Where oh where in the 1910s were all those 2Bs who hit like corner OF? In the NL, there was Larry Doyle, and...and...gimme a minute, they'll come to me. Sure you could find good freely available talent and they could field as well or a little better than Larry Doyle. But they hit like, well, freely available talent. They didn't hit anything like Larry Doyle.

More recently, you might find a Carney Lansford to be freely available, but that doesn't make it any easier to find a George Brett or a Wade Boggs.

defensible logic THIS:

Stephens leads in career WS, 265 (192 batting/running, 73 on def) to 231 (133 bat, 97 def).
Stephens played 1720 games to Scooter's 1661.

Per 154 games, then, we have
S.S. offense defense total
Vern . 17.2 ... 6.5 ... 23.7
Phil ... 12.3 ... 9.0 ... 21.3

Add in war credit, and Rizzuto's longer career and nice peak, and yes, I see him ahead, but it's certainly defensible to look at the above and conclude Vern was better.


I would argue that its indefensible not to look at War Credit for a guy who missed his age 25, 26, and 27 seasons to the war (and effectively lost his age 28 season to malaria, but that's more debatable). My point is that any voter who has Stephens sniffing his ballot and who believes in giving war credit, should definately have Rizzuto on his ballot ( I had forgotton that karlmangus just hates Yankees, I should have remembered from the assorted Sox Therapy threads).

I also think that, and no poster has addressed this concern, that we need to consider that Rizzuto was a awesome defender at shortstop. Contemporary opinion had him as a great defender, and BP's stats have him at +20. I loathe BP's defensive stats as much as the next guy, but that's what we have to go on.

My last piece of evidence is embarassingly anecdotal, but if anyone would know, its someone who played with Stephens and against Rizzuto:

"When Ted Williams was holding court at Fenway Park several years ago he was asked why the Red Sox won only one pennant during his tenure in Boston. Ted replied that if the Sox had had a shortstop like Rizzuto they would have won many pennants."

Good points, 'zop; altho I'd add that Ted also claimed Rizzuto was the #2 shortstop of all time, behind Wagner.

DL from MN--As I said on my ballot, Newcombe definitely pitched in an era that had a very low stdev of RA+ (which makes Robin Roberts and Warren Spahn even more impressive, by the way). But you can't correct for stdev of RA+ without also correcting for IP norms, and I'm still working on that. As soon as I have a draft of pitcher WARP I'll post my findings, but it's a long and complicated process.

All--I am one of Rizzuto's best friends, while Stephens isn't close to my ballot, but Stephens was clearly a superior player in 1949. I have Stephens at 6.9 WARP2 (behind Teddy Ballgame and Eddie Joost in the AL) and Rizzuto at 3.9. As for the rest of their careers, I'll take Rizzuto in a landslide, thanks. By the way, those 1949 Red Sox had INCREDIBLE position players--by my measure, five of the top 8 position players in the AL were Red Sox! (Williams, Stephens, Doerr, Pesky, and Dom DiMaggio). Plus they got terrific pitching seasons from Parnell and Kinder...Lost the pennant by one game to a Yankees team that seems pretty inferior to me (two games worse by Pythagoras).

TomH, it is only "defensible to look at the above and conclude Vern was better" if you are a robotic WS voter who doesn't recognize *any* of its flaws (such as systematic underweighting of defense) OR give war credit.

Mike Green, yes, I would say that the best evidence we have suggests that Nettles at his absolute best was better than Brooks at his absolute best. That said, I definitely consider Brooks a greater fielder overall, since he maintained his excellence over a much longer period of time.

Kenn, I have Puckett's fielding as marginally better than Murphy's--2.4 wins better over their careers. Puckett's sudden 1986 power surge lines up perfectly with the collapse in his defensive statistics, denying him of a true two-way player's peak (and serving as circumstantial evidence of PED's, for whatever that's worth which is nothing). Puckett's high putout totals from 1986 onwards are *clearly* the result of a low-K, high-flyball pitching staff. According to Michael Humphreys, "CF putouts as a percentage of total outfield putouts were almost exactly average throughout Kirby's career." Unless he was uniformly flanked by superb corner outfielders as well, his high range factors (which may have had something to do with his good fielding reputation) are a mirage.

I don't know if it got easier to play CF between 1980 and 1990. What I do know is that it was easier to find a good-hitting, average-fielding CF in 1990 than it was in 1980, which makes Kirby less valuable in his context than Dawson was in his. If you didn't have Kirby, you could have Lloyd Moseby. If you didn't have Dawson, you were stuck with Rowland Office.

Last time I checked, Doyle was a 2B not a 3B...if you are looking for 3B there are tons of deserving candidates! McGraw for peak and Nettles for career leap to mind. Anyways, the point was that Doyle shouldn't count as a "glove," not just because he fielded his position poorly but because his position was really offense-first when he played.

All--I am one of Rizzuto's best friends, while Stephens isn't close to my ballot, but Stephens was clearly a superior player in 1949. I have Stephens at 6.9 WARP2 (behind Teddy Ballgame and Eddie Joost in the AL) and Rizzuto at 3.9. As for the rest of their careers, I'll take Rizzuto in a landslide, thanks.

Dan-- My point isn't that Rizzuto was "better" than Stephens in '49. Rather, its that in a season when Rizzuto was so much inferior to Stephens offensively, contemporary observers still thought he was more valuable. This is a very good sign that Rizzuto had huge advantages in other facets of the game. Think of Ivan Rodriguez's 1999 MVP as an analogy; it was a mistake, but it was a surefire indicator that there was more to Pudge as a player than his offensive stats.

Shoot, yeah, Doyle was 2B. Don't know why he's listed at 3B on my compiled spreadsheet. He's listed correctly as a 2B on the original. I actually like Nettles quite a bit, though some pitchers probably go on ballot ahead of him, if I were to drop Puckett or any other OF. McGraw is too peaky for my tastes. On that subject, what's the knock on Cey, relative to Nettles?

Thanks for your take on Puckett's defensive environment. That's interesting, and indeed would result in a much smaller gap between Murphy and Puckett than otherwise, even if I still give some credence to reputation. By the way, what is Humphrey's baseline for "average" fly ball distribution between CF and corners - average distribution over a particular range of years? Most of the shift I see happens early in the century, then very gradually after that.

Also, I'd been wondering about Stephens, so that ongoing discussion is interesting to read.

> I'm confused. Aren't park and league incorporated into ERA+?

You're not confused, I am.

Sunnyday, I should have said 1B rather than OF to be precise, since back then OF was a little deeper than 1B. But to answer your question, there were plenty. Just look at, say, the 1911 NL. The starters at 2B were (from worst to best) Dick Egan, Otto Knabe, Dots Miller, John Hummel, Heinie Zimmerman, Bill Sweeney, Miller Huggins, and Larry Doyle. Egan, Knabe, and Miller were all excellent fielders, the rest carried legit bats. Not one of those guys played at worse than half a win below the overall league average (hitting + fielding) on a rate basis. By contrast, at 1B you had the carcass of Fred Tenney and the rookie season of Vic Saier, which both combined poor hitting with poor fielding. On the whole, NL 1B averaged 1.2 wins above overall league average per season that year, while NL 2B averaged 1.5 wins above overall league average. The gap was even starker just looking at the worst three regulars: 0.2 wins *above* overall league average for the 2B, 1.1 wins *below* overall league average for the 1B.

And don't accuse me of selectively picking my data points--look up the numbers yourself before you make empirically incorrect factual assertions. For the entire 1906 to 1915 peiod, 1B and 2B had the exact same positional average EqA (.272). Their worst-three-regulars averages were also roughly equal over that time. In Doyle's era, 2B was a right-spectrum position, a bit tougher than the outfield but equal to 1B and faaaar easier than 3B, SS, or C. There's simply no rational way to spin the data from those years that yields any other result.

Kenn, I like Cey plenty and used to vote for him. His peak/prime is a dead ringer for Nettles's (albeit with a different shape; more hitting and less fielding). But Nettles just outdistances him on career by a decent margin with his filler seasons at the beginning at the end. They're not worth much, but they add up and make the difference for me. If you exclusively vote on best X years where X is less than 11, Nettles and Cey are peas of a pod.

Just email Michael Humphreys to ask him how he calculates that. He's a BBTF member.

I'm confused. Aren't park and league incorporated into ERA+?

Yes, David.

yes, park and league are incorporated in ERA+.

>Shoot, yeah, Doyle was 2B. Don't know why he's listed at 3B on my compiled spreadsheet. He's listed correctly as a 2B on the original.

