious adjusting for all the other zones in LF at Fenway. Obviously a fielder plays closer and will make more plays on short fly balls than in other parks and might make fewer plays on line drives and "low" files near the base of the wall, because they cannot run full speed to the wall. Making those adjustments in those other zones is not that easy or workable for the following reasons. If you do it for small zones, you run into serious sample size problems. All of your adjustments will be serious biased toward the home player (in this case Manny), as almost half the Fenway data is comprised of him. If you combine zones or just use one adjustment factor for the entire left field, then you have a larger sample and thus a more accurate adjustment factor, but because you are "generalizing" an adjustment for all zones or for very large zones, if a player has a non-typical distribution of balls in play, his adjustment may be wrong. Knowing whether a ball is a wall ball is the least of the problems. In fact, that is not a problem at all, although, as I stated, I don't consider that, and can live with it.

If you read the original article in the NYT, you will see that at least half of Manny's damage is on the road,

i read the article and see it asserted, but i was hoping to see some actual numbers. is "at least half" 17 or 25?

thanks for the good work, and don't let a few jackasses deter you from posting here. your work is too important to the conversation.

I am not sure why logically a home fielder should catch more balls than a visiting fielder



Better knowledge of the effect of wind conditions, park angles, nearness of wall and other impediments.

Lastings Milledge looked lost at Fenway last year. He had no idea where the Wall was, and misplayed 2 balls in one inning, IIRC. Is he an awful LF?

The biggest "evidence" that Manny is a terrible left fielder is simply that there is virtually no such thing as a 35 yo average or better LF'er who was not a star defender when young.

Ehh...it's a house of cards argument though, since there's some evidence he was a plus defensive RF when young. I guess I am very leery of an argument that assumes because he's there he must be bad....aging patterns and such are true in large samples but not true universally, basically. Not that I doubt he's a bad defense LF, but I've long been in the -10 camp not the -20 to -30 camp.

mgl, what's your comment on the Arthur road data? That is, to me, one of the more interesting Manny defense points I've seen the last couple years.

Also, wasn't the HFA included in the calculation of the last UZRs you posted?

I'm not convinced of any of the arguments that Manny must be a poor LF.

1. He was an average fielder before, and he's older, so he must be horrible. But this should be true for all sorts of players, and Manny has been reported as historically bad. I think this sort of generalization is a severe misuse of the statistics - by definition we cannot apply them definitively to one player.

2. He's just as bad on the road. It seems that this is not entirely accepted, but even if it were, road performance is not the same as overall skill. It's half the sample size and not a random half either.

mgl, how can you account for Dial's observation - i.e. that there are a huge number of balls off the wall that are counted as in Manny's zone, enough to be a big deal, and not in line with your park factor?

If someone is not comfortable with a park adjustment for a player, especially one with many years of data, simply use his non-Fenway (or whatever park you are not comfortable with) numbers (road, and while playing for CLE). That should still give you a pretty good idea of a player's fielding value/talent. However, if nothing else, you can certainly infer SOMETHING more (using a larger sample) about a player's fielding by comparing his stats in a quirky park (like Fenway) to other fielders at that same park, again, adjusting for HFA.

If I were trying to calculate Manny's "true talent", i.e. neutral park fielding ability, then you are correct, using only non-Fenway numbers would be the preferred methodology. But I am not. I am trying to calculate the amount of runs that Manny cost the Red Sox while playing left field for them, in their park and the parks of the opponents that they play during the season. Calculating true talent, as you do, and runs cost or gained by a player in the context of his existing team are two different questions that require different methodologies. I believe that my methodolgy gives the best answer to my question, which also happens to be the question being discussed in the newspaper article. It is not surprising that we get two very different numbers when we are not measuring the same thing.

Knowing whether a ball is a wall ball is the least of the problems. In fact, that is not a problem at all, although, as I stated, I don't consider that, and can live with it.

That doesn't make sense to me.Is Manny catching 50 BIP that are short or not in his zone? And there's no requirement that the LF has to play so shallow he cannot get back on balls hit to the wall - that should be a tradeoff.

Fenway's LFs in ZR (and thus UZR) have always been significantly downgraded due to the Green Monster. That's so obvious to me after 10 years of using their data. I noted it in 1997, and that's cleearly the biggest factor in evaluating the BoSox LF performance. The other things you mention don't add up to be the same effect as 50 BIP that are too high off the wall.

Would Boston be a better team with someone like Jason Michaels in LF?

