DanR #200
So, as you can see, a team with Reggie Smith over 22 seasons would have accumulated 473 WS from his position, more than Cash (458), Cepeda (441), and last of all Pérez (435).

That's all
I don't believe in "bump"

189. jimd Posted: November 21, 2007 at 03:44 PM (#2622966)
> and one of the all-time great defensive teams; I am sure they ran up against the fielding WS cap.

Just about all of the good deadball era teams ran into the fielding WS cap.
(As did most good teams before then also.) The players we care about tend to be concentrated on the good teams in this era, and most FWS are capped, and PWS are correspondingly inflated.

jimd,
Do you know this by general knowledge or is it easy to calculate from stats in the Win Shares book or is it difficult to calculate. For example, Pittsburgh Chicago and New York 1908. Where do you look up or how do you deduce which of the three teams hit the ceiling for FWS?

Eric Chalek
So taking post-season into account, Jeter gains about 20% more PA, and he vaults four places up the PA ladder above one HOM SS with a career currently as long and three others (assuming we can all agree Larkin will be a HOMer) with longer careers. Given that what we're talking about here is opportunity, his opportunity alone is pretty hefty. The next biggest opportunity-getter is Reese, whose 573 3OCT PAs bump him up two slots. Setting aside the question of war credit for the moment, would those postseason PAs alone make Reese better than Glasscock and Wallace?

Plate Appearances?

Eric,
Then what do you say to using a Pete Palmer zero-sum rating for the postseason? In sum the post-season participants get zero, although you don't follow Pete Palmer in rating the season that way.

re: Win Shares and winning teams-
while some have shown that it's harder to accumulate individual WS on a poor team, it can also be harder to get them on a truly outstanding team; like some of those teams in the early 1900s that won two-thirds of their games. Once you get to that level, it tkaes MORE runs to win an extra game, since you're alreayd winning a lot of blowouts. A team of 'WARP = 8' players would underperform their combined WARP, because there are only so many games a team can win in one season.

So no, Chance did not benefit in WS from his fine Cubbie teammates.

Prelim.

My new system for hitters borrows a lot from Dan's work, adding some OPS+ to it. My pitching system is similar to what I used before, but uses BPro's NRA and DERA instead of ERA+. There's still other stuff I'm working on, namely the catcher bonus and the pitcher to position player translation. There are some significant changes from what my ballots used to look like before I lost all my older stuff.

1) Raines
2) Fred Dunlap
3) Vic Willis
4) Bret Saberhagen
5) David Concepcion
6) Luis Tiant
7) Eddie Cicotte
8) Albert Belle
9) Phil Rizzuto
10) David Cone
11) John McGraw
12) Bob Johnson
13) Tommy Leach
14) Bruce Sutter
15) Reggie Smith

I'm continuing to work on my WS over the median, and I'm sitting on my OF until I've finished them all but Juan inspired me to offer a sneak preview.

Left Field

1. Tim Raines +124 WS over the median for the sum total of his "prime" seasons
2. Albert Belle +108
3. Frank Howard +98
4. Jim Rice +82
5. Lou Brock +69
6. (Jimmy Sheckard) +64
7. Bob Johnson +47

For additional comparison, Wally Schang is +36. Other than that I am hard pressed to come up with a comp for Indian Bob. Next worst among my consideration set would be Phil Rizzuto at +47 but that's with no WWII credit. Pesky is +57 but ditto. Bresnahan is +57. Bobby Avila is +56 with no MLE credit. (Avila is better than Willie Randolph and very close to Lou Whitaker.) But back to my main point, Bob Johnson is a pretty bad candidate. Compare his +47 to Gavy Cravath, with his truncated career (no MLE credit), at +77. How about Dale Murphy at +83.5 or Kirby Puckett at +107. Ken Singleton +91. There's a couple dozen hitters who belong in the HoM ahead of Bob Johnson. I mean Tony Perez is at +73. I'd like to see +100 for a hitter. Something more than half of that would be nice.

Then Reggie Smith +71 versus Murphy +83.5, Puckett +107, Singleton +91, Jim Rice is even better at +82.

I understand some people don't use WS but if a guy is half of another guy by WS, no system that I'm aware of can make him better.

> Bob Johnson is a pretty bad candidate

Only if you assume performing at the median for extended periods of time is worthless.

"I understand some people don't use WS but if a guy is half of another guy by WS, no system that I'm aware of can make him better."

............

