Brock,

I think most of the systems that are out there ask a different question. Rather than trying to assess the DH's absolute defensive value in any way, they ask how easy it is to find someone who can do the same job. And since the job involves no fielding, the correct answer is "very easy." But it's not exceptionally difficult to find someone who can play first base or left field, either.

Eric - All the systems I have seen, or even read about, understand what you're saying. But that's not what they do. Instead, they assign a value of "zero" to anyone who plays right at the level where replacement talent would not only be "very easy" to find, but "freely available". That's what replacement rate means: It's an attempt to quantify just about how good a freely available replacement player would be. It's broken down, for position players, into Offensive and Defensive replacement numbers, but the replacement rate stays the same: Zero. Mathematically, it works out really well to assign the number zero the the replacement rate. That way, if you see someone with a negative accomplishment number, you can say that this guy needs to go back to the minors; there are freely available replacements who would play better. I've used the concept since I got hold of my first Bill James Baseball Abstract, back in 1981, and was introduced to it. And the only real problem I've had, other than trying to figure out just how high to set the replacement bar, was trying to figure out how much of a player's value is in his hitting and how much is in his defense, given that the various defensive positions have different values. But DH is different. Talking just about defense, DHs don't have a replacement value of zero. They produce LESS defense than any replacement who actually played the field would. Their defensive value has to be less than that of the people who play the field. That is, it has to be less than a really really bad defender; the worst defender that you can think of.

This throws everything off. For one thing, since a DH has a zero opportunity to play defense, you end up with formulas that divide by that zero, which blows them up mathematically. Also, the theoretical bottom end of defensive play - my hypothetical dead body - is way, way far below the replacement rate. If I were to actually try to compute out some answer to the question "what is the defensive value of a dead body?" I would end up with something ridiculously low. It might be low enough to offset all the DH's hitting. And that's not fair, because the DH has zero opportunity to show that he has at least a little defensive skill. So I end up reasoning in circles, devising formulas that end up dividing by zero, or getting results too low to use. What I want to know is if anyone has solved this in any way that they think is satisfactory. I do have the one that I am willing to use, but I'm also willing to listen to people who may have hit upon something I haven't noticed yet. That is, I'm sure of the theoretical trap and how bad it is, but I'm not completely sure of my solution. I'm willing to use it in the absense of anything else, but I'll take help from the dedicated formula-devisers here on BTF. That's all that's going on right now.

I think that's the right way to think about it. How bad of a bat are they willing to put at DH? Usually it is a league average hitter or better. Someone is going to DH, you measure the DH contribution over what other player they would have hitting that day.

But DH is different. Talking just about defense, DHs don't have a replacement value of zero. They produce LESS defense than any replacement who actually played the field would. Their defensive value has to be less than that of the people who play the field. That is, it has to be less than a really really bad defender; the worst defender that you can think of.

It's a question of absolute value vs. relative value. Win Shares, for instance, attempts to deal in absolute value, giving anyone who plays the field a positive fielding score. I believe Dan R's research indicates that this has resulted in an artificial compression of the range of fielding value (if that makes sense).

A relative value system, like Dan R or AROM's WAR, compares the player to a scrub who'd be asked to do the same job. At DH, this will be a Micah Hoffpauir type AAAA hitter, who you'd expect to put up something like a 100 OPS+. In left field, it'll be either the Hoffpauir type (100 OPS+, -15 defense) or, say, a So Taguchi type (80 OPS+, +5 defense). This will occasionally result in someone like Adam Dunn having a lower defensive value score (defense + position adjustment) than a DH. I suppose it could be argued that such a player should always be used at DH, and thus that the negative score beyond that point is the fault of poor deployment of resources by the team, rather than the player himself. Of course, that argument would apply better if Dunn didn't play in the NL.

Regardless, these systems aren't trying to assess absolute value at all.

How bad of a bat are they willing to put at DH? Usually it is a league average hitter or better.

Or at least someone they think is a league average hitter... *cough* Jose Vidro *cough*

Maybe another way to think of it is: would you ever trade a HOF-caliber position player at his peak to get Edgar or David Ortiz or Travis Hafner at their peak? The 100%-time DH who hits that well is extraordinarily rare, but that doesn't mean he's necessarily extraordinarily valuable. And if you'd never think of trading a HOFer even up for him, then he isn't a HOFer.

Evidently DL's is a reply to EricJ.

Quoting brock #101 103 105 here and there:

You can't just write off a DH's defense as zero, because "zero" is the replacement level of whatever your system is, and a DH has noticeably LESS defensive value than that.

No, you are disputing the very idea of replacement level.


That's what replacement rate means: It's an attempt to quantify just about how good a freely available replacement player would be. It's broken down, for position players, into Offensive and Defensive replacement numbers,

Only some systems really have separable zero points for batting, baserunning, fielding, pitching. I doubt that that approach is really workable, although it may be possible to answer the DH question adequately on those lines. Maybe we'll find out.


every system that I have ever heard of assumes that there are guys who go out in the field whose defense is worse, mathematically, than the system's zero point, which is its replacement rate.

On that line (separable components of replacement level), the players with negative fielding value have compensating plus value in other components or they are transient. They play while they establish a record, while management favors them at the expense of winning, while they some transaction cost is greater than their net minus value. (In the basic model they don't play.)


This throws everything off. For one thing, since a DH has a zero opportunity to play defense,

On that "separable components" line, you may need to recognize that every player has fielding value at every one of the nine positions, nine fielding values (typically unobserved); every player has pitching value (typically unobserved); every player has batting value (practically unobserved for AL pitchers).

A pure DH has no fielding record at any of the nine positions. A pure rightfielder has no fielding record at any of the other eight positions. Neither one bears any more responsibility than the other for the minus value of any teammates at those eight positions, which is zero responsibility in every system that I think I understand.

Your hypothetical team wouldn't use you at shortstop, Brock, if its DH Edgar Martinez were able to play shortstop. Nor would it do so if its RF Harry Heilmann were able to play shortstop. If you do play, we don't charge either Martinez or Heilmann with any part of their own minus values at shortstop. To charge Martinez partly for his inadequacy at shortstop would be a novel approach, as far as I know, but it should go hand in hand with charging Heilmann partly for his own inadequacy at shortstop, and there is no special challenge regarding Martinez.


That is, I'm sure of the theoretical trap and how bad it is, but I'm not completely sure of my solution.

I don't believe there is a trap. That is, designated hitters pose no special problems or replacement-level analysis.

I suppose it could be argued that such a player should always be used at DH, and thus that the negative score beyond that point is the fault of poor deployment of resources by the team, rather than the player himself.

Exactly. This points to the flaw (to me) in Brock's approach. Edgar is a rare animal: he's such a good hitter that his team refuses to risk injury by playing him in the field. He wasn't an Oliva or Cepeda type who was so hobbled that they couldn't play the field; rather, it was the team's choice not to have Edgar in the field, even though he was perfectly capable of manning first base, at a level likely above replacement level (except for his last few seasons). That Edgar calmly accepted the role the team gave him should be seen as a positive.

It goes back to the idea that we should neither credit nor penalize player's value for "what-if" hypothetical situations. I would also argue that we should not penalize players for decisions of their team in how they are used. In the common situation of great players performing below replacement level in some seasons (usually near the beginning or end of their career) I believe in ignoring these performances rather than deducting them from the positive value the player contributed in other seasons. It was their team's decision to keep playing them, with the perception being that they DID contribute value to the team.