While there are some legitimate points herein, the dicussion is out of hand. No matter how you spin it, the data indicate that Manny is most likely a poor fielder.

Misterdirt, my methodology gives you precisely what Manny was worth given his exact distribution of balls, both in Fenway and on the road, and not a league-average distribution. I don't know where you got the idea that it was otherwise. The only thing I do which "normalizes" the data to a league average player is when giving a UZR "per 150 games" I assume a league-average number of total chances per game. However, if you simply use Manny's raw UZR number (or his UZR per 150) and apply it to his actual number of games, you get his actual value as a Red Sox given his actual distribution of balls - IOW, how he did as compared to a league-average fielder.

FWIW, here is some more data:

Here is Manny's UZR, using my typical methodology, however, his results in each zone are compared to everyone else, home and visiting players, at Fenway. So, for example, all wall balls should have a zero or near zero caught rate, so that regardless of how many wall balls Manny had, he does not get docked or credited with anything.

03 -3 (-7 per 150)
04 +8 (+17 per 150)
05 -15 (-30 per 150 games)
06 -12 (-29 per 150 games)

Here are Manny's road numbers. Basically, I used the same methodology I always use, but I removed all data from Fenway.


03 -9 (-22)
04 -5 (-12)
05 -7 (-14)
06 -9 (-22)

If you do it for small zones, you run into serious sample size problems. All of your adjustments will be serious biased toward the home player (in this case Manny), as almost half the Fenway data is comprised of him.

But if you don't attempt to make those adjustments, you're biasing the analysis *against* the home team player, because every one of those quirks affects the home team player more than it affects any single visiting team player.

A left fielder in Fenway can't physically make the same set of plays that a left fielder can make in almost every other ballpark, no matter how skilled he is. He has to play 20-30 feet closer to the infield, cutting off his ability to get to balls hit toward left-center. He can't normally go all the way back to the wall to attempt a play if he's not sure of the carry, because if his judgment is off he winds up giving the batter an extra base or two by not peeling off and playing the carom. Any system that hopes to assess defensive value accurately has to take that into account in some way, and IMO it's better to try to make the adjustments for the odd quirks and run the risk of being biased in favor of the home player than to not make the adjustments and assure yourself of being biased against the home player.

-- MWE

Maybe I am not making myself clear. I am not advocating NOT doing any adjustments. I am discussing the up and downsides to using small or large zones to make the adjustments. Presently, I use one single number to adjust ALL of a player's stats in each park, regardless of the zone it was hit in. For example, in Fenway, I think the LF park adjustment is 81, meaning that LF'ers catch 81% of the total balls caught in all parks in the AL. So if Manny catches 60% in Zone A in Fenway, he gets credit for catching 60% divided by .81. Same in zone B, etc. That is not a great way to do it of course, but neither would having an adjustment factor for each of the dozens of zones in LF. There would be no good way to caluclate those adjustment factors for each zone, although I suppose you could use some kind of regression equation to make up for sample size (small sample sizes for each zone). I suppose the best compromise would be to assume a zero catch rate for all wall balls and then use one adjustment factor for everything else. What many people fail to get is that it does not make that much of a difference, Changing from for a poor park adjustment methodlogy (which I use) to a much better one, is not likely going to change the numbers all that much. The reason people don't understand this is that there is NO BIAS in a the poor methodology I use. For a large enough sample, my poor methodology (using one adjustment number for all zones, even wall balls) is going to generate the exact same results as using a perfect methodology! Heck, Dewan does not do ANY park adjustments at all and that is fine by me. It simply does not make all that much difference and as I said, using a flat adjustment factor of 81 in LF in Fenway is just fine. If in any one year or years, Manny or anyone else happens to get a larger or smaller than average proportion of wall balls, then so be it. Simply part of the noise in the data. That noise of course diminishes (and approaches zero in fact) as the sample size grows.

FWIW, here is some more data:

Here is Manny's UZR, using my typical methodology, however, his results in each zone are compared to everyone else, home and visiting players, at Fenway. So, for example, all wall balls should have a zero or near zero caught rate, so that regardless of how many wall balls Manny had, he does not get docked or credited with anything.

03 -3 (-7 per 150)
04 +8 (+17 per 150)
05 -15 (-30 per 150 games)
06 -12 (-29 per 150 games)

Here are Manny's road numbers. Basically, I used the same methodology I always use, but I removed all data from Fenway.