Funny, when Bob Johnson rates so far below those names, I wind up thinking, "I understand some people use WS, but if Bob Johnson rates only half those other guys, I'm aware that WS could be a lot better."

C'mon, DL. I didn't say it (he) was worthless. And I didn't say he played at the median, I said he was +47 WS versus the median. 47 WS better than the median. 47 WS worth of worthiness beyond the median of worthiness. Not worthless.

But, what if other players played further above the median AND played for a longer period of time? Would that not have more worthiness? Why not select the player who had the most worthiness rather than a player whose greatest qualification seems to be that he wasn't worthless?

I mean, Lou Brock and Tony Perez haven't been within miles of my ballot but both played above the median for a more extended period of time than Bob Johnson. Jim Rice played for about the same period of time but further above the median. I know, I know. Rice is over-rated and Johnson is under-rated. If this were the HoU, great, Bob belongs in that Hall, great. Rice was still better. Frank Howard played about the same period and played way further above the median. +50 WS vs. Bob, but of course his glove wipes all of that out.

Well, sorry, I'm ranting, but I didn't say he was worthless. But, man, is it ever easy to debunk a straw man argument. Tell me why he's the best candidate (other than Albert Belle) among the hitters.

Win Shares, for reasons we've discussed many times here, seem to really undervalue Bob Johnson, so any system based on Win Shares is going to do the same.

WARP1 has Johnson at 101.

Player Overall Wins has Johnson at a robust 36,just behind Raines and his 36 POWs.

Raines and his 39 POWs.

One reaction to the line of argument exemplified by post #107 and 112 on this thread (and numerous other posts as well):

If you're going to credit a player's value above some fairly high baseline while that player was in the lineup, and then not penalize him too severely for not not being in the lineup because he could have been replaced ...

Then for my purposes, I think the highest-ranking "hitter" candidate (other than the shoo-in Raines) should be Jack Clark. Or maybe a near-tie between Jack Clark and Norm Cash.

As it turns out, I did have Clark on by ballot once but then said, "Oops - he missed too much in-season time. I've got to dock him for that." And I haven't had him on the ballot since. And I probably will continue to support such long-career low-peak candidates as Perez or Brock ahead of Clark. But if you're going to take that high-baseline, durability-discounting approach, why not Clark?

Whether a baseline is "fairly high," or the penalty for missing time is "too severe," is the eye of the beholder...the contention here is merely that the correct baseline to use is the one above which teams are actually willing to pay for the production. Everything up to roughly 80% of positional average is worthless and not Meritorious because it is available for free. Everything above that is valuable, and receives credit. I don't think this is a "high-baseline, durability-discounting approach" at *all*--I think it simply is the system that reflects the way the game is actually played. I would call a system that uses a lower replacement level to be a "low-baseline, durability-emphasizing approach," since it rewards production that is worth not one cent in a team's pursuit of a pennant.

Moreover, I suspect that the only reason why the 80% of positional average baseline, so widely cited as MLB replacement level, is somehow considered here to be a "high-baseline, durability-discounting approach" is because this particular electorate developed its consensus approach (to the extent that there is one) through an extremely heavy reliance on BP WARP and Win Shares, both of which use replacement levels far lower than the actual MLB figure. I think if you took a broad experts' poll within the sabermetric community, you would find a very strong consensus in favor of something around 80% of positional average, and an outright rejection of the 52% of league average/1899 Cleveland Spiders level that seems to inform the "Oops - he missed too much in-season time. I've got to dock him for that" approach.

Now, as for Jack Clark and Norm Cash, it's because there are simply far superior candidates. Reggie Smith, to cite the obvious one, was just as good a hitter in the same career length and played CF and corner OF instead of 1B. That puts him in another category altogether. There are a bunch more outfielders with similar offensive value and more defensive value than those guys as well: Gavvy Cravath, Bob Johnson, Chuck Klein if you value peak, etc.

this particular electorate developed its consensus approach (to the extent that there is one) through an extremely heavy reliance on BP WARP and Win Shares, both of which use replacement levels far lower than the actual MLB figure.

naah, I don't think so.

There is a lot of WARP and WS used here; but I would say most who use them build in an adjustment, REALIZING that the replacement level is lower than it "should" be. It is perfectly reasonable to use tools such as WS and WARP with some kind of adjustment; just like some of use RCAA/RSAA and adjust the opposite direction. That does not obviate (3 dollar word warning, sorry, I don't use that word in conversation much, but it fit) the use of any of the tools. I for one think the electorate has taken far too much grief for its use of these tools. I use a screwdriver as a wedge in some circumstances; some of my friends get on me for using the "wrong" tool, but if it works....