Actually I have Doyle on my list with Nettles...and Elliott and Cey and so on. I list the old-time 2B and the modern 3B as "hybrids," players who were expected to hit the ball and also to play a reasonably demanding defensive position. Bill Monroe is on this list too and I'd guess that Bill Madlock is a pretty good comp.

I list old-time 3B (Williamson, McGraw, Traynor) on a list with modern-day 2Bs ("gloves") like (Randolph) and (Whitaker) who are HoM/not PHoM and also Bobby Avila, Red Schoendienst, etc. It works for me in trying to remember that the positions have changed.

I finally have an internet connection in my new apartment and managed to go through the numbers for this week's new candidates. Clark tops my ballot and the Evans' bump up just enough to leave my PHOM backlog. I have also revised (slightly) my weighting of pitchers (more) and glove(not as much) to bat, which results in a mini-shuffle in the second part of my ballot. Biggest surprise - Dick Redding makes my ballot for the first time in 25+ elections.
1)Will Clark - His snub by the BBWAA shocked me. Not that I expected him to be elected, but a first year rejection of such magnitude was unthinkable. He belongs. (PHOM)
2)Hugh Duffy (PHOM)
3)Ben Taylor (PHOM)
4)Tommy Bridges (PHOM)
5)Kirby Puckett (PHOM)
6)Lefty Gomez (PHOM)
7)Bus Clarkson (PHOM)
8)Dale Murphy (PHOM)
9)Vern Stephens (PHOM)
10)Gavvy Cravath (PHOM)
11)Bob Johnson (PHOM)
12)Tony Oliva (PHOM)
13)Alejandro Oms (PHOM)
14)Dick Redding (PHOM)
((14a)Darrell Evans)) (PHOM)
15)Bobby Veach (PHOM)
((15a)Dwight Evans)) (PHOM)
16)Bret Saberhagen
17)Dizzy Dean
18)Orlando Cepeda (PHOM)
19)Reggie Smith
20)Al Oliver
((20a)Andre Dawson))
21)Albert Belle - Fits in rather nicely with the next two on this list.
22)Jack Clark
23)Jim Rice
24)Wally Berger
25)Don Mattingly
26)Dan Quisenberry
27)Lee Smith
28)Bruce Sutter
29)Ernie Lombardi
((29a)Jimmy Wynn))
30)Ron Guidry

that we need to consider that Rizzuto was a awesome defender at shortstop. Contemporary opinion had him as a great defender, and BP's stats have him at +20. I loathe BP's defensive stats as much as the next guy, but that's what we have to go on.


As I described in the Mazeroski thread a couple of weeks ago, I've been working on a system of estimating the number of chances a position has in a given season, and how many of those are converted into outs, then calculating a Dial ZR +/- value.

It's still a bit cumbersome to work with, so but at the weekend I looked at the Yankees at shortstop during Phil Rizzuto's career. I'm not sure I'd characterize him as an "awesome" defender. Relative to his league, there was more often than not a better shortstop position elsewhere. However, during his first two seasons, 1941 and 1942, he was definitely in the "awesome" category. He had a third top quality season in 1950. I was very impressed with his career, which I've looked at before (check out his thread for my Brock2 calculations and ensuing discussion) and once you give him war credit his career value overall is clearly better than Albert Belle's.

However, in doing this analysis, I've come to the conclusion that sabermetrics doesn't have a very good understanding of the value of defensive play. This of course has implications for rating players for the Hall of Merit. Certainly for infielders, the range of what is acceptable is quite narrow. I wonder if we see here a cousin of the effect that Mike Emeigh noted related to BABIP - that guys who can't field (or control balls-in-play) are weeded out before they get to the majors.

Exceptional defensive play probably deserves to be looked at more as a 'geometric' (2, 4, 8, 16, 32) progression rather than a linear (1, 2, 3, 4) one. Rizzuto's 1941, 1942 and 1950 seasons defensively were indeed exceptional for a shortstop in his league and era. I would recommend ignoring defensive value as a method for evaluating merit except at the extremes of the spectrum, when it should be weighted quite heavily.

>I would recommend ignoring defensive value as a method for evaluating merit except at the extremes of the spectrum, when it should be weighted quite heavily.

Interesting. I tend to do this for the corners, e.g. I mean, c'mon, look at Frank Howard's batting numbers. Clearly meritorious. Why should I care about the rest--Frank didn't ;-) More to the point, his managers didn't.

Gosh, I really disagree. A really bad fielder at LF/RF/1B like Howard can easily cost a team 15 runs, taking a big bite out of their total value. Then again, Howard *is* an extreme.

The key issue with fra paolo's claim is this:

However, in doing this analysis, I've come to the conclusion that sabermetrics doesn't have a very good understanding of the value of defensive play.

Certainly stastical analysis is less good at interpreting the value of fielding than of hitting or pitching. However, statistical analysis of fielding has advanced greatly over the last ten years. Without some evidence to back up the claims that fielding value varies in a very narrow range and that statistical analysis of fielding value is too unreliable to be trusted in most cases, I can't give credence either to this assertion or to the claim that we should ignore defensive value except in extreme cases.

In any case, even if it is true that fielding value generally varies in a narrow range (a view that is consistent with WS treatment of fielding), we are now looking at players in the backlog whose values are very similar. Even a small advantage in fielding value could significantly affect a backlog player's ranking.

So what is the evidence?

DanR #8
1. Smith played CF for half his career, Johnson didn't.
2. Johnson needs to be docked for the war.
3. Smith has a higher peak.
4. Smith played in much tougher leagues with lower standard deviations.

I can't see a single counterargument for Indian Bob...would his supporters care to share one?

1. regarding Johnson's everyday play, and Smith not, there is the view "I have gleaned here", which I just summarized at "2005 Results" #157
2. Johnson gets 5% boost for 154->162
3. Johnson gets more than one season minor league credit
4. Smith gets no credit for Japan

Numbers 1 and 3 are most important, I guess. By the way, I believe that those factors carry weight where they are not explicit, whereas 2 and 4 depend on deliberation. Regarding 3, when someone's major league career begins so late, and he is immediately a star, it makes him look better, at least in long-distance retrospect. It is natural to feel "something is going on here" (emphasis feel, since my observation is that this is influential even without deliberation).


TomH prelim explanation for Cravath not in top 15
Gavvy Cravath - Yo. Let’s see, Frank Chance hit better, even though Frank didn’t benefit from a Really homer-happy park.

What is the measure here, TomH? Even at four seasons, selected to favor Chance, Cravath wins 160+ to 155+. During their four seasons (10 years apart), Chance played about 125 of 150* games, Cravath about 145 of 154*.
*number of games scheduled


DanG
I'm not really the guy to explain this one, bt my understanding is this is due to the fact that negative win shares are not credited. Even with its super-low replacement level, some players (especially on bad teams) actually deserve <0 win shares. In effect, those guys are rounded up to zero while productive players like Johnson get rounded down to compensate. IOW, there aren't enough wins by the team to credit productive players for what they truly deserve.
42. TomH Posted: October 02, 2007 at 11:56 AM (#2556130)
yup

TomH reported a study, maybe an envelope model, with a different emphasis. (In the Charley Jones thread re 1875 and 1876?) IIRC he looked at teams with Pythagorean expectation or actual W-L something like {.200 .300 .700 .800}, a severely "extremist" selection that leaves even most Bob Johnson teams in the ordinary middle range. At these win rates, runs do not deliver many extra wins.


45. mulder & scully Posted: October 02, 2007 at 12:12 PM (#2556168)
I was doing some experimenting with my system, trying to better recognize defensive excellence. . . . [substance deleted] . . .

C'mon, Kelly. We value sharply worded opinions so don't pull your punches!

So what is the evidence?

Well, using the methodology I described on the Mazeroski thread, which results in a defensive winning percentage being calculated, the best season so far I have seen for a 2b is Bobby Avila's 1954. AL team DWPs at 2b, from best to worst:

.521 (Avila)
.502
.502
.501
.497
.496
.493
.491

One of worst I've seen is Bobby Richardson's 1961. Again, AL team DWPs at 2b:

.5068
.5065
.504
.503
.502
.502
.501
.495
.493
.487 (Richardson)

You get the same clustering in season after season, with most DWPs being within 3 percentage points of .500 either side, and a couple of outliers at each end.

Certainly stastical analysis is less good at interpreting the value of fielding than of hitting or pitching.

I'd prefer to say that whereas statistical analysis probably is accurate at interpreting the value of average fielding, it underestimates the impact of exceptional fielding, whether good or bad. An exceptionally bad glove is a serious disadvantage relative to the average, because it's so easy to find average, and an exceptionally good glove is a considerable advantage, because they are very rare. Hitting seems to offer more opportunities to add value, so there is more of a continuum in quality.