So it is with Edgar's defense. His team made the decision that his bat was so impactful on the good fortunes of the team that they did not risk injury by putting him in the field. The team was then constructed with this decision in mind, so as to minimize the downside; not having much flexibility with the role of DH. What, so Edgar should have said, "No, for the good of the team and for the future analysis of my HOF worthiness, I refuse to fill the role that the team thinks I am most valuable filling. I must play the field." The team then tells him, "Nope, we got it covered. Just keep pounding out that 160 OPS+ for 600 PA every year and you'll earn your HoM plaque." Which he did.

As far as "computing" the defensive value of a full time DH, it's zero. The reality is he didn't play the field, there are no numbers to compute, neither positive nor negative. Set aside the hypotheticals and phrases such as "a DH's defense".

I agree with Paul Wendt. Brock, I'm afraid you've left the reservation and are descending into karlmagnus/yest territory with this one (although I suppose you might have joined them long ago with your quixotic fetish for 5th-tier deadball pitchers :)) There are only two ways to look at this, IMO:

1. A value approach, as taken by myself (see http://www.nytimes.com/2009/12/06/sports/baseball/06score.html), the vast majority of HoM participants, and all uberstats. This sets DH replacement level right around league average (or a bit lower if you give credit for the supposed difficulty of DH'ing, as CHONE does).

2. An ability approach, in which case you'd ask yourself the hypothetical question, "If Edgar had been asked to field, where would he have played on the diamond, how good would his defense have been, and how would it have affected his durability?" In that case, you'd probably put him at 3B through say 1996, descending from league average at the beginning to somewhere below by the end, and then to 1B, starting above average and drifting to well below by 2002-04. That would, of course, actually *increase* his raw value. But then you'd pencil in maybe 30 missed games a year, and a career maybe 2-3 years shorter, reducing his value probably to a bit below where he would show up by the first method--say, 50ish WARP instead of 55.

No DH would be -100 in the field. The most you could conceivably penalize them would be to call them, say, a -20 1B (compared to the actual aggregate defensive performance of primary-DH's playing 1B, which is -9 runs per 162 games).

Dan says, "No DH would be -100 in the field. The most you could conceivably penalize them would be to call them, say, a -20 1B (compared to the actual aggregate defensive performance of primary-DH's playing 1B, which is -9 runs per 162 games)"

This is, right now, just about what I'm doing. I guess I'll see what happens to the ranking. I have him lumped in with the first basemen and am treating him as if he were a very bad defensive first baseman (Dick Stuart). I don't think that this approach is over the river and through the woods. Its just that I can't justify taking the approach that a DH can have a higher defensive ranking than someone who actually plays some defense. Until we have voted on enough DHs to figure out what seems to work, I've got to theorize. Edgar just happens to be the first test case.

Approach #2 has its appeal, if I thought I could reconstruct a man's history with that many distortions. Ending up with 50 WARP instead of 55 would suit me just fine. But it involves even more guessing than what I'm doing now.

Dan G's comment makes a lot of sense, too, with regard to Edgar. But Edgar's peer group is soon going to consist of a lot of guys who are DHing because they really would be unacceptable at any defensive position. I may just let Edgar's defense sit there at replacement rate, because even I remember that he was not helpless in the field when his team asked him to DH. But some of his upcoming peers, like David Ortiz, aren't like that.

Oh, I ought to say thanks for the help to everyone. I realize that some of you think I've gone off the reservation, but I'm not getting flames. That helps. And if it makes any difference, I now realize that I've got to start making more adjustments to deadball era pitchers than I had been. I doubt that Phillippe or Leever will be on my next yearly. If so or if not, though, they'll be right together. I just need to plow through a lot of pitcher analysis that was supposed to be done in September or October, when it turned out I was sick for 6 weeks. This year, without positionals to consider, I'm taking the time to sort out the pitcher adjustment thing right after the holidays. I don't even know whether Newcombe or Dean will survive on the ballot.

Until we have voted on enough DHs to figure out what seems to work, I've got to theorize. Edgar just happens to be the first test case.

Is there any candidate like Edgar on the horizon? You mention Ortiz, but his career is sinking fast, and as a candidate he's still somewhere below Frank Howard and several miles short of Edgar. Frank Thomas would be another DH candidate but he's obviously far above the line. Harold Baines and Chili Davis are non-starters as candidates. No, Edgar looks like a unique case.

Its just that I can't justify taking the approach that a DH can have a higher defensive ranking than someone who actually plays some defense.

A simple way to do that would be to put a limit on negative defensive value during the DH era, with the assumption that anyone below a certain level should have been used as a DH.

Treating DH's as very bad defensive 1B's seems perfectly fine to me. It's just a question of whether you see "very bad" as -9 (which is their true value) or something south of that (which implicitly could serve as a way of counteracting the extra longevity that DH'ing provides).

One problem with this discussion is that there is that, in contrast to the case of hitting, there is no such thing as "zero fielding value". There is in fact no real lower limit to how bad a fielder can be; a team could, say, play Jason Giambi at short for an entire season. Basically every groundball to short would be a hit, and he would probably cost the team something like 300 runs per season. Surely you can appreciate that giving a DH -300 runs because that is the worst a fielder could possibly be doesn't make any sense?

Is there any candidate like Edgar on the horizon?

This is a fair question that rightly questions the necessity of worrying too much over Edgar's rather unusual case, but it's also pertinent to consider if there is any candidate like Edgar in the past? Is there a group of players who are serious HoF/Hom candidates, against whom a career DH like Edgar Martinez could be unfairly advantaged are great-hit, no field players from the pre-DH era (or from the NL, but when the DH exists in the AL, players like that would presumably gravitate to the AL DH positions)? This is more a "Hall of Merit" question than a "Hall of Fame" question, but I can see why one might be uncomfortable simply assigning zero fielding value to the DH.

The three players who come to mind who might fall into this category are Gavy Cravath, Frank Howard, and Jack Fournier.

Myself, I'm content to accept that players' value can't be completely divorced from context, context including both the nature of the game and the managerial decisions that influenced the player's opportunities. It's worth trying to correct for systemic injustices like segregation or generational circumstances that affected most players of a generation adversely, like compulsory military service, and one might go farther and correct for players being trapped in the minors when they were actually very good or even great ballplayers. So I'll give MLE credit to Cravath and Fournier for their time in the high minors when they should have been in the major leagues. Correcting for whether the DH was an option or not goes beyond what I am inclined to adjust for, however. I can see how others might want to do it, though. But do you zero out or reduce defensive penalties (and loss of playing time) for players like Fournier and Howard, or do you reduce the value of modern DHs below their actual value to their teams to accomplish this? I am disinclined to see either adjustment as necessary or fair. The nature of the game was what it was, and the responsibility of the player is to play that game as well as he can, given the rules he is playing under. The game has changed a lot over time, but we honor the players who excelled in the game as they found it. That to me means assigning merit to DHs according to their contextual value. But I can understand why someone might not be comfortable with that.

WELL, now. THIS is what I call getting some help! Thanks to all. To just write up where I am now, I think I'm going to go with a zero defensive value for Edgar, because it's true that he could have played the field if his team hadn't wanted him to. That puts him right on the borderline of my ballot, but I think it's fair to Edgar's HoM candidacy.

To respond to a few of the comments above:

Eric J (107) syas, "It's a question of absolute value vs. relative value. Win Shares, for instance, attempts to deal in absolute value, giving anyone who plays the field a positive fielding score. I believe Dan R's research indicates that this has resulted in an artificial compression of the range of fielding value (if that makes sense)."