03 -9 (-22)
04 -5 (-12)
05 -7 (-14)
06 -9 (-22)

***

This is great! I would consider this pretty much the final word on Manny's defense, since now we're finally comparing apples to apples. Mickey, just for the heck of it, could you tell us what those home numbers look like using your current UZR methodology (e.g. compared to league average and park-adjusted)?

Any thoughts beyond noise on why Manny's defense went from good (-7/+17) to awful (-30/-29) at home while his road defense remained unchanged (-22/-12) and (-14/-22)?

FWIW, the Sox ERA+ in those two two year periods went from 105/116 to 93/96. Before we declare the final word on Manny's defense it would be nice to know if the number of uncatchable yet in play wall balls against the Sox pitching staff remained roughly the same or jumped as their performances sank.

Before we declare the final word on Manny's defense it would be nice to know if the number of uncatchable yet in play wall balls against the Sox pitching staff remained roughly the same or jumped as their performances sank.

***

Since wall balls are removed in that data (or more properly, the probability of catching a wall ball is 0), the number doesn't really matter for how Manny is rated. And how is -3 runs "good"? It seems pretty clear to me from the data that Manny has been a poor and declining fielder who had a fluke good season at home in 2004.

I think Philly is asking how Manny could go from below average (-7) to GG quality (+17) and then to to historically bad (-30) at home in consecutive seasons.

When you see data that is so wildly inconsistent, it causes one to raise ones eyebrows and question how reliable the data is. I'm not willing to just accept it was "a fluke".

***

You're talking about 75 games! In 2005, Angel Berroa had a .787 OPS at home -- does that make him a good shortstop? (Berroa is the first guy I looked at, BTW, I'm sure you could find even better examples.) A lot can happen in a small sample -- that's why we use as much data as we can find.

Here is Manny's UZR, using my typical methodology, however, his results in each zone are compared to everyone else, home and visiting players, at Fenway. So, for example, all wall balls should have a zero or near zero caught rate, so that regardless of how many wall balls Manny had, he does not get docked or credited with anything.

But that simply doesn't work. Well, it might, but you an't know if it works primarily because the BIP distribution may not even out - teh Red Sox may have far fewer (or more) "wall balls" for the opponents LF to miss. This position assumes a relatively even Red Sox wallballs/Opponents wall balls.

my methodology gives you precisely what Manny was worth given his exact distribution of balls, both in Fenway and on the road, and not a league-average distribution.

I assume this is somewhat off: the run value of each play that hits off the wall - you use an average of those plays, not the exact value of hte play - is that not true - because using the exact value of the play is wrong.

I suppose the best compromise would be to assume a zero catch rate for all wall balls and then use one adjustment factor for everything else. What many people fail to get is that it does not make that much of a difference, Changing from for a poor park adjustment methodlogy (which I use) to a much better one, is not likely going to change the numbers all that much.

It will likely make a significant difference for Manny, when he has a FB staff - you note that given a large enough sample size it should even out, but when we want to evaluate whether or not Manny is worth his contract (or his next one), it will matter, because the difference between -32 and -17 is quite a bit, i terms of millions of dollars.

The reason people don't understand this is that there is NO BIAS in a the poor methodology I use. For a large enough sample, my poor methodology (using one adjustment number for all zones, even wall balls) is going to generate the exact same results as using a perfect methodology! Heck, Dewan does not do ANY park adjustments at all and that is fine by me.

You know, I have taken tons of heat over the last decade regarding my statements about park factors for defense, and how they were mostly not necessary - and usually I tried to make the "Park adjustment" with a "pitching staff adjusment".

FWIW, here is some more data:

That doesn't agree with Joe Arthur's work.

And how is -3 runs "good"? It seems pretty clear to me from the data that Manny has been a poor and declining fielder who had a fluke good season at home in 2004.

This is getting extremely nitpicky, but I didn' say -3 was good. I said -3 and +8 over a two year period was good. Actually I used the per 150 figures (-7 and +17) which makes it look more substantial. I suppose I should have said average to perhaps slightly above. And then he suddenly became awful. I think that's a more accurate read of the data than that he started poor with a single fluke season, but whatever.

I do find the tone of defensiveness (a lighthearted pun) and finality in the UZR camp to be very much offputting and quite a bit contrary to the spirit of open and engaged research. That mgl finds his numbers in this case "good enough" is a perfectly valid point of view. I happen to think that far too much time is spent trying to attain theoretical accuracy. "Good enough" is usually plenty "good enough". But in this case we know that there is a flaw in the data. And for some people that "good enough" is not "good enough". That's fine too.