I actually like Nettles quite a bit, though some pitchers probably go on ballot ahead of him, if I were to drop Puckett or any other OF. McGraw is too peaky for my tastes. On that subject, what's the knock on Cey, relative to Nettles?

For Nettles, unlike Cey, there seems to be a telling rhetorical question, "Why don't all the career voters have Nettles high on their ballots?"
Cey has no obvious clientele (wrong term). If you believe 3Bmen have been underrated overall you are more likely to look at times when few 3Bmen have been elected, and see Leach or Elliott.

I suspect there was some learning or sheer emulation. Why else so many relatively good batters suddenly at 3B? Merely within league-decade Cey is head to head with Schmidt and Evans, here elected unanimously and promptly. (Why else? The stock answer is, because the position became absolutely easier to play)

TomH prelim explanation for Cravath not in top 15
Gavvy Cravath - Yo. Let’s see, Frank Chance hit better, even though Frank didn’t benefit from a Really homer-happy park.

What is the measure here, TomH?

OWP. Includes baserunning, very impt in Chance's day.
Unfortunately I don't have OWP ##s right here since I gave my copy of Sinin's BBencyc to a friend.

The measures I've been using offensively are pretty closely related to OWP. How do I have Chance/Cravath?



The kicker, of course, is that this is MLB only and does not show any of Cravath's Minneapolis years or PCL years. And there's the deduction for Chance for missing time during seasons. Considering those factors, it looks pretty close to me.

Paul Wendt, I think the career voters don't have Nettles high on their ballots because a lot of them tend to be straight total WS voters, and since WS understates the value of defense it sells Nettles short (compared to, say, Rusty Staub). Maranville is hurt for the same reason, although I don't support Maranville.

The other issue for Nettles is that there are so many good 3B during his day and he was not Schmidt or Brett or Boggs or Brooksie. He may be the best of the rest but that by itself is faint praise.

Hah, well my contention is that he was Brooksie more or less, of course.

There's something to that. Brooksie, in hindsight, is pretty over-rated. At the time, of course, there was no real precedent for him and so he was regarded appropriately to that. The 2nd Brooksie is doomed to be regarded as the 2nd Brooksie, and even then he had competition for that title. I personally kind of like Robin Ventura.

Oh yeah, if you don't think Brooks should be in the HoM then obviously you won't vote for Nettles. But they are so damn similar that it seems awfully inconsistent to me to have voted Brooks in resoundingly on the second ballot and not have Nettles anywhere close. That seems to me to be a clear case of confusing Fame with Merit.

I remember being underwhelmed by Brooks to a degree (& being surprised by it), but still voting for him. Maybe I gave the reputation too much respect? Could be. But it doesn't inspire me to correct the wrong by voting for Nettles.

OCF, I never got around to thanking you for the Tommy Bridges info. And now I've got another request (although you've probably posted it before). Could we see a 3B Candidate/HoMer list by your RCAA (or whatever it is, I'm working too hard these days) method from post #122? I'd like to see Brooks, Boyer, Elliott, Nettles, Cey and Bando, and anybody else you might think is appropriate. I don't know if Leach or McGraw really fits in such a comparison. I'm trying to see if I'm still comfortable having Cey ahead of Nettles & Elliott.

Thanks in advance.

You guys take rookie votes, right?

1. Will Clark
2. Mike Stanley
3. Whatever else

My two favorite players of all time on the same ballot.

Oh, and my rankings tend to focus partly on Win Shares, partly on grit and hustle, and mostly on blind allegiance. My third vote should actually be for Major Applewhite. No Win Shares, but he's got all the grit, baby.

I would echo that OCF's modified (for run environment) RCAA tool is a very helpful one.

I would also like to remind everyone that I've explained at length before my contention that the technique of comparing players by decscending-order-of-goodness-years is a pile o' hooey that is given too much play.

tom, explain why it's hooey again, please. thanks!

Devin - I'll be away this weekend and might not get to it before then. Maybe next week. I also think I'd like to throw some more recent players into the mix - Matt Williams, Robin Ventura, maybe even Howard Johnson.

pretend we have the ultimate 0-10 player value scale of agreed-upon accuracy.

player A, in chron order, puts up seasons of
4 6 9 7 10 7 2 8 6 5 0 4 2

while player B puts up precisely equal career value years of
4 6 9 7 5 7 5 8 6 5 2 4 2

Obviously, I merely swapped a 0 + 10 for a 5 + 5.

re-ordering by the declining years method shows
pl A: 10 9 8 7 7 6 6 5 . 4 4 2 2 . 0
pl B: . 9 8 7 7 6 6 5 5 . 5 4 4 2 . 2
diff: .. 1 1 1 0 1 0 1 0 -1 0 -2 0 -2
cumdif1 2 3 3 4 4 5 5 .4 4 . 2 2 .0


Now, I am sympatetic to the argument that the 0+10 is worth a little more than the 5+5; better "peak". But by the above, player A looks better by a huge amount. Which I say is utter nonsense. You could replace the 0+10 with a 5+7, and "A" would STILL look a lot better. This system just utterly overvalues peak years and under values rotten years.

Which is basically, IMHO, how Jimmy Wynn got in so easily; many voters ignored his 1971 when he turned a potential pennant contender (7 games out on July 7) into an also ran by putting up an OPS of 597.

Which is basically, IMHO, how Jimmy Wynn got in so easily; many voters ignored his 1971 when he turned a potential pennant contender (7 games out on July 7) into an also ran by putting up an OPS of 597.

I didn't ignore that season, Tom, and hardly gave him any credit for it. But I certainly didn't deduct points off of his career tally, however.

But by the above, player A looks better by a huge amount.

I think I'm among the peakiest of the peaky, and those two look pretty close to me--as you say, player A comes out a bit ahead (and that little bit can mean quite a few ballot spots with our crunched-up backlog these days), but I wouldn't say he's "a lot better."

when I say 'ignored', I mean that some peak/prime voters did not appear to discount his peak/prime; the season got counted as if it were his age 39 swan song or age 22 debut.

It makes no difference unless you care about consecutivity. I don't think many voters do, although obviously there are some.

It makes no difference unless you care about consecutivity. I don't think many voters do, although obviously there are some.

It doesn't to me, Dan, so I appear to differ with Tom on that point.

under values rotten years

Don't care about them. They are still more valuable than not playing at all, unless they are below replacement. Then it's still not the player's fault that management has no replacement option available should a nagging injury occur, management says "Suck it up", and the player proceeds to do so. I would need evidence that the "rotten year" was deliberate before I would penalize a player for it.

Over the weekend, I finished going through Dan R's spreadsheet determining an all-star team for each league in each year 1893-2005. If anyone would like a copy, let me know. I tried to maintain a LF/CF/RF split in the outfield, but there were a few years (Cobb/Speaker or Aaron/FRob/Mays, for example) where I just picked the three best. Other times, there were 3 or 4 players with almost the exact WARP and WARP/season and I listed them all.

I'm also playing around with a Dan R. Gold Glove team spreadsheet.

Rizzuto was the AL gold glover in 1941, 1942, War, 47, 49, 50, 52 and 4th in 46, 5th tied in 48, 2nd in 51, 5th in 53, and 5th in 54.

6 years out of 9 he led AL shortstops in FWAA2.

I'm not sure how much added value there is there, at least for seasons before 1987, since FWAA is just an average of Fielding WS and BP FRAA. Also, the standard deviation adjustment is irrelevant if you are looking at league ranks--if you lead in FWAA1, you lead in FWAA2. That said, I am extremely glad you are finding my data useful and interesting!! Let me know if I can answer any questions or if you happen on anything that seems particularly surprising or notable.

I am heading home from work. I'll post more later.

I wanted to thank Dan R, sunnyday, TomH, zop, Chris Cobb, and everybody else for some great debate over the last week.

I'm with jimd.
Lousy over-the-hill seasons are worthless either way.

You don't cut the rate stats, nor credit the counting stats, basically.

And once a guy gets down to mediocre at that age, it becomes fairly irrelevant. Obviously better than lousy, but at that level it's just management skills on what they have available.

Hello guys, please consider this my ballot. I would be much obliged if someone would copy and paste it into the 2006 ballot thread when voting begins. I will be off to the police academy for a couple of weeks of training and will not be on the computer. Thanks in advance. I'll be back in time for the next vote.

Clark debuts down in the 60's with Tony Perez and Lou Brock. I have Hershiser almost tied with Saberhagen and Cicotte in the 90's. I have Bob Johnson in the 40's, Puckett in the 30's, and Reggie Smith in the 80's. Bus Clarkson is the only NeL player that will recieve any votes from me (without any new data).