Actually, Win Shares rests on a system called "marginal" contribution, which is a well-defined economics math term that I don't remember enough economics to define. That's why he complains that he isn't using a replacement rate. His mathematical model doesn't work that way. Dan, who generally knows what he's talking about when it comes to math, may be able to help here, and may well be right about the compression issue.

Paul (109) is a very helpful post. I think (could be wrong) that the crux of it is in this paragraph:

"Your hypothetical team wouldn't use you at shortstop, Brock, if its DH Edgar Martinez were able to play shortstop. Nor would it do so if its RF Harry Heilmann were able to play shortstop. If you do play, we don't charge either Martinez or Heilmann with any part of their own minus values at shortstop. To charge Martinez partly for his inadequacy at shortstop would be a novel approach, as far as I know, but it should go hand in hand with charging Heilmann partly for his own inadequacy at shortstop, and there is no special challenge regarding Martinez."

As a practical matter of managing a ballclub, this is entirely correct. They'd have Dick Stuart out there at short before they'd have me out there. And every system I know of would give him a negative number for his defense, because he would be way below the replacement rate. Pete Palmer, who assigns league average as the replacement rate, would give him a REALLY lousy number. (All this is assuming that I was Ted Williams as a hitter; otherwise, I get shipped out on the train with Buzz Arlett).

But as to the general point, it's not true that, say, first basemen do not get charged for their weaknesses at playing shortstop. Skipping the additional complication of replacement rate, all the methods that I know of assign an "arena of opportunity" to each position. That is, Bill James, say, assigns, say, 3 games (in a year) to first base defense, and maybe 10 games to shortstop defense (both numbers may be high and I didn't include a replacement rate). So a first baseman, no matter how well he plays, is limited to 3 games of contribution, while Ozzie Smith could pile up as many as 10. The way that these systems work is to figure out what percentage of these defensive games the player actually "won" and multiply by the arena of opportunity. So let's say that Ozzie is a .900 defender (whatever that means). .9 x 10 = 9, so Ozzie gets credited with 9 defensive games won. But a first baseman who fields at a .900 rate gets only .9 x 3 = 2.7 games worth of credit in his WAR. The problem with DH is that there is no arena of opportunity, so you (or at least I) constantly end up frustrated when it come to the multiplication stage. Without an arena of opportunity, a .900 defensive DH would get .9 x 0 = 0 and a .100 DH defender would get .1 x 0 = 0. And that's assuming that I have a percentage to work with at all. If I try to divide plays made by possible plays made, in order to get the percentage, I end up with all sorts of formulas that generate 0/0 as the number I want for DH defense, and 0/0 is undefined. That's my problem.

Dan G (113) may be right that there's no real DH candidate near the borderline coming real soon (I agree that Thomas is a gimme), but I would also like to assign some defensive number to people like Dave Winfield and Jim Rice, who played several years there. That is, this issue has been building in my head for a while here. Edgar is the first time I've ever had to turn in a ballot that might well depend on what number I assign to someone's DH defense. But there will be almost-entire-career DHs coming along, and someday one of them is going to end up on the borderline.

Dan R (116) really helped clear things in my head. He's absolutely right. If I assign a defensive value to a DH that is what a very bad first baseman would get, I'm probably going to be OK as a practical matter. Since my area of math is applied math, I'll end up comfortable with that. I can compute out an "arena of opportunity" by looking at how many defensive innings first basemen get per plate appearance. That, too, will be an approximation, but we're in applied math here, where I am much more comfortable than I am trying to think out theory. So, just to ask, what does "-9" turn out to mean? It's too large an absolute value to be games at first base defense. Is it runs?

Blackadder says (117), "One problem with this discussion is that there is that, in contrast to the case of hitting, there is no such thing as "zero fielding value". There is in fact no real lower limit to how bad a fielder can be;"

Well, yes there is, and it's the cause of all this angst. The lowest limit a fielder can have is related to the arena of opportunity. That is, no shortstop, not even me, can be worse than making no plays at all in his arena of opportunity of ten games. He can't have a Fielding Winning Percentage less than zero, so he can't lose more than ten games for you with his lack of glove. It's the lack of an arena of opportunity that makes DH unique. We can all agree that they make no plays, but as a practical matter, they also have no arena of opportunity, so you end up dividing zero by zero again. As far as I know, every system I've seen just looks at that 0/0 and says it's zero. But that's not mathematically correct. Mathematically, it's undefined. That's why I asked for help. I got to the 0/0 stage with no problems, but completely hung up right there.

Chris (referencing 118) -

I don't know if there are any current candidates, but if you go back to the right field positionals and look at my posts comparing Harry Heilmann to Dave Winfield in terms of career length, you can probably see that I was already trying to think through the can't-play-defense-and-there's-no-DH-available concept. That thread was indeed a precursor to this one.

Again, thanks to everyone for thinking rather than flaming. Boy, do I appreciate that. Right now, I'm in general down with Dan's post 116 above, although I think I'll end up giving Edgar more credit than the absolute lowest, because he could have played in the field for at least some of the years he was DHing. And who knows? Maybe it will just turn out to be impossible to hit well enough to have a borderline HoM career if you were never really athletically able to play defense at all.

Thanks! - Brock

Brock,
Paul (109) is a very helpful post. I think (could be wrong) that the crux of it is in this paragraph:

"Your hypothetical team wouldn't use you at shortstop, Brock, if its DH Edgar Martinez were able to play shortstop. Nor would it do so if its RF Harry Heilmann were able to play shortstop. ... "

That paragraph is supposed to illustrate the two that precede it. It's the crux if concrete examples tend to be crucial for you, but to continue the farce of putting you at shortstop may have been a distraction. The two preceding paragraphs are the crux for me.

quoting myself
>>
On that "separable components" line, you may need to recognize that every player has fielding value at every one of the nine positions, nine fielding values (typically unobserved); every player has pitching value (typically unobserved); every player has batting value (practically unobserved for AL pitchers).

A pure DH has no fielding record at any of the nine positions. A pure rightfielder has no fielding record at any of the other eight positions. Neither one bears any more responsibility than the other for the minus value of any teammates at those eight positions, which is zero responsibility in every system that I think I understand.
<<

That is a pointed reply to Brock #101,
Your [conventional replacement-level rating] system will have some glove butchers with negative defensive W%[WS?] and WAR, because they are worse than replacement, and a DH has to have less defensive value than that, or he would be playing the field and the other guy would be the DH.

Yes, the DH does have less fielding value at shortstop than does the incumbent shortstop Donie Bush, whose fielding value is negative by postulate. RF Harry Heilmann also has less fielding value at shortstop than does Bush. Still, we debit SS Bush for a big share, and RF Harry Heilmann for no share, of the Detroit infield's inability to turn the doubleplay. (Some of us blame management if SS Bush doesn't bat well enough to give him net value at least zero.)

Paul -

First, thanks for trying to work this out with me. Yes, I work a LOT better with concrete examples than theory, which is why I'm better at applied math than theoretical math. I want word problems to solve, not theorems to prove. I was aware that the paragraph I cited was an example summing up the two preceding paragraphs, but when I try to write responses, it's always easier for me to go concrete (matches the thickness of my head and the stiffness of my glove at shortstop).