There is a lot of WARP and WS used here; but I would say most who use them build in an adjustment, REALIZING that the replacement level is lower than it "should" be.

Correct. I'm one of those who do so.

DanR #213
Everything up to roughly 80% of positional average is worthless and not Meritorious because it is available for free. Everything above that is valuable, and receives credit.

Is FAT simply all players under contract at the major league minimum salary?
If not, do some analysts use that definition?
Do some infer free availability from transactions --maybe interclub transactions, maybe including or focusing on waivers and releases?

There is a lot of WARP and WS used here; but I would say most who use them build in an adjustment, REALIZING that the replacement level is lower than it "should" be.

This assertion is incorrect. I was similarly skeptical of Dan R.'s argument, Tom, so I decided to go back and count up which ballots were using which systems. I tried to categorize each ballot as best I could, awarding half-credit to a category if the ballot was an identifiable mix of 2 different systems. The numbers come out as follows:
----------
WS, adjusted for rep level: 9.3%
WS, unadjusted for rep level: 14.8%
BP WARP, adjusted: 0%
BP WARP, unadjusted: 7.4%
OPS+ (mostly position adj.): 12.0%
Dan R. WARP: 7.4%
Traditional Stats: 7.4%
Quantitative Mix*: 21.3%
Qualitative: 20.4%

*quantitative mix represents ballots using proprietary systems, or ballots which acknowledge using a blend of WARP/WS/OPS+ etc without indicating which dominated. About half of the mixed ballots represent sophisticated quantitative ranking systems; but the remainder definately are not adjusting WS and WARP for replacement level based upon who's being put on the ballot
---------------------
Conservatively estimating that 1/3rd of the quantitative mix ballots are not adjusting for rep level, ~30% of the HoM ballots are using WS and WARP and not adjusting for replacement level.
This is a conservative estimate: assuming that many of the qualitative group is influenced by the uber-stats, the true number is probably much higher.

I conclude that Dan is right; the HoM consensus is strongly biased by an inaccurate choice of replacement level influenced by ue of unadjusted WARP and WS.

On a separate note, I'd like to express strong criticism of the ballots from the following voters:

Sean Gilman
Rusty Priske
Rob Wood
Adam Schafer

When looking through the ballots, I found it impossible to ascertain any rationale for their votes, even a qualitiative ranking system. Their ballot comments generally were thinks like "Great Peak" or "superlative defender"; comments like this don't allow an observer to figure out the logic of the ballot order, which is, if I'm not mistaken, the point of ballot comments. Right now, their ballots are no less arbitrary than an uncommented ballot. I encourage Joe and John to ask these voters to provide an explanation for their votes in their 2008 ballots.

I encourage Joe and John to ask these voters to provide an explanation for their votes in their 2008 ballots.

I think it's pretty clear what type of rationale each of those longterm voters use.

Now, if there is a candidate on one of their ballots that sticks out among the rest, then feel free to point it out to them. I have done so in the past with quite a few ballots during these four years together and feel that's reasonable.

Nate Silver has FAT as over age 27 making less than twice the league minimum. I have just piggybacked on that work.

John, the rationale is clear? I disagree. Take Rob Wood's ballot (this is his 2006 ballot; I used that year for my study because 2007 was a low-participation year)

2006 ballot from this career voter (low replacement level):

1. George Van Haltren - deserving star of the underrepresented 1890s
2. Graig Nettles - super fielder; I am surprised by his lack of support
3. Bob Johnson - solid hitter, solid career (w/1 year minor lg credit)
4. Bobby Bonds - good combo of peak and career
5. Tony Perez - good career though he was barely an adequate 3B defensively
6. Will Clark - great peak, would be even higher if he had played longer
7. Bob Elliott - good 3B mired with woeful Pirates and Braves
8. Tommy Bridges - luv the strikeouts & win pct, with minor league and wwii credit
9. Rusty Staub - good peak + good career (similar to Perez)
10. Reggie Smith - boost from center field play and japan
11. Chuck Klein - very good peak and career (even after adjusting for park)
12. Bus Clarkson - everybody should give him another good look
13. Rabbit Maranville - better career than most realize (with credit for 1918)
14. Pie Traynor - was so overrated he is now underrated
15. Hack Wilson - great peak, short career

OK, so first off, we have a self-proclaimed career voter with Wilson, Klein, and Bob Johnson on the ballot.