The collegial thing to do is to step aside and let them pursue it to a point where it is "good enough" for them. It's not to unilaterally declare that something that hasn't been measured probably doesn't make much of a difference and then declare the debate over.

Was it over when the German's bombed Pearl Harbor? Yes, I'm cheekily using that as "collegial humor". It was, afterall, set at a university.

That's the equivalent of about 225 plays. That's a decent sample size.

Eh ... not really, at least according to MGL, et al., who have said in the past that two years of defensive data have the same significance as one year of offensive.

There are about 1.8 plays per game for corner OF'ers. When DSG (whom I did not know was in the "UZR camp") says "the final word" he means that is about all you can do with the data. I agree.

As far as "why a player's numbers fluctuate" there is no answer to that when you are looking ar relatively small samples of data chunks. Any "why's" that might exist are drowned by the noise.

Most of the noise is (essentially and practically) random and cannot be eliminated completely. As far as defensive data is concerned, most of the noise is:

Where within a zone a ball is actually.
The hang time of the ball withing the various hit types.
The starting position of the fielder (which you may or may not consider part of his "skill set").
The base lines that are established for each zone - those are based on sample data and have an uncertainty level.
Park and environmental (e.g. weather) variations.
More

Some of this noise can be mitigated by using more data and watching video.

I think DSG is asking for his home values using my basic methodology. I am not sure. I gave the home values as compared to all other players in Fenway only, which means that the adjustments for Fenway are essentially per zone. My basic methodology just uses one adjustment number for all balls hit to LF at Fenway (park index of 81). The results should be similar. I don't know which would be better. One clearly is not differentiating between zones and the other is suffering from sample size issues for each zone (of which there are dozens in LF). Clearly for some large number of games, using the Fenway data only is best and for some small number of games, the individual zone adjustments for Fenway are almost meaningless and will put a lot of noise in the results for any one player.

Anyway, the home numbers using my basic methodology are the total ones minues the road ones which I posted above. These are absolutes and not per 150 games.

Year Home Road Total
03 +1 -9 -8
04 -2 -5 -7
05 -37 -7 -44
06 -24 -9 -33

Here are the home numbers side by side using both methodolgies

Year Basic Fenway only/adjustments per zone
03 +1 -3
04 -2 +8
05 -37 -15
06 -24 -12

If we were to use the "new" methodology for Manny at home and combine that with his road numbers, we get total UZR (not per 150) for Manny of:

-12
+3
-22
-21

Much better than what I have, but, as I said, pretty much no matter how you shake it or spin it, he comes out looking very bad.

Remember that we are not arguing over someone like Betancourt or Cano, someone who is young or even relatively young, someone who there is some disagreement among the scouts, or between the scouts and the statheads. That is where this "argument" mystifies me. We are talking about a slow, lazy, 25 yo left fielder who is almost universally regarded as a poor fielder. Even if we had no data at all, what would we estimate his runs cost/saved per year, given what we DO know about how much the best and worst fielders cost/save a team? -10 -15 -20? I generally tell people that, depending on the position of course, a bad fielder is -10, a very bad one is -15, and a horrendous one is -20 (a few are probably legitimately worse, like Griffey, who is data-wise the worst fielder in baseball, bar none). With Manny, take your pick.

35 yo of course, not 25.

Speaking of Griffey Jr... Ken Rosenthal reports that:

When the Reds' Ken Griffey Jr. makes his Grapefruit League debut, it will be in right field, according to a source with knowledge of his situation. Griffey, recovering from a broken left hand, does not mind shifting from center to right, the source said. However, Griffey was under the impression that the Reds would delay their decision until midway through spring training while attempting to acquire a legitimate center fielder. Instead, he learned at the start of camp that he would be in right.

Around the horn: Griffey to start in right

I gather Griffey would still be a below average defender in RF?

We are talking about a slow, lazy, 25 yo left fielder

Manny being Manny.

Would Boston be a better team with someone like Jason Michaels in LF?

That would mean that either Ortiz or Ramirez sits, or is exposed to an even more challenging position every day. I don't think so.

Now I think Boston would actually be well-served by trading Ortiz for some other help - I think he's great, and I believe the perception around baseball has him as the "Duke of Clutchiness". Not believing that clutch hitting is repeatable, I think that his trade value will likely never be higher than it is today.

This has the benefit not only of helping the Sox out in other ways, but of getting Manny in the DH slot on a very regular basis.

On the other hand, in the real world, the city of Boston would explode if this were done. But be on the lookout for the Boston writers to start slapping Ortiz around - that's the way to see that the FO is ready to jettison him.