1. Gavy Cravath - there's no arguement that he took advantage of his park. i say great for him. no one else on his team was able to do it to the extent that he did. You give the man proper credit, and he should be appealing to the peak and career oriented voters.


2. Lee Smith - I am sure this won't be popular. Someone has to be a fan of the reliever. Might as well be me. I voted for Fingers, Gossage, Wilhelm, and continue to vote for Sutter and Jack Quinn. As far as I'm concerned, he is the Eddie Murray of relievers. Never really stood out, or pops into your mind as a stud, but quite consistent, and for quite a long time. I would have no problem with either the HOM or HOF opening their doors to more relievers.

3. Orlando Cepeda - very consistent for a long time.

4. Bucky Walters - not as much career value as I typcially like, but enough peak to offset for the lack of career.

5. Don Newcombe - if you give him any NeL and Korean War credit, he has quite a case.

6. Bus Clarkson - I am willing to speculate that he was better than Stephens.

7. Vern Stephens - a power hitting, perrennial MVP shortstop is someone I can vote for.

8. Bruce Sutter - basically the same arguement as I have for Walters.

9. Elston Howard - I admit, I missed the boat with him a long time ago. Given proper credit, and my love of catchers, he should've been on my ballot long ago.

10. Chuck Klein - see Cravath

11. Bobby Veach - Tons of credit for pitching in 1918...ok, just kidding. Good career value though. Similar to Cepeda in the sense that he never stood out, but was consistent for a long time.

12. Jack Quinn - giving him credit for PCL play, being a fan of relievers, and a ton of career value makes him a viable candidate for my ballot. if he had been in the majors before the age of 25, I believe he would've got 300 wins and that milestone would've made a huge difference to some voters.

13. Ernie Lombardi - I understand why some people don't like him, I really do. I never imagined that his support would be so small though.

14. Lefty Gomez - he played for great teams...so did Red Ruffing...i'm not penalizing him for that.

15. Johnny Pesky - obviously only makes my ballot with war credit

Not that it's going to matter, since Clark is going to be elected easily and Adam isn't going to revisit his ballot, but

Will Clark in the 60s and Orlando Cepeda at #3?

They played the same position. Clark was a better hitter and a better fielder in a career that was approximately one season shorter. The only advantage Cepeda has is 150 games of playing time.

Both WARP and WS over-value playing time, and they give Clark more career _value_ than Cepeda.

Chuck Klein over Will Clark also leaves me scratching my head. It seems like a pretty straightforward comparison.

I support Clark and will vote for him, but I'm surprised people are talking about him as a virtual no-brainer. He seems to me fit in rather comfortably near the top of the backlog. That might be because position is much more important to me than it is to, say, WS voters, where really career league-adjusted Runs Created are almost all that matter. Clark--and all right-spectrum position players with decently long careers and in-season durability--tend to be their favorites.

Well, it's basically that the backlog is so weak. Depending on a voter's preferences, an individual might have 5 or 10 players ahead of Clark, but he's going to make a lot of ballots, and he's competing against players who are barely making 40% of ballots. All Clark has to do is hit 50%, and he will breeze to election. Back in 1996, we elected Keith Hernandez on his first ballot, and I think it's pretty likely that the electorate will view Will Clark as being at least as good as Keith Hernandez.

From another perspective: no candidate who has done as well in my system as Clark does has not achieved election quite easily. It would take a significant departure from past voting patterns (or a huge error on my part) for Clark not to be elected in 2006.

I agree that Clark will get in, and that he deserves to. But I think that even 15 "years" ago he would have debuted well in the backlog.

"I will be off to the police academy for a couple of weeks of training and will not be on the computer."

My brother is a cop. Hope this isn't the training where they hit you with the taser.

I don't think you're valuing defense properly which is probably helping to lead you toward Orlando Cepeda far ahead of Will Clark and Ernie Lombardi on the ballot.

M&S;,

Not sure my message got through. Please send me the All-Stars listing for DanR's stuff. Just use my btf email. Thanks!

eric

I dunno, W. Clark is clearly superior to all the post-expansion era hitters in the backlog...as hitters, anyway. And I have Frank Howard and Reggie Smith and Cepeda in my PHoM. Clark is better. Of course, Belle was just almost as good as Clark at his peak. If Charley Keller fans don't see anything in Belle, they're not looking hard enough, possibly allowing the character issue to enter in without acknowledging it. I wonder if a group like this 25 years ago would have treated Dick Allen the way we are going to treat Belle here? I'd say, yes.

I don't see enough in Belle - but then, I wasn't a Keller fan. Belle is a "bat" from a time period rich in bats, his career isn't long, and (for me, at least), his peak isn't high enough to overcome that lack of career length - run context has to be figured into his raw stats. I think Clark's peak is high enough, and McGwire's peak is high enough. (Don't want to deal with Sosa just yet, and Ramirez is way too still-active to think about.)

Eric,

Do you want the spreadsheet compatible with Excel 07 or 03?

Kelly

M&S;2003, please!

Thanks again!

eric : )

I see practically no difference between Belle and Kiner.

Eric,

I'll email you when I get home from work.

Kelly

I definitely don't see Belle and Keller as remotely comparable. Keller's 1942-45 is certainly a dead ringer for Belle's 1994-5-6-8, but he blows Belle away in the remaining seasons--his 1941, 1946, 1940, and 1939 were waay better than Belle's 1992, 1993, 1997, and 1999. Kiner is a better comp--his top 7 years are right in line with Belle's--but he too has a bit more padding in seasons 8-10 (1946, 1954, and 1955) than Belle does (1997, 1991, and 1989).

2006 prelim

1. W. Clark--WS is a little too nuts about him, but even taken down a buncha notches, he's better than the backlog, if only slightly.
2. Saberhagen
3. Dean
4. Williamson
5. E. Howard
6. Willis
7. Cravath
8. Tiant
9. Rosen
10. Singleton
11. McGraw
12. Pesky
13. Belle--I was expecting him to end up higher, but the numbers don't put him there. I see him as well below the Kiner/Kellers of the world.
14. Rizzuto
15. Oms

Catcher Gold Gloves by Dan R, 1901 to 1910:

I hate it when you hit the wrong button!!

Catcher Gold Gloves by Dan R, 1901 to 1910:

AL, NL
1901: Lou Criger, Malachi Kittridge
1902: Ossie Schreckengost, 2 way tie with Malachi Kittridge and Bill Bergen
1903: Lou Criger, 2 way tie with Pat Moran and John Warner
1904: Lou Criger, 4 way tie with Bill Bergen, Heinie Peitz, Admiral Schlei, and John Warner
1905: 4 way tie with Criger, Schreckengost, Mike Heydon, and Malachi Kittridge, Red Dooin
1906: Schreckengost, Johnny Kling
1907: Schreckengost, Fred Jacklitsch
1908: 2 way tie with Lou Criger and Bob Schmidt, Johnny Kling
1909: 2 way tie with Ira Thomas and Paddy Livinstone, 2 way tie with George Gibson and Bill Bergen
1910: 2 way tie with Jack Lapp and Ira Thomas, George Gibson

These are going to take forever to post. I'll probably finish the complete gold gloves this weekend. Let me know if anyone wants the file.

I'd like to see the modern guys: 80s and 90s

Just to reiterate: before 1987, my FWAA are nothing more than a BP FRAA/Fielding WS average, so I wouldn't look for too many surprises. After that I have Chris Dial and eventually UZR which does add value. Ideally I'd use DRA, but Michael Humphreys is still developing it and isn't doing it for all player-seasons in MLB history unfortunately.

..which is why I'd like to see the modern guys :)

I just finished the gold gloves for 1987 - 2005. All the information is in spreadsheet and I don't know how to move that information onto the HoM.

Some random notes:

Most valuable defender in any one season? 3 way tie of Terry Pendleton 1989, Jeff Cirillo 2001 and Placido Palanco 2001. All NL 3B. All 2.5 FWAA2.

Number of seasons with 2.0 FWAA2 or better in 19 years? 26.

Who has the most? Robin Ventura 3, 1992, 1998, 1999.

What is the positional breakdown? 17 3b, 4 CF, 3 2b, 2 SS.

What players have the most GG (including ties) at their position?
C: Ivan Rodriguez 6, Brad Ausmus and Charles Johnson 3 each. Each Molina brother has at least one.
1B: Mark Grace 6, John Olerud and Todd Helton 4
2b: Jose Lind and Mark Lemke 4. 2 A's tied in 1991 - Blankenship and Gallego.
3b: Robin Ventura 5, Scott Rolen 4 (add a 5th if you want to allow a 3 way tie in 2001 when he had a 2.3 to the 2.5 of Cirillo and Palanco.) Terry Pendleton, Jeff Cirillo, and Scott Brosius all had 3.
SS: Ozzie Smith 5, Greg Gagne 4, Cal Ripken, Barry Larkin, and Rey Sanchez all had 3.
CF: Devon White, Lance Johnson, Marquis Grissom, D Lewis, and Andruw Jones all had 3. Qualcomm Stadium must have some mojo going on. 5 CF were the best in the NL over 12 years: D Jackson 92, Finley 96, Ruben Rivera 2000, Kotsay 2001 and 03.
LF/RF: Geoff Jenkins 4, Barry Bonds, Larry Walker, Paul O'Neill, and Jermaine Dye all have 3.