But actually, I don't think thats so much a sin; what I'm pointing out is the complete inability for the observer to figure out the logic behind the ballot order. Just take the top 3:
"1. George Van Haltren - deserving star of the underrepresented 1890s
2. Graig Nettles - super fielder; I am surprised by his lack of support
3. Bob Johnson - solid hitter, solid career (w/1 year minor lg credit)"

Why Van Haltren #1? The only substantive comment we get is that he is "deserving". More deserving than Nettles? Than Johnson? Why?

Bob Johnson has a "solid career". Then why not put him 2, over Nettles? Nettles and Johnson are about as disparate as two candidates could be; a long-career glove v. a short-career bat. How is he comparing the two? We have absolutely -no- information to go on.

Long-term participation shouldn't excuse people from explaining their ballot analysis. That ballot, despite having no rhyme or reason, counts just as much as Chalek's, Dan R.'s, or Joe D's, to name some of our more "communicative" voters.

zop, excellent job culling thru our voters and presenting useful data. But I disagree that we are "strongly biased by an inaccurate choice of replacement level"

If between 22% and 35% of our voters used unadjusted WS/WARP; i.e., repl level too low; we can't conclude we are biased as a group unless we also see how many voters are using repl levels that are "too high" (by the 80% metric DanR and others have put posited with solid backing). We have some true peak voters that In Effect use a higher repl level. I don't know exactly how to quantify it, but I would guess that our HoM repl level is likely somewhat lower than the HoF's (not that I wish to invite comparison as a bonus pt in our favor...).

I do echo zop's sentiment that some of our voters could be more explanatory in their comments

I think that if a voter explains his voting method thoroughly, expansive comments on each candidate become less necessary. That has been my approach, more or less.

Juan, I couldn't agree with you more; in fact, I've taken to putting only a line or two on commentary on my ballots. What's important is (1) if you're using a system, you explain the system sufficiently so that observers can figure out your logic. (2) If you're not using a system, that you clearly explain the logic of your ballot placements.

Here's another less egregious example from 2006. This is the top half of Sean Gilman's ballot. I think the commentary for Leach and McGraw is adequate, but then it degenerates. Duffy is described as the best borderline OF candidate but is placed below Dale Murphy on the ballot. The Van Haltren and Oms comments are without any substantitve meaning; he could have written "black guy" and "white guy" in lieu of his actual comments, and the ballot would be no worse.
Lets say I want to try to persuade this voter to consider a candidate I support; what is he looking for? OK, so he wants a good peak and good career, but what measure is he using to rank peaks and career? An uberstat? OPS+? All of the above? A heuristic assessment? Again, I'm not arguing that a qualitative ranking system is bad; I'm arguing that, no matter what type of system you use, your comments need to be sufficiently descriptive so that all voters understand a voter's logic, not just the ones that remember his witty posts in the Sam Crawford discussion thread back in 2003.

2006

1. Will Clark (-)--I have no idea what the writers were thinking.

2. Tommy Leach (3)--May be the most underrated candidate out there. Great career value, fine peak and played two premium defensive positions. (1942)

3. John McGraw (4)--He’s got the best non-Browning peak of the backlog, but still a shortage of career value. (1997)

4. Dale Murphy (5)--A great prime with a decent career value despite the decline phase. Bumped up this year as he’s got the best peak of the outfield glut, and the career value difference is minimal. (2000)

5. Hugh Duffy (7)--High peak, medium length career, the best of a massive group of borderline OF candidates. (1964)

6. George Van Haltren (8)--Almost a HOMer not too long ago, will he make it eventually? (1966)

7. Alejandro Oms (9)--Another good, yet underrated, all-around outfielder. (1986)

I'm not sure if anyone ever looks at it, but the latest edition of my at-bats file is up at the Yahoo group. What's bad this time is I remember that I put the 2007 numbers in when I did the 2006 ones, because we all knew who they were going to be. I just never got around to posting it until now.

Speaking of updating things, I spent a lot of Friday putting all the new WARP numbers into my database. Wow. Those were not minor changes, especially for the 19th Century guys. I didn't trust WARP3 before, but I sure don't now. It's fairly obvious that he was adjusting for leagues having fewer teams in them (the 1900 NL makes it crystal clear), although it's also noticeable that the numbers are higher through the 00s-20s as well, but not as dramatically. Pitchers didn't change much, though.