I THINK we are talking about the same thing. I call it "arena of opportunity." Shortstops have different arenas of opportunity than right fielders do. It turns out that the arena of opportunity for a shortstop is much greater, in terms of impacting games, than the arena of opportunity for right fielders, so the number that represents it in a mathematical system is greater than that for RFs. But you are certainly right to say that the arena of opportunity for shortstops involves no long flies to right field, nor does the arena of opportunity for right fielders include turning the double play. The problem, for me, starts when the arena of opportunity for a DH comes into consideration. It's mathematically zero, of course, and so you can't divide by it to get a defensive winning percentage. (I did mean "negative Defensive Winning Percentage" when I wrote W%, rather than Win Shares; sorry about that). Sabermetric systems compute percentages (not that this is new info for anyone here) by dividing things. With DHs, you end up trying to divide by zero, because one of the things that you have to divide by is the arena of opportunity. That's the problem. You try to divide by zero, you get an undefined number. So, in all sabermetric systems I know of, the inventor has to decide what that undefined percentage is. They have always defined it to be zero in the systems that I have seen. That's fine until you realize that there are real defenders who make real plays, and the system regards their contributions as worth less than that zero that the DH got (Donie Bush was a hilarious example). I want the DH to have the lowest percentage, nothing greater. I just wasn't sure how low to go. Dan's idea of just using the defensive percentage of a really bad first baseman is a good enough approximation for me. This is especially true since a lot of this discussion is theory, where I am weakest.

I brought up the absurd idea of my playing shortstop because it would never happen, so I will never have an arena of opportunity at shortstop. A sabermetric system would therefore assign my defense at shortstop as zero. That does not mean that I would be better at the job than Donie Bush was, even though Donie would get a negative number in the same sabermetric system. It just means that my arena of opportunity is zero, and a sabermetric system (well, all that I know of) have to divide by it to get a Defensive Winning Percentage to compare to the Winning Percentage that the system uses as its replacement point, or zero level. Does that make any sense to you?

Actually, Win Shares rests on a system called "marginal" contribution, which is a well-defined economics math term that I don't remember enough economics to define.

This is true - but I have the impression that they're still intended to represent absolute value, and James couldn't get the math to work without using marginal contribution.

That's why he complains that he isn't using a replacement rate. His mathematical model doesn't work that way.

Of course, when James uses (say) raw Win Shares total for a season to pick the best player, he's using the system's built-in (incredibly low) replacement rate whether he acknowledges it or not.

On Win Shares' compression of fielding value, particularly at offense-first positions: It's this simple. The range of Fielding WS per season given to corner outfielders is what, 2.0 to 4.0? Maybe 1.5 to 4.0, max? So we're talking less than 1 win a year. In other words, according to Win Shares, every LF or RF is within 5 runs of average per year. The true talent range is 3 times as big (-15 to +15), and the actual seasonal figures are far bigger than that.

And yes, -9 is runs. It means that an otherwise league-average defensive team who sticks a primary-DH guy at 1B would be expected to allow 9 runs more than a league average team.

That puts him right on the borderline of my ballot, but I think it's fair to Edgar's HoM candidacy

You do realize that there are no more ballots to put Edgar on, correct? He was elected already.

DL - Are you sure? The title of this thread looked to me like this was the discussion thread for a BBTF simulation of a HoF ballot, which will have its own ballot thread on December 21. Edgar is listed in the Eligible Candidates section in the thread header. It is this "HoF" ballot that I'm working on here. I hope it does actually exist and Edgar is actually eligible. I'm going to feel really dumb if it doesn't.

Dan R - Thanks for the compression note. That would have been my guess. In the terminology I've been using (and sometimes inventing on the fly) here, you're saying that Bill gives corner outfielders too little "arena of opportunity." This may be true. But there are an awful lot of things clamoring for defensive space. Pitching and all the other positions' arenas. The arena for first base must be really tiny. Just to ask, how do you break down the arenas of opportunity between offense, pitching, and fielding? (If you don't use that approach to starting your system at all, the question is meaningless, and there's no reason to answer it beyond saying that.)

Actually, I just took out my copy of Win Shares and looked up what Bill does. It starts on page 67, "Converting the Claim Percentages into Fielding Win Shares", where he notes, about halfway down the page, that first base gets an arena of opportunity of 6% of the arena for fielding, while outfielders altogether get 29%. Later, page 68, item #3, he notes in passing that the outfield shares are distributed 1-2-1, meaning that left and right field each get a quarter and center field gets half. A quarter of 29% is just about 7%, so the corner outfielder arena of opportunity is only 7% of the total fielding arena, while that for first base is 6%. That does seem just a bit off. It's also 7% of the total arena for fielding, which is (page 26, paragraph 2) 32.5% of the total arena for defense, which is half the total arena of accomplishment, both offense and defense combined. So the portion of the entire arena of Win Shares assigned to a corner outfield position is 7% of half of 32.5%, which is about 1.1% of the total, about 2.3% of the defense total. That's a tiny arena, no doubt. An average team, of course, wins 81 games, for 243 Win Shares, so the arena for a corner outfield spot is about 1.1% of 243, or about 2.5 Shares, just a bit shy of a whole win, which is just what you said. I've heard for years that baseball is "15% fielding," so it's compatible with that estimate (half of 32.5% is 16% or 17%), but yes, the arena for corner outfielders is very small, all but identical to that for first basemen.

On the other hand, defense gets half the Win Shares. Pitching has to get a lot of that. Of the remainder left over for fielding, first base and corner outfield are going to get the smallest percentages. So maybe Bill is actually in the right ballpark after all, saying, essentially, that baseball is about 1% right field defense and another 1% left field defense. I do think that corner outfield is getting shortchanged here, but I'm not going to be able to figure out how to rearrange the percentages tonight. Or this month. Or, very likely, this year.

Also, I was hoping it was -9 runs (9 runs below average, not replacement rate). That looks to me like a very good estimate of the defensive value of a DH. I'm going with it until I get a really good reason not to.

Ugh, forgive me, but I can't even begin to penetrate this drivel. Whatever James is doing is simply wrong, flat out empirically wrong. He's trying to force the square peg of the reality of baseball into a round theoretical hole, and it doesn't work.

Just take a look at any play by play defensive stat, or an average of them. The best and worst fielders will be around +20 to -20 runs a year true talent, and +30 to -30 on observed results, at 2B, 3B, SS, and CF; +15 to -15 and +20-25 to -20-25 at LF and RF; and +10 to -10 and +15-20 to -15-20 at 1B and C. Until the gap between the league leaders and trailers in fielding WS in a typical year is 20 WS at the first group of positions, 15 WS at the second, and 11 WS at the third, I will continue to assert that it is a junk stat with no meaningful relationship to baseball reality.

A position player's value is the gap between his total runs above average (batting and baserunning measured vs. league average, and fielding measured vs. positional average as above) and the total runs above average (in practice below average) that a replacement player at his position would accumulate in the same playing time. I don't know what an "arena of opportunity" is, and I don't want to. Forgive me for sounding arrogant, but anything other than this very simple truth just muddies the waters.

If you value a DH the same as a 1B who is 9 runs below average with the glove per 162 games, you will be perfectly in line with the sabermetric consensus (at least among those who don't give credit for the supposed difficulty of hitting without having to field, as CHONE does). That method puts Edgar very close to the HoM borderline, but on the right side of it in my book.

OH! First off, I apologize. I had NO idea that you thought this of Bill's work. I would never have cited it in your direction, much less in this detail, if I had known that. Regardless of whether I agree more with him than with you, it would have been very very wrong of me to do that if I have known. I am in no way offended; I just hope that you're not mad at me. I didn't know. That's all I can say.

As for the issue of DH defensive deductions, I found myself in agreement with you yesterday, so I have good reason to trust your work, regardless of whether or not I understand how you got there.