Darrin Erstad makes the list 5 times: at 1st in 1999 and 2005, in CF in 2001 and 2002, and LF/CF in 2000.
Devon White makes the list 4 times: 3 in CF and 1 in LF/RF in 1987 for Cal.
Andruw Jones makes the list 4 times: 3 in CF and 1 in LF/RF in 1997.
Placido Palanco makes the list 3 times: at 3rd in 2001 for StL, 2nd in 2003 for Phi, 2nd in 2005 for Det.
Mike Bordick makes the list 3 times: at SS in 1998, 1999 for Bal and 2B in 1992 for Oak
Otis Nixon makes the list 3 times: at CF in 1988 for Mon and 1992 for Atl, and in LF/RF in 1991 for Atl.

If anyone wants the Excel spreadsheet for 1987-2005, post here either 03 or 07, and I'll get in touch. I am having problems with the email that is linked through BTF.

Can't wait for Robin Ventura to get eligible. Dunno if he's a HoMer but I'm thinkin' yes.

Still, what's with all that AA value at 3B? SS have so much more value, you'd think they would also have more differential value. Or is it that there are more baaad 3Bs than SS?

I definitely would not support Ventura.

Again, sunnyday, this is the same issue that came up in our discussion of Doyle: the difficulty of a position to field is not the same as the spread of fielding performance at the position. Your hypothesis as to why that might be the case--that teams will stick awful fielders at corner outfield or third base to get their bats in the lineup--is a very good one. But the point, again, is an empirical one. Bad fielders at right-spectrum positions hurt you just as much as bad ones at left-spectrum positions, and good ones at right-spectrum positions help you just as much as good ones at left-spectrum positions (ignoring 1B and C for now, since their contributions are not primarily tied to range). This is what Win Shares gets wrong, and why I think you are so wrong in largely ignoring Doyle's poor fielding because 2B was an offense-first position, and why I think Frank Howard voters are so misguided as well. The fact that they played easy positions has absolutely no bearing on how much they could hurt their teams in the field-none whatsoever.

Re. Ventura

I'm not sure what you're saying. Doyle was a lousy fielder, we know that. Ventura was a great fielder. Bad fielding is bad. OK. Then great fielding is great, no?

I haven't looked at his numbers yet, BTW, it's just that I keep hearing what a great fielder he was.

Re. Doyle

I'm just saying that his offensive value (big) and defensive value (small) still add up to a good value. e.g. +10 WS vs. the median 2B during his career, and -1 WS on defense, equals +9 WS per season over the median 2B. And sure, the -1 is rounded off, it might be 1.5. That's still a good value.

Yeah, Ventura was definitely a great fielder. And yes, bad fielding is bad, and great fielding is great. You could look at him as similar to Nettles, except he doesn't have nearly as much career. If Ventura had lasted longer I'd support him.

My point is that because Win Shares' standard deviation for fielding is way too low, Doyle should be more like 3 WS below the median, not 1.5. Conversely, the best fielders should be more WS above it than they are.

You never responded to my post showing my methodology for calculating Concepción's Win Shares above positional median. Did you figure out why your results are coming out so differently from mine?

A question not pertinent to any particular candidate:

What, if any, are the sources for official win shares for seasons after 2001? I seem to recall that someone said they are included in the STATS All-Time Major League Handbook. Is this true? Are those stats available anywhere else?

I've been relying on the _Win Shares_ book and the electronic supplement for the lifetime of the project, but those only cover to 2001.

Note that this question is not an endorsement of the win shares metric. I do, however, use win shares as one element in my system, along with WARP1 and WAR, and I would like to keep my system consistent after passing the 2001 line.

I haven't had time to re-check Concepcion yet.

Chris, WS back to 2004 at least are at http://www.hardballtimes.com/thtstats/main/?view=winshares. Dunno about 2002-03.

Chris Cobb, a few questions about your treatment of the shortstops. You have them as Maranville-Bancroft-Concepción-Pesky-Rizzuto-Harrah-Stephens-Bartell-Campaneris.

1. What makes Maranville the best of the batch? Are you using unadjusted WARP1 rather than some adjusted form of it or WARP3 for the BP chunk of your rankings? Maranville's gaudy WARP1 total is entirely due to BP's nuts treatment of prewar infielders. If you buy their take on Rabbit, where's Tommy Corcoran on your ballot? WARP3 takes the wind out of Maranville's sails a good bit, although it still sees him as more comparable to guys like Concepción than I do.

2. Why so unfriendly to Campaneris? Are you making any adjustment to BP WARP or WS to account for his superlative, historically great non-SB baserunning?

3. How do you get them spread so far apart? You have Maranville at #4 and Campaneris at #58. If my system really makes up half of your evaluation, that seems a bit incongruous. Is the separation among them caused by your method of breaking up the backlog into decades? If so, would you mind explaining the procedure in a bit more detail?

Thanks very much.

On a broader note, I feel like the electorate is pretty divided on bats vs. gloves--you're either largely dismissive of positional value as a whole (which would apply to most WS voters) or you have a ballot stacked with left-spectrum position players (people who use my system, many BP voters, and just defense-oriented voters in general). Given how divided the backlog is, if the glove-friendly voters could manage to agree among themselves which of the shortstops are the most deserving, we could probably get one or two into the HoM, whereas if we continue to split our top glove votes among a variety of candidates, those spots will be taken by right-spectrum position players or pitchers whom we all agree are less Meritorious. This isn't a call for strategic voting or collusion--rather, it's a request for discussion within the glove-friendly contingent of the electorate to see if we can reach a semi-consensus ranking of players within the category. I personally see Concepción, Pesky, Rizzuto, Campaneris, and Bancroft as all deserving and fairly closely bunched together in value--too close for my system to conclusively distinguish between them--so I'm wide open to arguments as to their relative Merit.

I'm a WS voter but there's plenty of peripherals worth checking out in migrating from the raw numbers to a final ballot. I mean, if BJ uses a bullshirt dump to translate his own system, well, I sure am going to have one. Of course BJ's raw numbers need a lot of adjustments anyway.

I've revised by prelim based mostly on some pitcher work. Bottom line: Bucky Walters replaces Addie Joss.

Yes, the newbie batters are 1-2 but there's not another pure bat anywhere on my ballot. Sheckard (17a), Cravath (19) and Singleton (24) would be the next 3. But it's true that I have Frank Howard, Reggie Smith, Orlando Cepeda and Don Mattingly in PHoM. They've moved down a bit on the top 50 since I started re-evaluating the gloves, both their rankings among them but also their relative standing vs. the bats. Ironically, Trammell and Whitaker dropped down in that re-eval (and Willie Randolph revealed himself to be an exceptionally weak choice, not as good as Nellie Fox in my view) while Pesky was moving up. The next gloves are Leach (16), Vern Stephens (admittedly a special case, more of a hitter but at a glove position, at 21) and Dick Lundy (22). I do also have Thurman Munson in PHoM.

1. Will Clark (new, PHoM 2006)
2. Albert Belle (new, PHoM 2006)—coupla mashers. WS peaks are beyond question, Clark at 44-37-34 and Belle at 37-34-34-31 (with 1994 and 1995 adjusted appropriately). Clark’s total moves to 341 with adjustments.