And one thing that did stick out to me is that even in WARP1, the system is not at all fond of Hugh Duffy. Even just compared to his contemporaries, Van Haltren, Ryan and Griffin, he's significantly behind them (although WARP has always been in love with Griffin's defense.)

Say what you want about WS and Bob Johnson. WS is just a tool. The underlying truth is that everybody had a big hitting LF. If Bob Johnson had never been born, his teams would have had a guy out there who was just as good or at least *this* close. The same cannot be said of Albert Belle or even Jim Rice or *even* Lou Brock. All 3 of them would have been vastly more difficult to replace.

wow, sunny, that sounds like an exaggeration, altho I know the basis for it and why it is likely to have some truth to it.
Anyone have RCAP (which would account for quality of guys who play same position in same era) numbers for Johnson/ Belle/ Rice/ Brock?

220. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 25, 2007 at 03:26 PM (#2624764)
in reply to me:
Nate Silver has FAT as over age 27 making less than twice the league minimum. I have just piggybacked on that work.

Hi, DanR
Recently you referred to a consensus of sabermetricians that replacement-level production is about 80% of league-average production --or is it 80% of some regular player's average?

In order to measure replacement-level production, if I understand correctly:

Silver measures the production of freely available talent as defined above.
Woolner measures the production of players who replace others in emergency or off-season (maybe in intermediate circumstances too).
Are there other recent measures or does the consensus represent assent to Silver's and Woolner's work?

Pete Palmer's old "Total" and current Baseball Encyclopedia measures may be interpreted as setting replacement level at 100% of some league average where I don't recall the treatment of pitcher and designated hitter batting.
Clay Davenport's WARP ('R' for Replacement) is defined in terms of a replacement level less than 80%.
Bill James Win Shares may be interpreted as setting replacement level at 52% of some league average.

1. 80% of the weighted average production of all players at the position is the standard.
2. Tangotiger has certainly researched the replacement level question in depth as well, and his replacement levels agree almost precisely with mine (some slight positional differences but the overall range is exactly the same). If you hunt around on his two websites (tangotiger and insidethebook) I am sure you can find his methodology.
3. Brandon Heipp has a massive treatise summarizing the various baselines that different analysts have deployed to calculate value. He advocates a "chaining" approach for hitters, where players are compared against a progressively increasing baseline the more PA they accumulate. Of course, nothing in it suggests a baseline anywhere near what WARP/WS use.
4. I highly suspect that if there were any credible studies suggesting a replacement level substantially different than 80% of positional average, we would have heard about them, no? It'd be quite a major finding.
5. That's right, Palmer's numbers were compared to league average. But they didn't claim to measure replacement level, they claimed to measure value relative to average, simple as that.
6. Clay is on record saying his replacement level is the 1899 Cleveland Spiders (who, indeed, were 1.4 WARP1 as a team). So that's a .130 baseline, which translates to 53% of average (53% fewer RS, 47% more RA). This is, presumably, why BP WARP/3 and WS run so closely together.

I was doing some thinking about Bob Johnson and the whole pre-integration mess in the 1930s. I think we performed MLEs to see which players would have been stars assuming they were the only player added to MLB. Likewise we performed our analysis on MLB players assuming that MLB was the highest talent level league. By making these assumptions we virtually expanded the league and contracted it again after the war.

It would be an interesting exercise to re-do this section, assume there are only N major league teams, decide who would and would not make the cut, re-calculate replacement level and/or perform MLEs on the major league players to the hypothetical integrated league. After that we could re-vote and keep Bob Johnson out (along with Cool Papa Bell, Willard Brown, Earl Averill and Ducky Medwick). As for our current hall, Bob Johnson meets the established standards of his era.

I am not a FoBJ but I do not agree with sunnyday's contention that "If Bob Johnson had never been born, his teams would have had a guy out there who was just as good or at least *this* close." The worst-regulars average for corner outfielders in the 1930s was right on its 1920-1980 average. If positional average was higher, it was because there were more stars, not because there were better replacement players at the position than there were in other eras.

DL from MN, that is exactly what I have been saying for virtually every single ballot since I started voting, and is the "segregation penalty" I mention at the top of my ballots. I am thrilled to see someone else understands this dynamic the same way I do!

My voting is indeed based upon a system. I think I have told this story several times before. After Bill James announced he was writing a historical abstract, but before its publication, a friend and I tried to anticipate his results by developing our own system. We each ranked the top 100 players of all time (pitchers and position players separately). We then integrated our two lists. We then wrote down what were the key factors in our rankings. Based upon that information we then reverse-engineered an analytical formula to mimic our rankings. We have tweaked the formula a bit over the last 20 years, but it essentially remains the same as our original formula.