One thing I will ask is where the heck you go to get a play-by-play database to analyze? I've wanted one for maybe 25 years now, since I know how to program, but don't know of one that a single person can even try to afford access to. My old versions of the BBBA had a computer dump from the ones at STATS, Inc. at the time, but it all had to go through Dick Cramer, who did the programming. As a result, there was, really, just the one big dump that I worked out at the beginning and Dick was kind enough to program out for free. I couldn't ask him to program me up another one, complete with specialty queries, every year. If you don't yourself have access to a play-by-play DB, can you tell me who produces systems that start from them? I know there are people who have access to such things, and that they have systems, but I don't know who has what, much less which methods are most trusted. I would certainly use the info if I had it, but as of now, I don't. The best I can do is try to sort through the various methods threads here at BTF and see if I can decipher what's going on. It's frustrating, and it also is what keeps leading me back to simple analyses of seasonal data from the big encyclopedias.

Again, sorry to toss stuff out there that you don't respect. Had I known, I would never have done that. I will make every effort to NEVER do that to you again. - Brock

DL - Are you sure? The title of this thread looked to me like this was the discussion thread for a BBTF simulation of a HoF ballot, which will have its own ballot thread on December 21. Edgar is listed in the Eligible Candidates section in the thread header. It is this "HoF" ballot that I'm working on here. I hope it does actually exist and Edgar is actually eligible. I'm going to feel really dumb if it doesn't.

You're not hallucinating, Brock. :-)

You're right, I got confused. I kinda assumed the HoM guys had already worked up Edgar.

bjhanke--haha, you don't have to apologize!!! We're just exchanging views on this board, like everyone else. The more I learn about how Win Shares is calculated, the more it seems to me to be a total crock. But there are literally dozen of HoM participants who rely on them like the Bible, and it's not like I get into flame wars with them. PLEASE post absolutely whatever you want, regardless of what I or any other HoM participant thinks of the metric or idea in question. That is how we move forward as a group.

BTW, if what you are looking for is to break down responsibility for wins and losses by position and role, the methodology as I understand it would be the following. Take the stdev of team wins and multiply it by 10 to get the stdev of team runs. Square that to get the variance of team runs. That should, in theory, equal the sum of the variance of all the different components (hitting, running, fielding, and pitching by each player/position). So then RF defense as a "percentage of baseball" would simply be the square of its stdev divided by the prior number.

Everybody uses Retrosheet for PBP data. I myself don't know how to download info off it, but there are tons of people who do--AROM, Tango, etc. You could ask any of them to give you a hand or to teach you.

Brock,
Maybe you should read some of the articles at Play-by-Play Data Files (Event Files) to begin. I feel that I have seen instruction on the web for making a MS Access database from retrosheet pbp event files, and I just revisited this page thinking that it may be the place. Evidently it was somewhere else or I dreamed it.

P.S.
The "Retrosheet era" now spans 1952-2009. That is the run of recent seasons with play-by-play data available for download from the Retrosheet website. From a pilot project or two, pbp data may be available for stray earlier seasons, but generally there is nothing available or box score data only. From some year late in the 1900s back to 1952 Retrosheet has compiled the data from sources like paper scorecards. STATS may be the compiler from that time going forward. Seeing 2009 in the directory is a surprise to me.

Brock - re: DH defense. I don't assign numbers to defense myself, rather a letter-grade score that gives me an approximate value I am comfortable with. I do, though, understand the drive to look at things numerically. What you might want to consider for DH-defense, is the value of the guy whose bat he is replacing in the lineup - the pitcher. The DH is not really a negative in the game with the leather, because his presence in the lineup does not require the loss of a defensive position. His team will have a player seen by them to be at least somewhat capable (we would hope) at all nine defensive positions at all times. 8 of those nine (in the AL) will also have opportunities at bat. 1, the pitcher, will not. He is replaced in the lineup by the DH. How much fielding value do you give to pitchers?
I don't give very much, although I give more for old-time ones. I recall doing a study back when Rube Waddell was still a candidate that showed that pitching fielding range has declined precipitously in the last century. From 1984 through 2003, the AL average range factor per 9 innings for pitchers ranged between 1.62 (2000) and 1.87 (1989), generally trending downwards. Seen like this, a pitcher has very little opportunity to affect a game with is glove (with the caveat that I'm pretty sure that pick-offs are calculated under pitcher range factor as assists). Even the worst fielding pitchers will have a RF of approximately 1/9 innings. Why not view the DH's fielding contributions in that light?

First, Dan, Whew! As I've posted before, I absolutely HATE flame wars. You said nothing ad hominem about me, or even about Bill James, in your post. I was hoping that you were just getting intense about an idea that you have strong opinions about, and that you didn't think I had any ad hominem reasons for citing Bill. The main reason I resort to Win Shares as examples is that I understand the method. The book itself is essentially one long methods glossary with all sorts of examples; just the kind of thing my mind works best with.

One of the oddities of this whole exercise for me is that Win Shares cannot handle DH defense rankings at all, or I would almost certainly have used them. It is an EXTREMELY value oriented system, and doesn't handle rates at all well. And the main benefit I see in Win Shares in general is that its top down approach to fielding works very well. Except for DHs.

The variance approach may well work. I'll have to try it and see. At the very least, I retain enough of my old college statistics that I know what you're suggesting and why to use variance instead of standard deviation.

Dan suggested I try Retrosheet for pbp data, and Paul gave me a link. The link contains all the info I need. It will take me a long time to actually convert Retrosheet files into what I need, because they are set up as text files with records containing data items, with commas separating the data items. The various records do not seem to carry the same number of fields, which is hell if you're trying to convert a text file full of commas into a database. You need each record to have the same number of commas, even if they are empty of data, because the conversion method is looking for commas. On the other hand, comma-delimited files of this sort are the easiest to convert, once you get the number of commas thing worked out. The other problem is that I have a Mac and all the Retrosheet software is build for Windows (actually for the underlying MS-DOS), so I'll have to write all my own software. But once I get the data into records with the same number of fields and commas, I should be able to convert into Access and have what I need to get started. I've programmed Visual Basic and C++ over Access databases, so once I get the data into Access, I should be OK.

Rawagman's idea of giving a DH the defensive value of an average pitcher does have its appeal, and might well come up with reasonable results. Right now, though, I'm going with Dan's -9 runs per 162 games. Just estimating off the top of my head, that number seems like what the DH deduction ought to be. It's not nothing, but it's not horribly huge, either. In any case, DH defense is just vaporware that you need only for purposes like the HoM/HoF, where you're trying to compare players from different positions, using their rates as well as their values.

And to conclude, I really appreciate all the time and effort that various people have put in trying to help me figure this out. It wasn't any easier than I thought it would be. The whole idea of trying to compare people who have actual defensive numbers to people who don't is one big fat pain. I am so glad I didn't try to do this while I was still sick.