3. Dizzy Dean (2-2-3, PHoM 2002)—for a peak voter, the one big oversight of the HoM project to date; 6.5 more WS per year than the median #1 starting pitcher (that is, the “ace” for each team) in the NL over a 6 year period; I haven’t yet found anybody with a better peak than that in the HoM or PHoM backlogs

4. Ed Williamson (3-12-6, PHoM 1924)—more peak and more glove than any other available IF

5. Elston Howard (5-5-5, PHoM 1994)
6. Don Newcombe (6-7-9, PHoM 1997)—these 2 guys missed more opportunities than anybody; Newcombe coulda been Robin Roberts, for Howard the comp is Mickey Cochrane; Newk was 3 WS per year better than the median “ace” for 7 prime seasons, vs. Saberhagen, for one, also at +3 and Stieb at +2.5

7. Tommy Bond (8-11-12, PHoM 1929)—he’s baaaack; pretty much the all-time WS peak monster even after I give his defenses half his credit

8. Larry Doyle (9-8-13, PHoM 1975)—at his best, a deserving NL MVP on a pennant-winning team; about 9 WS per year better than the median NL 2B

9. Phil Rizzuto (4-9-14, PHoM 1995)—lots of Yankees on this list; fact is, I hate the Yankees, but they had the horses or in this case the gloves

10. Bucky Walters (45-38-36, PHoM 2006)—the big mover in my pitcher re-eval, 6 WS per year better than the median “ace” over 7 years and this is AFTER discounting his WWII years

11. Johnny Pesky (11-15-24, PHoM 2004)—28 WS (in his rookie season) and 34 WS in the years before and after spending 3 years fighting WWII; adjusted WS peak (adjusted to 162 games), then, is 36-30-30, compared to Rizzuto’s 37-27-26 and Trammel’s 35-29-26; rate is 23.8 to Rizzuto and Trammel’s 22.5; rates below Rizzuto due to his years in the wilderness of 3B

12. Al Rosen (12-18-38, PHoM 2005)—the #1 WS peak of anybody in my 30 hitters—Dale Murphy never had a 42 WS year (44 when adj. to 162 games)—but, of course, the shortest career of the lot; but, also fills a position imbalance

13. Hugh Duffy (13-34-26, PHoM 2005)—I understand WS over-rates him, that’s why he’s here and not higher up

14. Kirby Puckett (15-3-4, PHoM 2001)—drops down after re-eval

(14a. Alan Trammell [15a-24a-11a])

15. Dale Murphy (23-16-16)—no Albert Belle but 33-32-32-31-29-28 is a damn nice peak

I also noticed that Greg Gagne wins 4 GG in (is it) DanR's system (?). There are those in these parts who believe he was the MVP of those Twins WS champs. I don't quite go that far, Gagne was not a leader, not an inspirational kind of guy, and frankly he sometimes struggled when expectations were put upon him. Still he was just consistent as hell at SS and made the pitching staff look good, not to mention the rest of the IF. I will say that those were outstanding defensive teams overall, however, with Kirby (A+!) and Hrbie, who was really quite good in the field. 2B was not always strong until Knobber showed up. Gaetti was good but not great and the corner OF were passable, Gladden more so than Bruno. Then there's catcher. As a catcher Brian Harper was a good hitter and, by all accounts, a good Christian. But the whole package was pretty good and Gagne was the heart of it. I'd rank him second behind Kirby in terms of his value over those 5 years 1987-1991. 'Course that's the answer to sort of a trick question since so few guys played on both of the WC teams. Not a pitcher. Just Kirby, Hrbie, Gagne and Gladden, if I recall. But anyway, Gags could pick it.

Hrbek was an outstanding defensive 1B. The numbers (esp. Zone Rating) show Kirby as merely average in the field after 1985 (which is, coincidentally or not, right when he started to hit). His high putout totals are attributable to a low-K, flyball pitching staff (CF putouts as a percentage of total OF putouts were perfectly league average during his time with the Twins, according to Michael Humphreys).

Maybe that's why Gagne never seemed to screw up, too, I don't know. He was sure always in position for the cut off.

Kirby "looked" a lot like Torii in the OF, meaning making that catch up over the OF wall to rob the HR. I've seen more of Kirby and Torii than other CF, but I wonder if other guys have made that play as often as the two of them did.

What do the numbers say about Kirby's arm? What I saw was a guy who came up throwing, very quick release, and very accurate. Made a lot of plays at the plate, though I say "a lot" as an impression, haven't looked at the numbers in years. But better arm than Torii for sure.

The over-the-wall catch to rob a HR is most definitely captured by Zone Rating and its more sophisticated cousins.

I haven't seen a proper study comparing Kirby's kill and hold rates to league average, but you could probably do it without too much trouble if you're handy with Retrosheet. John Walsh of the Hardball Times outlines the methodology at http://www.hardballtimes.com/main/article/cannons-and-popguns-rating-outfield-arms/.

I'm not sure whether Peter Gammons handled the Baseball Notes column in the Boston Globe throughout the relevant period. Maybe not.

As I recall "Gammons" was very high on Gagne, maybe the best shortstop in the league, certainly one of the few players (a dozen?) whose exceptional defense was sometimes noted. Moreso than Puckett, I know, or Ventura, I guess (I was no longer a regular reader after Gagne's prime years). "Knowing" Gammons as Mr. Rolodex, I suppose that Gagne was frequently or highly praised by baseball people around the league.

If Greg Gagne batted like Vern Stephens he would be in the Hall of Merit despite a short career (meas. in plate appearances), and he would have batted more often, too.

As I recall "Gammons" was very high on Gagne, maybe the best shortstop in the league, certainly one of the few players (a dozen?) whose exceptional defense was sometimes noted.

Hrbek too. Maybe Gaetti but I may be confusing my Gs.

If Greg Gagne batted like Vern Stephens he would be in the Hall of Merit

How many players can that be said for? If Ozzie Smith hit like Arky Vaughan, he'd have been the greatest player ever...conversely, if he had fielded like Howard Johnson, he'd never have had a MLB career.

Gagne was a pretty bad hitter. He never took more than 30 walks as a Twin, though he took a few more later on. Career OBA .302 over a BA of .254. He hit 111 HR and 296 2B and 50 3B for a SA of .382, which is not horrible I suppose. But more than 3K for every BB is not so good for a guy whose ability to put the bat on the ball could not so stellar in the first place.

It has been said that giving up on Gagne was the worst decision Tom Kelly ever made. From a strict value standpoint, Gagne was who he was, but with those limitations, you got a pretty good value out of him. But over 8 years he never taught himself not to chase the ball down into the dirt and hundreds of execrable PAs were just driving Kelly nuts. (Actually I suppose you could argue that Gagne improved a bit, from 100+ K in 1986 and 1988 to no more than 83 as a Twin (through 1992) thereafter. But still. Some of the worst ABs I've ever seen from an established MLer. With just a bit more plate discipline he could have been so much more valuable, or so Kelly thought.)

But then, the Twins decline can also be traced to Gagne's departure. 90 wins in '92, 71 in '93, and the Twins had to beat their pythag in '93 to do that. Team ERA+ from 109 to 92 in '93 and 86 in '94. Defensive errors from 95 to 100 to 112.5 (pro-rated) in 1994. DP down from 155 in '91 to 141 despite many many more opportunities by '95.

If the Twins had had Gagne these past 5 years, they'd sure be a lot better off.

Twins All-Time Team

C- Battey. Mauer's got a couple years to go.
1B- Hrbek
2B- Knoblauch. Really don't want Carew out here.
SS- Gagne
3B- Gaetti
LF- Allison
CF- Puckett
RF- Oliva. Can't put Hunter in LF and Puckett in RF. I want Tony.
DH- Killebrew. Nope, can't see Carew here either.

Hard to believe I can't get Carew or Hunter in there, but I like the up the middle defense I've got and I like Hrbek's combined bat and glove at first. Actually I would try and see if I could teach Carew to play LF, then I'd be a happy man.

Just killin' a rainy Sunday afternoon. I'm not a football fan, you know.

Chris, WS back to 2004 at least are at http://www.hardballtimes.com/thtstats/main/?view=winshares. Dunno about 2002-03.

Are those win shares using the same exact formulas as Bill James win shares? My recollection is that they handle pitcher hitting differently.

On shortstops: first, answers on the system, then a couple of lists.

1. What makes Maranville the best of the batch?

Essentially, it's his career.

Are you using unadjusted WARP1 rather than some adjusted form of it or WARP3 for the BP chunk of your rankings? Maranville's gaudy WARP1 total is entirely due to BP's nuts treatment of prewar infielders.

Yes, WARP1. I apply a period adjustment to scale WARP1 to WAR in my system, which pulls a good deal of this excess out. Their approach isn't quite right, but as my ballot isn't choked with pre-1930 infielders, I don't see a systemic problem with using it. Maranville stands out, and his contemporaries don't.

If you buy their take on Rabbit, where's Tommy Corcoran on your ballot? WARP3 takes the wind out of Maranville's sails a good bit, although it still sees him as more comparable to guys like Concepción than I do.

The period adjustment for the 1890s, plus the win-shares take on Corcoran leave him nowhere near my ballot. Maranville does well by win shares.

2. Why so unfriendly to Campaneris? Are you making any adjustment to BP WARP or WS to account for his superlative, historically great non-SB baserunning?

I'm not downgrading Campaneris below where the numbers place him. I am not adjusting WARP or WS for baserunning value. That's not what sinks Campaneris: it's that WARP doesn't see him as nearly the player Concepcion was. That's what accounts for almost all the distance between him and Concepcion in my rankings.