For position players we consider career on base pct, slugging pct, total bases including walks and hbp, runs, a measure of domination (based upon home runs and RBI), all adjusted for era and park effects. We also consider defense, post-season, war years, and a deduction for performance in WWII and the Federal League.

I finished up my OF last night. I've already provided some of them because confirmed my belief that Bob Johnson would be a gross mistake. But setting that aside.

The two numbers are 1) the players WS above the position median in his median year (in an 11 year prime, e.g., it would be the player's 6th best/worst season as measured by WS above the position median), and 2) the cumulative total WS above the position median during the players prime years. The prime is defined as ≥100 games and ≥100 OPS+.

The first number goes to peak/prime. The differential rank order from the first number to the second relates (obviously) to career length, but also (less obviously) to the fact that it is the median not the average. E.g. see Brock and Rice, or Puckett and Murphy.

My consideration set is the top 3 on my 2007 ballot, the top 3 in 2007 voting consensus, and anybody who is HoM/not PHoM, plus a couple of wild cards.

Left Field

1. Tim Raines +7 and +124--not as dominant as I expected but the #2 career total available
2. Albert Belle +9 and +108
3. Frank Howard +10 and +98
4. Jim Rice +1.5 and +82
5. Lou Brock +5.5 and +69
6. (Jimmy Sheckard) +5 and +65
7. Bob Johnson +5 and +47

Center Field

1. Kirby Puckett +7.5 and +107
2. Hugh Duffy +7 and +88.5
3. Dale Murphy +12 and +83.5
4. GVH +8 and +74.5
5. Reggie Smith +7 and +71
6. Jimmy Ryan +3 and +51.5--as long as I was doin' GVH and Duffy...does anybody remember that they made up one singular OF, I think in Chicago in '89

Right Field

1. (Dwight Evans) +3 and +97.5
2. Ken Singleton +9 and +91--I prefer Singleton to Evans, look at the yearly median
3. Gavy Cravath +13 and +45--with no MLE credit, I prefer him to Evans or Singleton with MLE credit
4. Dave Parker +5.5 and +71
5. Rusty Staub +5 and +54

And reaching back to my 1B results...

1. Cepeda +8 and +104
2. (Beckley) +6 and +87
3. Cash +4.5 and +84
4. Mattingly +6.5 and +79.5
5. Perez +4 and +68.5

BATS (Prelim)

1. Raines
2. Puckett--passes Belle due to defense and he was a better teammate
3. Belle
4. Cravath--his +13 is the best seasonal number on the board
5. Murphy--his +12 is second best in that category
6. Duffy

(probable ballot line here)

7. F. Howard
8. Cepeda
9. Rice
9a. (Dw. Evans)

(big gap here)

GLOVES

1a. (Alan Trammell) +9 and +147.5
1. Ed Williamson +11.5 and +109.5
2. Dick Lundy +7.5 and +130.5
3. Elston Howard +9.5 and +61--with no MLE credit
4. Larry Doyle +8 and +116

(probable ballot in/out line)

4a. (Joe Sewell) +10 and +117--but against an unusually weak cohort
5. John McGraw +13 and +105.5--moves way up, still probably short of ballot
6. Fred Dunlap +13 and +94
7. Johnny Pesky +6 and +57--against a historically great cohort
8. Tommy Leach +13 and +102
9. Thurman Munson +10 and +85
10. Al Rosen +11 and +83

11. Phil Rizzuto +5 and +47--no WWII credit factored in (yet)
11a. (Quincy Trouppe) +9 and +108.5
12. Bobby Avila +9.5 and +56--no MLE credit factored in (yet), one of the most underrated players ever
13. Bill Monroe
14. Vern Stephens +9 and +82--WWII discounts applied
14a. (Lou Whitaker) +6 and +100
14b. (Roger Bresnahan) +9.5 and +57
14c. (Willie Randolph) +7 and +88
15. Bob Elliott +9.5 and +81

5. John McGraw +13 and +105.5--moves way up, still probably short of ballot

C'mon Sunnyday, take the plunge and vote for McGraw! Aren't you a peak voter? Nearly as much value above positional median as Doyle, and in about half the playing time! :)

'the "segregation penalty" I mention at the top of my ballots'

Except the HoM as a whole didn't institute a segregation penalty. I think the standard has been set and we can't penalize the players clearly better than the standard established. Unless we start unelecting players we have to apply logical consistency to the 1930s and treat all pennants as equals.