So thanks again to everyone, now I've got to look at Kevin Appier. - Brock

Brock,
MS Access does directly import .csv files as Tables with null fields at the end of each record, regardless of whether the .csv records end in the complementary number of commas. Try it with the table Retrosheet IDs and convince yourself. At least, this works if there is a full-length record at or near the top, such as a complete set of field names or a first substantial record that is complete, Aardsma in the IDs table.
>>
LAST, FIRST, ID, DEBUT
Aardsma, David, aardd001, 4/6/2004
...
Abadie, John, abadj101
Abbaticchio, Ed, abbae101
Abbey, Bert, abbeb101
Abbey, Charlie, abbec101
<<
(I don't know whether the retrosoftware does generate .csv files with variable numbers of values in each record, such as this one. This afternoon I did import 193276 events for 2009 without glitch.)

unhappy to be awake now, i'll try to correct that, must stop typing

Thread?

re: DH value, relative to other positions

I think the best (well, maybe I should caveat with "a very good") way to look at it is pretending you are playing a simulation game. Let's say you have to draft a team of hitters, directly against 30 other HoMies. In this sim, you have people available such as (numbers for example purpose only, please don't argue with vaules assigned)

player ...... pos def rating (runs per year +/-)
W McCovey 1B ... 0
T Williams .. LF .. -8
M Ramirez . LF .. -15
E Martinez .. DH .. X

Obvious pt: Ted is 7 runs per year better than Manny in LF. But what is Manny's value on defense relative to Edgar? ANYONE is allowed to play DH in this sim. So, Manny and Edgar are exactly even when it comes to being a DH. The fact that Manny can play a lousy LF; better then Edgar (maybe the SIM allows Edgar to try to play LF with a -30 rating?); makes Manny more valuable by some small amount. If you work 1,000 drafts and Manny is a DH 98% of the time (& if Edgar winds up trying to play 1B or LF very poorly 2% of the time), Manny's value over Edgar in this draft set up is about 0.3 runs (15 times .02).

If the sim accounts for non-DH hitters swinging more poorly when they get put in the DH spot by penalizing them 3 runs per year in that scenario, than Edgar woudl actually go FIRST over Manny in the draft, if they were equal hitters.

Obviously, for those who value plausible "whatifs" in addition to "actual", the fact that E Martinez could have played other spots than DH, and Manny could have DH'ed far more than playing LF in other plausible worlds, also comes into account.

I conclude that a career DH can potentially have more value than the worst 1B/LFer, IF one accounts for the harder-to-hit-well-when-DH effect. I also conclude that the worst LF/1Bers should have their rotten defense cost capped at the DH level, by sliding the scale of rotten defenders such that the Mannys of the world are penalized more than a DH is.

Of course a career DH can "potentially have more value than the worst 1B/LFer," WITHOUT accounting for the harder-to-hit-well-when-DH effect...all he has to do is hit the snot out of the ball. If Teddy Ballgame had been a career DH, he'd still be an inner circle Hall of Famer just for the raw batting wins above average....

the ballot thread will be posted December 21 (the election will end on January 4).

the ballot thread will be posted December 21 (the election will end on January 4).

Hey, I didn't say what time I would post it, DL. :-)

(numbers for example purpose only, please don't argue with values assigned)

Well, I'm glad I went back and re-read that. The problem with scanning through a comment thread is sometimes something like examples are posted and you end up responding to the made up example as I ALMOST just ended up doing. I was going to rant on about the defensive difference not being as dramatic as posted. We see a player screw up and it is played repeatedly on SportsCenter for laughs and we reach a conclusion that the player in question is costing his team a lot with his lousy defense. I just don't think there is an incredible difference between an average OF and a "bad one." Especially, now with the DH position available where you can hide a masher. In other words, Manny Ramirez does some goofy stuff in the field but even for him the goofy stuff is the exception not the norm. There is a marked difference between a great OF (Gutierrez in Seattle for example) and a bad one, but not so much between an average one and a bad one.

Alomar
Jack Morris
Blyleven
McGwire
Edgar Martinez

msilva177:

If you post that on this thread, we won't count it. Anything you want counted needs to be on the other thread.

But - why do you hate shortstops? In particular, what's wrong with either Larkin or Trammell?

136. TomH Posted: December 21, 2009 at 11:28 AM (#3418594)
re: DH value, relative to other positions

[...]
player ...... pos def rating (runs per year +/-)
W McCovey 1B ... 0
T Williams .. LF .. -8
M Ramirez . LF .. -15
E Martinez .. DH .. X

[... modeled in terms of a fantasy league draft] If you work 1,000 drafts and Manny is a DH 98% of the time (& if Edgar winds up trying to play 1B or LF very poorly 2% of the time), Manny's value over Edgar in this draft set up is about 0.3 runs (15 times .02).

... I conclude that a career DH can potentially have more value than the worst 1B/LFer, IF one accounts for the harder-to-hit-well-when-DH effect. I also conclude that the worst LF/1Bers should have their rotten defense cost capped at the DH level, by
sliding the scale of rotten defenders such that the Mannys of the world are penalized more than a DH is.

1.
IIUC the last clause should be truncating, or compressing, or
"sliding the scale of rotten defenders such that the Mannys of the world are not penalized more than a DH is."

2.
I think TomH means to model Manny Ramirez in LF as a labor market inefficiency (if longterm; an emergency role if shortterm). In the example, that labor market inefficiency is corrected 98% of the time but replicated 2% of the time as a draft inefficiency; the draft is the fantasy league's labor market.

There should be some tension here at the HOM (maybe not for TomH).

Should Manny R and Edgar M be excused for their negative fielding value if they play the field by some market inefficiency? They should be two of the game's 14 regular DHs (we suppose along with TomH). If so, then it is the fault of management when they play the field regularly, not merely in emergency. Doesn't that fit a treatment of "less than zero" seasons that is popular here?

Yes, I could certainly see a strong argument that no position player should ever be credited with fewer WARP than his batting + baserunning wins above average (the DH baseline). That would cap the defensive scale for 1B at -9 and corner OF at -15.

Dan R. - will you look at the argument I've been having with Sugar Bear Blanks on the Edgar thread on the "mainland"? It's all about how good an offensive player Ozzie Smith was in the mid/late 80's (compared to, among others, Hubie Brooks). Am I completely off the deep end on that?

wherebouts?

here

As I write this, Andre Dawson is on 29% of the BBTF Hall of Fame ballots. This is especially surprising to me because there have actually been very few small-hall ballots. By comparison, Edgar Martinez—whose value is roughly identical to Dawson’s using Dan R’s WARP (Edgar at $154 million and Dawson at $156 million in the last version I saw)—is listed on 67% of the current ballots. In my opinion, the vote on Dawson is one instance in which the actual Hall of Fame voters are right, and the BBTF consensus is wrong. For those of you who haven’t voted yet, I’m going to try to convince you to have him on your ballot.

The biggest criticism of Dawson is that he didn’t have a high enough OBP to warrant enshrinement. It’s certainly true that he didn’t have a high OBP. In the 16 years that make up the entirety of his positive value (1977-92), Dawson had a 327 OBP compared to a league average of 330. On the other hand, Dawson was one of the best sluggers of his era. His SLG was a full 100 points above league average over those 16 years (491 compared to 391 lg avg). In spite of playing most of his prime in a ballpark that suppressed power, he was in the top 10 in home runs 9 times and in the top 10 in slugging 8 times. Of course his middling OBP is a detriment to his case, but it was less important when he played, and NOT because managers didn’t care about it. Dawson played in a low-offense environment for most of his career. According to B-Ref, his AIR (offensive environment, with average = 100) was 96.7 for 1977-92 and 95.0 for his 1980-83 peak. OBP is more important relative to SLG in high-offense conditions, where outs are more costly. When runs are scarce, extra bases gain value relative to the cost of an out. I am not trying to say that he was the ideal hitter for his era. Rather, I’m saying that his 124 OPS+ in his 16-year prime does not vastly overstate his offensive contributions due to a SLG/OBP imbalance as some have argued.