3. How do you get them spread so far apart?

You can see the composite numbers on my ballot. I rank them in order, interfiling them with other players as those players come up. See below for a compact list.

You have Maranville at #4 and Campaneris at #58. If my system really makes up half of your evaluation, that seems a bit incongruous. Is the separation among them caused by your method of breaking up the backlog into decades? If so, would you mind explaining the procedure in a bit more detail?

Your system does make up half of my evaluation, but the players in the backlog are all closely bunched, so where your system, filtered through my weighing of career and peak, doesn't see differences, differences in other systems may be influential. The separation may be influenced by the way I handle individual decades.

The system. For each of WARP, WS, and WAR, I sum career value, total peak (wins above average in all seasons above an average player's seasonal total), and peak rate X 5 (peak rate is the player's top rate over 5 or more consecutive seasons). For win shares, I give a small fielding bonus to pre-1930 infielders and a 10% bonus to career and total peak for all infielders. I convert win shares to wins, and scale WS and WARP to WAR's level by subtracting the average difference between their totals and the WAR totals for the players 10 spots above and below the likely in/out line for that decade. Then I multiply the WS and WARP totals by .5 and add them to the WAR total, and rank the players in descending order. The all time in-out line in this system is around 240 for position players. If I were using WAR only, the line would be around 120.

Here are the shortstops, as they would be ordered if I were using only WAR scores:

Harrah 130.15
Concepcion 129.15
Bancroft 126.85
Campaneris 126.7
Rizzuto 126.05
Pesky 120.7
Maranville 120.4
-----------------
Stephens 119.95
Bartell 110.05

Here are the shortstops, as they are ordered when I combine WAR with WS and WARP

Maranville 256.77
Bancroft 250.78
-----------------
Concepcion 238.22
Pesky 234.87
Rizzuto 233.57
Harrah 228.38
Stephens 225.14
Bartell 223.54
Campaneris 220.9

Here is how my system would rank them, if I used scaled win shares only:

Maranville 118.73
Harrah 117.67
Bancroft 109.85
Pesky 109.83
Rizzuto 109.3
Stephens 105.13
Concepcion 93.88
Campaneris 91.4
Bartell 91.3

Here is how my system would rank them, if I used scaled WARP1 only:

Maranville 154
Bancroft 138
Bartell 135.85
Concepcion 124.25
Pesky 118.45
Rizzuto 105.75
Stephens 105.25
Campaneris 97
Harrah 78.8

These lists should make it clear which system is moving a given player away from the very closely bunched group that WAR finds.

What do the numbers say about Kirby's arm? What I saw was a guy who came up throwing, very quick release, and very accurate.

According to John Walsh's study in THT Annual 2007, Kirby was +1.9 runs per 162 games and +13.9 runs overall with his arm in CF from 1986 onward. His yearly rate was almost definitely somewhat better than this for his career because his best assist totals came in 1984-85, when the necessary Retrosheet PBP data was unavailable for all players. The interesting thing about Puckett is that his Hold rate generally remained around average (102), though his Kill rate was very high (138). This seems especially odd because the beginning of his career, when his arm would logically be less of a known commodity, is not included in the data. Most guys with excellent arms seem to establish and maintain high Hold rates as their careers progress based on reputation. In contrast, baserunners continuted to test Kirby, and he kept throwing them out.

Well, there's my answer. Fascinating that putting my numbers through your system makes Harrah #1, since I have him near the bottom of the group (and well behind his contemporaries Campaneris and Concepción). Is it possible that Campaneris is getting caught between the 1960s and 1970s? Could that make a difference?

My only suggestion is that I think your period adjustment to WARP1 may not be strong enough, judging by the fact that the the three players with the earliest careers are also top three players on the WARP1 list, and by a large margin.

FWIW, I think BP BRAA is a rather good (albeit flawed) metric, and FRAA is probably the best publicly available pre-PBP defensive stat (at least judging by its correlations to PBP measures). The problems with BP's numbers are all in their replacement levels. If you just apply a sensible replacement level (FAT, 80% of positional average, whatever) and subtract it from BRAA+FRAA, you're likely to get fairly accurate results.

#182
If Greg Gagne batted like Vern Stephens he would be in the Hall of Merit despite a short career (meas. in plate appearances), <u>and he would have batted more often, too</u>.

184. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 14, 2007 at 03:54 PM (#2576373)
> If Greg Gagne batted like Vern Stephens he would be in the Hall of Merit

How many players can that be said for? If Ozzie Smith hit like Arky Vaughan, he'd have been the greatest player ever...conversely, if he had fielded like Howard Johnson, he'd never have had a MLB career.

It depends what you make of the underlined clause, which I tacked on as a light-hearted afterthought, but would be crucial. Because Gagne actually finished with only 6207 plate appearances, many voters wouldn't consider him --but the enthusiastic support of a large minority would carry him.

FRAA is probably the best publicly available pre-PBP defensive stat (at least judging by its correlations to PBP measures).

Duh, that approach should be obvious. Has it been executed systematically for several uber-fielding ratings?

Yes, Michael Humphreys has published all the correlations. His own DRA, unsurprisingly, leads the way with a .76 correlation to a UZR/Dewan Plus-Minus/PMR hybrid, which he says is comparable to the correlation that the PBP systems show among themselves. Pete Palmer's Range is a distant second at .6, BP FRAA is close behind at .58, Palmer's Fielding Runs is on its tail at .55, and Fielding Win Shares holds up the rear at .47. Unfortunately, he has not made the complete set of DRA data available to me, so I can't use it in my WARP. I'd never heard of Palmer's Range and would need it in spreadsheet form to include it.

Fascinating that putting my numbers through your system makes Harrah #1, since I have him near the bottom of the group (and well behind his contemporaries Campaneris and Concepción). Is it possible that Campaneris is getting caught between the 1960s and 1970s? Could that make a difference?

Yes, I'm also surprised that Harrah looks better to me using WAR than he does to you. 1960s vs. 1970s doesn't make any difference in my system's adjustments. I'd have to study BP's numbers closely to see why it like Concepcion so much better than Campaneris, because that's where the difference enters my rankings. I do adjust W1 for the DH (and WAR), and obviously Harrah isn't hurt by the DH, so that's not the reason.

My only suggestion is that I think your period adjustment to WARP1 may not be strong enough, judging by the fact that the the three players with the earliest careers are also top three players on the WARP1 list, and by a large margin.

Yes, I agree that is something of a factor. As a matter of convenience, the adjustment i have applied is not position-specific, so insofar as the FRAR-FRAA difference does not vary at the same rate for all positions, the adjustment will not treat all positions even-handedly. I haven't bothered about that for the most part because I view the over-rating of infielders in WARP as an offset to their being underrated in win shares, but it may be that pre-WWII inflelders are being overrated with respect to post-WWII infielders. I'll look into that a more before I cast my ballot this year. I in no way have time to make a systemic fix, but I'll look closely at whether I ought to make some modest subjective adjustments here.

You adjust my WARP for the DH? That explains it! The DH adjustment is absolutely already built into my WARP--replacement levels are 0.6 wins per 162 games lower in DH leagues than in non-DH leagues across the board. If you're making a further adjustment, you're double-correcting for the DH, which would absolutely account for Harrah's placement atop the leaderboard. BP's WARP1 defintely do need a DH adjustment, however.

Also, that's clearly what's going on in the BP portion. BP does not over-credit prewar OF's at all relative to postwar OF's--just to take one example, FRAR-FRAA per 162 games for RF is 18 in 1925 and 17 in 1998. But the change for infielders is dramatic: it's 49 for SS in 1925 and 25 in 1998. That's a *massive* differential--2.7 wins a year--which your system is misallocating. If you don't correct for that, prewar infielders will systematically come out waaay too high in the BP portion of your rankings, as your chart of the SS rankings according to BP indicates.

You adjust my WARP for the DH? That explains it!

I am using an older set of WAR numbers, before you built a DH adjustment into the system, also before you made some changes to keep replacement level steadier over time, if I recall correctly. The numbers I have been using have a single replacement level for all major league players at a given position. If I weren't making those adjustments, Campaneris would not come out close to Concepcion using the WAR numbers I have.

I will consider how to best to adjust the WARP1 numbers. I don't accept all the league-strength jiggerings that go into WARP2, nor the approach to season-length adjustment for WARP3, and I think that some acknowledgement of the greater defensive role of infielders pre-lively-ball is needed in each system, so I not going to just normalize the early fielders to the modern numbers.

for 2002 WS:
http://www.baseballtruth.com/winshares/bbt_winshares.htm

If you look at the explanatory article, it includes this line: "Besides, the inventor of win shares -- Bill James himself -- gave me his blessing to go ahead and run them."

for 2003 WS:
http://www.baseballgraphs.com/2003.html

for 2004-2007, the hardball times as previously mentioned.