I think each voter has his own standard, DL from MN! My standard is the following: of course a pennant is a pennant, and the relationship between being among the best X% of baseball players in a given year and pennant-winning is stable after adjusting for league size and standard deviation. So being in the top X% of all players of all races in the 1950s should be worth exactly as much as being in the top X% of all players of all races in the 1930s. In order to maintain this consistent standard, pre-integration stats from both MLB and MLE stats need to be deflated. Otherwise, we are electing the top 1.5X% or whatever of players in the 1930s vs. 1X% of players in the 1950s. Which, of course, is exactly what we've done. And I think it's a mistake, and I am exercising my inalienable right to suffrage to register my protest. :)

> we are electing the top 1.5X% or whatever of players in the 1930s vs. 1X% of players in the
> 1950s

Agreed, there were more available pennants in the 1930s. The standard has been chosen (albeit without direct debate).

For position players we consider career on base pct, slugging pct, total bases including walks and hbp, runs, a measure of domination (based upon home runs and RBI), all adjusted for era and park effects. We also consider defense, post-season, war years, and a deduction for performance in WWII and the Federal League.

Cool. How can convince you to take another look at Dick Lundy?

there were more available pennants in the 1930s

Says who? I say there were two, distributed over three leagues.

Says who? I say there were two, distributed over three leagues.

If the winner of a given league is called the "pennant" winner, then the number of available pennants equals the number of leagues. Therefore, periods of time in which there are more than two leagues of approximately major league quality have more than two pennant winners. By definition.

Having said that, I agree with Dan that its a mistake to elect substantially more players from eras where there were three or more equivalent leagues than from other eras.

Which is also a 19th century issue in that we've elected more players from the '80s with its 2 and sometimes 3 leagues than from the '90s with its 1 league.

How do you define "major league quality?" Tautologically, as "the major leagues?" The NgL's most certainly were not leagues of approximately major league quality, although they had a surfeit of major league quality players mixed in with a lot of retreads. Again, I prefer to think of it this way: the HoM standard is the top X% of players in the game, regardless of if there were two leagues or twenty. The more leagues, the fewer players per league.

Anyone have RCAP (which would account for quality of guys who play same position in same era) numbers for Johnson/ Belle/ Rice/ Brock?

RCAP
Johnson - 319
Belle - 310
J.Rice - 155
Brock - 56

> RCAP
> Johnson - 319
> Belle - 310
> J.Rice - 155
> Brock - 56

There's a big difference in defensive value between Johnson and Belle.

Yeah (tee hee). Indian Bob stays on my ballot.

I don't think that we elected more guys from the '80s because of the AA, at least not in the sense that we elected a lot of guys whose value was largely AA. OTOH if you mean it was easier to dominate the NL in the '80s than the '90s, and when faced with direct choices of '80s guys head to head vs. '90s buys we went with the '80s guys without thinking about league quality enough, well, then, yeah, I can't really dispute that.

It's been said before but wouldn't it be interesting to go back and re-do those first few elections, knowing what we know now. Of course, they might not come out any different, though I think they would.

Do you know this by general knowledge or is it easy to calculate from stats in the Win Shares book or is it difficult to calculate.

The formula for the WS team fielding cap is .32375*GP. Any team that is at 49.9 FWS is usually capped, unless they played more than 154 games. FWS counts such as:

42.7 (132)
...
45.2 (140)
...
49.2 (152)
49.5 (153)
49.9 (154)
50.2 (155)
50.5 (156)
50.8 (157)
51.2 (158)

occur with regularity if you scan the deadball (and pre-deadball) team numbers. The league fielding leader in many (most?) seasons is capped.

In 1908, the Cubs (51.2 in 158G) and the Naps (50.8 in 157G) appear to be capped.

typo: 45.3 (140)

That would certainly explain why WS isn't as fond of Lajoie's fielding as FRAA is, since Lajoie played on historically great defenses (especially 1906).

The formula for the WS team fielding cap is .32375*GP.

See page 33 of the WS book.

'zop, I'm just wondering, what did I get classified as? (I will assume that whatever category I was put in, I got dinged for not adjusting for replacement level.)