The other reason I believe many people undervalue Dawson is that they see him as a corner outfielder and use that classification to judge his candidacy. The fact that he ended up with more innings in RF should not take away from his initial 7 seasons as a CF. His entire peak and most of his value came as a CF. Whether you think Dawson is unqualified, borderline, or clearly qualified for the Hall of Fame should depend on your evaluation of his CF defense. The defensive statistics available do not completely agree on this issue, though Dawson comes out somewhere between average and elite depending on what you trust (unlike Murphy, who is somewhere between horrific and below-average). FRAA says he’s average, and so does DRA in a limited sample. Another advanced system for analyzing range, Tango’s WOWY, pegs his annual CF defensive value at +28 plays by park, +28 plays by hitters, and +13 plays by pitchers (Posts 59 & 65 at http://www.insidethebook.com/ee/index.php/site/comments/hardball_times_annual_2008_starring/) At .8 runs per play, that’s +10 runs per season at the low end and +18 runs using the average. In addition, John Walsh’s arm ratings have Dawson at +35 runs total in CF, or +5 runs per season. Putting the numbers together, that would make him approximately a +15 to +25 defensive CF annually. As an average CF, I think Dawson would be borderline by the Hall of Fame standards. If he’s +15, I think he’d have to be comfortably in.

Dawson was also good baserunner, with more than 300 steals at a 74% success rate, and he had quite a long career (roughly 10,000 PA in his 16 useful years). His only significant flaw as a candidate is his mediocre OBP. Meanwhile, Edgar Martinez is doing very well based primarily on his high OBP, with significant flaws elsewhere. Compared to Martinez, Dawson was much worse at getting on base, but he had much more value as a defender and as a baserunner, plus an edge in career length and a slight edge in slugging relative to league. The one thing that Martinez did better is very important, but it doesn’t explain the difference in the balloting. Dawson is being overlooked.

You know why I didn't vote for Dawson? Because catch-all stats are hugely unreliable, and defensive statistics hugely unconvincing, and on the whole his case rests almost entirely on those two things. Since I don't buy either one, I don't buy Dawson as a HOFer.

#143; as usual, Paul does a fine job using English to artfully explain someone else's (in this case, Tom the statistician's) attempt to use words instead of their natural language of data.

defensive statistics hugely unconvincing, and on the whole his case rests almost entirely on those two things.

Dawson's case rests on his having been a good centerfielder. I understand that one shouldn't look at BPro's numbers, take them at face value, and be done with it. But defense matters and there's plenty of evidence that Dawson was a good to excellent defensive centerfielder.

Contemporary opinion? Dawson won 8 Gold Gloves. AROM's Total Zone has him at +10/yr in his CF years. DGLM laid out the rest of the case.

To dismiss his defense because you don't trust defensive statistics seems awfully close to saying that since we can't measure defense well, we can't use it in a Hall-of-Fame case at all. Which seems to be slipping down a slope that says that Ozzie Smith doesn't belong in the Hall of Fame.

#101, but who cares? Seriously. It's one thing to ask whether he was the best 1B in the game. I can see how that would be useful. But to care about whether he was viewed as the best 1B in the game? Of what relevance is that?

I don't care. I was responding to Pretoric's comments which cited contemporary opinion. More relevant is the point he was making about dominance relative to contemporaries re: Blyleven and McGriff. For me, that's McGriff's biggest problem, not a point in his favor. His statistical profile is borderline, but he is lower down on a list of 1Bmen from his time period. I like both Clark and McGwire more. I like Thomas more but he only half counts. I like Bagwell more, but he's not as much of a contemporary. I'm not really sure.

Mike Emeigh:

Well, since Brock mentioned it:

I need to repost the study data that I have for Blyleven, but the basic issue with him is that he did very poorly when handed a lead, especially early in the game, compared to other HOF pitchers of his era. My expectation from an HOF starter is that when you give him enough runs to put him him in front, he stays there, and Blyleven didn't do that. (Neither did Morris, FWIW.)

The problem is that, even assuming arguendo this is a skill, Blyleven is far enough above the line as to render this utterly meaningless.

For McGwire, my belief is that steroid usage kept him in the game when he very likely would have been forced out of it because he couldn't stay healthy. He doesn't have a HOF resume if you don't include 1996-1999, and I think it's better than 50-50 that he wouldn't have made it through those years.

So steroids (1) help foot and other injury problems, but (2) only between the years of 1996-1999. I think I'll pass signing on to that logic.

Are we going to have to call in Rich Lederer to beat up on you?

First fact about Blyleven: his actual W-L records are considerably below the records a pitcher of his effectiveness should have had. That leaves us divided into two positions here in HoM threads: is actual evidence of pitching effectiveness (ERA+ or RA+, or DERA, or FIP), combined with usage (IP) the only criteria that should be used in judging him? Or is there something about that W-L record that should cause us to value him less? (Call the latter, if taken to extremes, the MWE position.)

You said, "I would rank him below era contemporaries like Seaver, Carlton, Palmer, Gaylord Perry, Don Sutton, Fergie Jenkins, Phil Niekro, Nolan Ryan, Bob Gibson, Marichal." The thing is, we don't rank him below that whole list, or even most of it. In our ranking vote for the 1960-1990 pitchers, the order went:

1. Seaver (that was unanimous; nothing else was unanimous)
2. Carlton
3. Gibson
4. Niekro
5. Perry
6. Blyleven
7. Palmer
8. Jenkins
9. Marichal
10. Ryan
(Sutton was several places further back, behind Koufax, Wilhelm, Drysdale, Bunning, and Saberhagen).

And the thing is - I take that vote order as evidence that we did NOT adopt the first position (ERA+ or DERA being all the matters) but rather explicitly downgraded him for the actual W-L record in some way; if we hadn't we should have at least had him ahead of Niekro and Perry.

Then you name a whole bunch of pitchers on a next lest. Let's just take one of them: Catfish Hunter.

Here are my seasonal RA+ equivalent records for Hunter for a 10-year period, 1967-1976:

17-12, 12-14, 15-13, 14-15, 16-14, 21-11, 15-13, 23-12, 24-13, 17-16, for a total of 173-135.

There are three legitimately great years in there in 1972, 1974, and 1975 - but the other 7 years out of the 10 reveal that most of Hunter's value was tied up in being a near-average innings eater.

Now here's an overlapping seasonal RA+ equivalent records for Blyleven for 10 years, 1970-1979:

10-8, 19-12, 19-13, 24-12, 19-12, 19-11, 20-13, 17-9, 16-11, 15-12, for a total of 179-113.

While it's not as many total innings in the 10 years as Hunter, that's a whole lot better than average for many more years than Hunter has, and there's a very large difference between 173-135 and 179-113.

But that was pretty much it for Hunter - the outer five years of his career don't add much. Whereas Blyleven has a whole other peak from 1984 through 1989: equivalent 18-9, 20-13, 17-13. 16-13, 9-14, 18-9.

To use CYA and All-Star votes against him is double jeopardy. You're not only discounting him because his actual W-L was worse than his equivalent record, you're doing it twice - he didn't get those CY votes because he didn't have the flashy W-L records. (And the lack of All-Star selections is, in some cases, just weird.)

Here's a question: was Tom Glavine dominant enough in his own era for you?

Mike Emeigh, that's one peculiar way to look at things. What about pitchers who give up 1 run in 7 innings, while the opposing team scores 1 or 2 runs over that span, but then the pitcher's team scores 2 in the ninth to win the game? Did that SP do nothing for his club??

Look, the argument that the distribution of a pitcher's runs allowed is significant is obviously a valid one. Clearly, a P who gives up 91 runs in one start and 1 run in each of the other 29 starts is more valuable than a P who gives up 4 runs in each start. But focusing exclusively on situations when a P has a lead, while ignoring all other situations, seems nothing short of crazy to me.

Yes.