I think that tht (and bbgraphs) handles all negative player value differently than Bill does.

I believe that the annual editions of The Bill James Handbook include WS calculated according to the original formulas.

I've been re-eval my pitchers. As a WS voter, it's true that modern starters have a hard time measuring up. I've adjusted for that in various ways, and here's another way. Along with raw WS, now I'm looking at WS above the median "ace." By "ace," I mean the #1 starter for each team based on WS and I find the median #1 starter by WS and simply compare the subject pitchers to that median. It's a manual process so it takes awhile. I'm working my way through my backlog pitchers and yours.

I include only what I call "prime" seasons, which are ERA-eligible seasons of ? 100 ERA+. IOW words a non-consecutive prime. You'd be surprised at how few such seasons our backlog pitchers have. Well, ok, no, you wouldn't. You already knew that. And I didn't see the point of calculating any MLE-type years, since this is a rate and MLE years would be based on the existing rates. And like any rate stat, you've then got to consider the length of the prime, and how you do that would determine how you get from this list to an actual ballot.

There's any number of ways to slice it from there. Take Dizzy Dean as an e.g.

His median total is 27.5
The median of medians is 21
His differentials in his 6 prime seasons are: +4, +1, +15, +10, +10, -1

So his results are:

1) the differential of the medians 27.5-21=+6.5
2) the median of the differentials is 10+4/2=+7
3) and the average of his differentials is 39/6=+6.5

In some cases you get big swings among these numbers, so it just helps to not get blindsided by some statistical oddity or other. But unless that is the case, I use the average of the differentials as my main number.

Here's who I've got so far.

1. Dean +6.5 for 6 years
2. Bucky Walters +6 for 7 years--and so he has moved on to my ballot, wins tie-breaker with Ferrell which is the other 2 measures, and OH! this is AFTER discounting his WWII years
3. (Wes Ferrell) +6 for 8 years--HoM not PHoM
4. (Jim Bunning) +3.5 for 10 years--HoM not PHoM
5. Wilbur Cooper +3 for 11 years--wins tie-breaker
6. Burleigh Grimes +3 for 9 years
7. Bret Saberhagen +3 for 6 years--I started this little project because of Saberhagen, BTW
8. Don Newcombe +3 for 7 years--loses the tie-breaker, you might say, but look who he out-performs, I didn't see the point of calculating MLE-type seasons
9. (Dave Stieb) +2.5 for 10 years
10. Vic Willis +2 for 9 years

11. (Early Wynn) +1.5 for 10 years
12. Luis Tiant +0.5 for 10 years
13. Jack Morris 0 for 10 years--wins tie-breaker, I still say he's a lot better than given credit for
14. Addie Joss 0 for 8 years--drops off my ballot
15. (Billy Pierce) -0.5 for 12 years, and his very best year is just +3, 3 WS better than the median ace
16. Tommy Bridges -1 for 9 years
17. Lefty Gomez -1 for 9 years, like Bridges playing time issues
18. Eddie Cicotte -1.5 for 9 years,wildly inconsistent

Conclusions:

1. Walters is #4 and Dean #18 in the backlog. Obviously Walters had a few below-average years in addition to the 7 prime years, and Dean didn't. Still this seems like to large of a differential to me.

2. I may have have missed on Wes Ferrell and maybe Jim Bunning. I would need a larger sample to know if I missed on Dave Steib or Early Wynn. I didn't miss on Billy Pierce, who was a below-median "ace."

3. Saberhagen is viable, but so is Don Newcombe and Newcombe has the same advantage over Sabe that Walters has over Dean. You guys are missing on Newk. Cooper and Grimes are viable as well. The next backlog pitcher, in short, really ought to be Dean, Walters, Sabe, Newk, Cooper or Grimes. Anybody else would be a mistake IMO.

4. The especial mistakes would be Joss (my mistake), Cicotte, Bridges and Gomez, who are below median "aces."

5. Tiant is barely better. Now I know what you're going to say. Giants roamed the earth, etc. etc. Following are the median aces of Tiant's prime seasons and Tiant's differential vs. the median, starting in 1967. Earl Wilson (-1), Culp (+12), Odom (-7), Ryan and Hunter (-5), G. Perry (-3), Busby (+7), Busby and Tanana (-5), Gura and Goltz (-2), Flanagan and Koosman (-11). My point is not all of the median pitchers are giants. And you had an expansion right in the middle there, too. Jack Morris was just as good as Tiant.

The medians are volatile in this system. In any given year, there are multiple medians depending on who the subject is. IOW the median is for the other 7 teams. So here is how the medians have evolved, expressed as a range.

1900s Range 19-27
1910s 19-28
1920s 17-23
1930s 15-26 (the 15 was 1939, otherwise a low of 17)
1940s 17-25
1950s 15-22
1960s 16-21
1970s 15-24 (the 24 is quite extreme, mostly it was 17-21)
1980s 15-18
1990s 15-18 (small sample)

I would guess that the declining medians gives guys like Stieb and Saberhagen a slight advantage, though OTOH they were also affected by whatever forces were suppressing the median. Bucky Walters was +23 in 1939, which might be regarded as a bit flukey. There is not another +20 in the entire study, though obviously if I were doing HoMers rather than backloggers there would be. But if the mediann in 1939 had been more normal, say 20) he would still be +18 and his prime average would drop only 0.5. I can't see any reason to do that, however, it's not like it was a war year. I did discount his war years, however.

I am moving on to the hitters. I don't have much yet, but thought I ought to toss this out there. Same methodology as for the pitchers (above). So far, just 5 guys.

1. Kirby Puckett +8.5 for 11 years--10 years CF, 1 year RF
2. Dale Murphy +8.5 for 10 years--and, yes, I again followed the position switch
3. Reggie Smith +6.5 for 11 years--ditto
4. Bob Johnson +6 for 10 years
5. Gavy Cravath +5 for 7 years--extrapolate to however many seasons you like including MLEs

But here's the catch. I'm also calculating the differential versus the #1 player at the position (or #2 if the subject player is #1).

Versus the #1 (or #2) player at the position

1. Cravath +3--again, this is per season
2. Puckett -5
3. Murphy -5
4. Smith -5.5
5. Bob Johnson -7.5

I have not seen anything like Cravath's swing, with a smaller margin versus the median and a bigger margin versus the best or next best player at the position. i.e. +3 over the 2nd best and only +5 versus the median. So obviously there was a pretty tightly grouped little bunch of players at the top. The median RFs were Evans, Titus, Snodgrass, Stengel, Johnston and Griffith, Griffith, and Johnston Griffith and Southworth. So maybe this is just a totally flukey weak cohort. Your call.

Meanwhile, Bob Johnson fares the worst of this small group when compared to the big dogs. And it's not like he's leading against the median either, that is compared to Puckett, Reggie and Murphy. Johnson's medians are Vosmik (rookie year), Roy Johnson and Ben Chapman, Roy and Goslin, Radcliff, Vosmik and Mills, Dixie Walker, Walker and Averill sharing the position at Detroit, Chapman again, Case, and Case. Certainly a better group than Cravath's.

Reggie's medians Pepitone (yes as a CF), Jay Johnstone, M. Stanley, Blair and Monday, (moving to RF) Buddy Bell, Coluccio, (moving to NL), Gross and Crawford, Gary Carter (!), Kingman, Murcer, Griffey and Valentine. Pretty good group.

Murphy's are Youngblood, Landreaux and Ru. Jones, Landreaux, M. Hall and McGee, Dernier, M. Wilson and McReynolds, Gladden, (moving to RF) Maldanado, Brnansky, Dawson and L. Walker.

Puckett's are Gary Ward, Lynn and Pettis and Yount, O. McDowell, Burks and W. Wilson, Moseby and Gallagher, D. White and Griffey (rookie), Yount and L. Johnson, B. McRae and P. Kelly, D. White and J. Gonzalez, McRae and B. Anderson, Anderson and Johnson, (and in RF) O'Leary.

Your call whether any of these guys faced especially strong or weak cohorts or whether it matters. Interesting to remember some guys who played CF or OF generally--G. Carter, B. Bell, Joe Pepitone, Juan Gone in CF, Earl Averill in LF for 3/4 of a year after being traded to Detroit, and so on. Or some guys who were actually median players at one time. I don't remember Mel Hall ever being that good, or Bob Dernier, or Candy Maldanado or Brian McRae or Troy O'Leary or Bob Collucio. Oddibe McDowell is forgotten now but had a nice run, Greg Gross, Lance Johnson...