Here's a question: was Tom Glavine dominant enough in his own era for you?

As I recall from Mike's work, Glavine showed up pretty poorly on his list w/r/t "protecting leads." If I'm right about that, does Mike support Glavine for the HOF?

The argument for McGriff comes down to, other guys cheated. Why didn't you say so?

I have to put McGriff ahead of those three, and I swear to God I think he would be ahead of those three if they had not been artificially enhanced, becaue I think McGwire would have broken apart from injuries like Dave Kingman

Here's a newsflash: McGwire did break apart from injuries.

It's really bizarre how people conclude that steroids are some magic elixir that Healed McGwire, even in the face of the obvious: if McGwire was using them they did not prevent injuries.

Agreed, Ray. I think McGwire was using since early in his college days and, I dare say, I would guess they contributed to his poor health. How much they helped him powerwise, I have no idea.

I'm not Mike but I would absolutely say two things are true about Glavine: he's far below Pedro, Maddux, Clemens and Johnson (of his peers). But he's also well above the line for being a HOFer. If you have to be of the caliber of those four pitchers to get in, it would be a tiny, tiny hall.

Of course, I'd put Blyleven in despite Mike's data. So, maybe I'm a bigger hall guy than I think I am.

sunnyday2 said:

I meant that McGriff IS a slippery slope. There's an awful lot of good 1B out there. If McGriff, then Boog Powell, e.g. And Steve Garvey has a good claim to have been the best 1B for a stretch of years, too.

Hey, I voted for McGriff. The slope isn't really all that slippery. Someone (or some 10) has to be the weakest candidate that you'll vote for. And someone else the strongest candidate that you won't vote for. I voted for McGriff, but I'm not going to invest much if any energy trying to persuade anyone else to go along. It's not like he's Blyleven or Raines and stunningly obviously qualified. Similarly, I didn't vote for Dawson. Again, I'm not going to invest much energy into arguing that point, either. And I wouldn't vote for Boog Powell or Steve Garvey, even though there are numerous significantly worse players in the HoF.

bunyon said about Glavine:

... he's far below Pedro, Maddux, Clemens and Johnson (of his peers). But he's also well above the line for being a HOFer.

Which is exactly why I dragged him into the conversation.

I had forgotten that I had posted in this thread.

I want to make it clear that I wasn't attempting to be snarky or to brush aside Emeigh's arguments. I know full well who is he and where he stands in the sabr community. I mean no disrespect. All of the questions I asked where meant in good faith as questions to which I was genuinely curious of the answer. There was no rhetoric intended there.

And honestly I don't really understand Brock's post and what "book" he is talking about so I'll just leave that alone.

The other reason I believe many people undervalue Dawson is that they see him as a corner outfielder and use that classification to judge his candidacy. The fact that he ended up with more innings in RF should not take away from his initial 7 seasons as a CF. His entire peak and most of his value came as a CF. Whether you think Dawson is unqualified, borderline, or clearly qualified for the Hall of Fame should depend on your evaluation of his CF defense.

As one of those who did not vote for Dawson, I will say that I still consider him borderline and wouldn't be horrified if he were to be elected. If I viewed him solely as a corner OF, a very good defensive one, I wouldn't consider him even in the discussion.

His peak is very short in the CF years... we're really talking about only 4 seasons (one shortened) as an elite offensive CF, and that's 1980-1983. His 9000-ish innings are closer to 6 seasons than 7 as a CF, and it isn't exceptional enough for his RF years.

In terms of peak, I find Dawson to compare unfavorably to guys like Bernie Williams and Jim Edmonds, who are career CF. He's in the ballpark with Kirby Puckett, but in 75% of the innings in CF (and I'm not a huge fan of Puckett as a HOFer). In terms of career, Dawson is somewhat of an accumulator. His last four seasons were below average offensively, and his last "great" season was probably 1990, with 6 more seasons to pad his counting numbers.

@38:

Like I said - I don't think I'd vote for him - but I think Morris supporters miss the mark entirely when they try to highlight a single game or claim he was a 'big game' pitcher. Morris' best case isn't one of isolated incidents of brilliance -- it's that he was consistently healthy and good (but never really great) for an extended period of time, and for duration of his career, not a lot of pitchers can make the same claim.

Except, Jack Morris wasn't even that. Let's take what's probably the lowest reasonable definition of a "good" season for a starting pitcher. Let's not make it particularly difficult. In fact, let's keep things simple and make the criteria easy to meet. No 5 WAR seasons. No iron man 270 IP requirements. Let's go with 200 innings pitched, with an ERA above the league average. That's all. All Morris has to do to get credit for a "good" season is throw a--historically--fairly low number of innings for a starter, and all he has to do in pitching those innings is post an ERA above the league average. That's it. He can be 1/100th of a run better than average, to be considered "good" for the sake of this inquiry. Morris pitched in 18 major league seasons. In how many of those seasons did this potential Hall of Famer meet those two simple, low standards? 12? 14? More? In actuality Morris failed to meet this basic, easy-to-meet definition of "good" in over half the seasons he pitched major league baseball. He was "good" in only 7 of 18 seasons, and that's with setting the definition for "good" as low as I think anyone would accept it.

Jack Morris was durable. I'll give you that. But he wasn't all that good. In fact, the majority of the time, he wasn't even turning in good seasons.

Dawson's case rests on his having been a good centerfielder. I understand that one shouldn't look at BPro's numbers, take them at face value, and be done with it. But defense matters and there's plenty of evidence that Dawson was a good to excellent defensive centerfielder.

Contemporary opinion? Dawson won 8 Gold Gloves. AROM's Total Zone has him at +10/yr in his CF years. DGLM laid out the rest of the case.

To dismiss his defense because you don't trust defensive statistics seems awfully close to saying that since we can't measure defense well, we can't use it in a Hall-of-Fame case at all. Which seems to be slipping down a slope that says that Ozzie Smith doesn't belong in the Hall of Fame.

While there are some valid points here, I think that trying to measure defense for players of this variety actually makes the situation worse rather than better -- it's not that there's noise in the data, it's that the data is almost all noise. When it comes to Dawson in particular, I look at it this way:

+ Because he used such an exceptional number of outs, his defensive output would have to be very, very exceptional to cross my personal line.
+ Because he's a CF and not a 2B or SS, the standard is going to be even higher. CF is a glove position, but not that much more than 3B.
+ Regardless of the reasons for it, he was moved off CF halfway through his career. I realize that this was not because of limited ability but because of limited durability, but again it severely impacts his value long-term.

Almost every case I see made for Dawson relies heavily on WAR data or something similar, which just isn't convincing to me for a variety of reasons, but largely because the defensive element of it is junk. I don't mean to disrespect the people who put a lot of time and effort into these statistics, but I have always objected to them because they presume that defense can be measured with a degree of accuracy that it just cannot yet, and certainly could not 30, 40, 50 years ago. Dawson's case relies not only on his having been a good CF, but having been a really great one, at least as far as I'm concerned, and I'm not convinced. It's not even necessary to cite the Jeter example to argue that a string of Gold Gloves isn't really that meaningful; the award is often given out repeatedly to the same players long after it should be out of, for lack of a better word, inertia.

To me, the HoF case for Dawson assumes facts not in evidence. When a player is borderline, relying on quantifications of his defensive value to get him over the line is usually going to leave him short for me, especially for a guy who plays CF rather than 2B, SS, or C, and especialy for a guy who played half of his career at another, far easier position. I realize that to some degree this may result in a personal HoF that is bat-heavy, but that's a price I'm willing to pay.