(probable ballot line here)

7. F. Howard
8. Cepeda
9. Rice
9a. (Dw. Evans)

Marc,
if i count correctly, Dwight Evans is the first RF, fifth OF, and sixth "bat" (Cepeda) by the prime measure that you use to order the player lists in #235. Why is he only tenth and off the ballot? In particular I wonder he is behind Jim Rice by mistake, for he beats Rice on the median-prime-season or peak/prime measure too (97.5 to 82, 3 to 1.5). Or is that "1.5" for Rice a mistake --the smallest number in #235 with Evans second smallest.

Marc,
Evans and Rice don't matter, of course, nor do most of those you rank above Evans matter in your system at this point. It looks like you can't do much better than vote for Raines plus 14 "gloves" and pitchers.
--

jimd #250
DanR #230

Thanks!
--

243. Dizzypaco Posted: November 26, 2007 at 11:45 AM (#2625232)
> Says who? I say there were two, distributed over three leagues.

If the winner of a given league is called the "pennant" winner, then the number of available pennants equals the number of leagues. Therefore, periods of time in which there are more than two leagues of approximately major league quality have more than two pennant winners. By definition.

245. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 26, 2007 at 11:53 AM (#2625240)
How do you define "major league quality?" Tautologically, as "the major leagues?" The NgL's most certainly were not leagues of approximately major league quality, although they had a surfeit of major league quality players mixed in with a lot of retreads.

There may also be a telling response from inside the framework Dizzy espouses. What is the number of Negro Leagues pennants in 30 years 1920-1949? about 45? (Jim Riley counts a Negro major league appearance for Ernie Banks in 1950 but that is unusually liberal, right? Others drop the Negro AL after 1949 iirc.)

Regardless of the answer this isn't adequate. The HOM is full of black players from the 1910s when there were no pennants.

254. Devin McCullen and the Epileptic Trees Posted: November 26, 2007 at 05:07 PM (#2625575)
'zop, I'm just wondering, what did I get classified as? (I will assume that whatever category I was put in, I got dinged for not adjusting for replacement level.)

IIRC, I had you as 50% quant. mix, 50% qualitative mix.

You cited OPS+, WS, and WARP in your post, so I had to consider you as partially quantitative, but didn't indicate a preference for either. You also indicated that you're not going solely by the numbers.

'Zop, I would guess that almost everybdy would be curious to know what manner of beast they are. Could you publish your entire list?

Hard to believe this is the last ballot. Final consideration set for last PHoM slot. Frankly, I could just leave a slot empty, not turned on by any of these guys. Listed in the order that I am leaning to right now.

1. Joe Sewell--rejected all these years, short career, and even then didn't stay at SS throughout; revises because of astronomical WS above position median scores, but his cohort was an unusually crappy one; still, value is value

2. Dale Murphy--short career, short prime, fell off a cliff and, like Sewell, didn't stay at an important defensive position; still, great peak with value on both sides of the ball; waaay short of Kirby Puckett overall, however

3. Wes Ferrell--short career, short prime, but a high peak within there

4. John McGraw--short everything including short seasons even at his peak; I'm a peak voter, not a rate voter; but his WS above position median are pretty good

5. Ken Singleton--probably the premiere candidate remaining whose value is pretty much all stick, but even with the stick he was no Albert Belle

6. Vern Stephens--excellent value on both sides of the ball, held his own against one of the great cohorts of all time--that is, AL SS 1940s, need to apply war discount of course, but a great great prime still remains

7. Jim Bunning--probably an OK HoM choice, unlike some other pitchers I didn't support, such as Billy Pierce who was approximately half the pitcher Don Newcombe was

Only one measly slot for all these 7 candidates. I hate the idea of admitting I'm wrong and I've been rejecting Sewell, Ferrell and McGraw for an awfully long time. That really leaves Dale Murphy....

"The NgL's most certainly were not leagues of approximately major league quality, although they had a surfeit of major league quality players mixed in with a lot of retreads."

Due to segregation, you can make the same claim about the white major leagues.

Due to segregation, you can make the same claim about the white major leagues.

I'm no expert, but I don't believe this is the case. The question isn't whether segregation affected the quality of all leagues, white or black - it did. I also don't think the question is the overall quality of the Negro versus white leagues. The point I think Dan was making was that the talent was distributed differently, which sounds right to me. That is, the level of talent among the best players in the Negro Leagues may have been roughly comparable to the level of talent among the best players in the white leagues. However, the level of talent among the median player, or among the lesser players was not as good in the Negro Leagues as in the white leagues.

I don't know if this is true, but I suspect it is. It might also matter whether we are talking about the Negro Leagues of 1921 or the Negro Leagues of 1937. If it is true for any league where MLEs have been developed, it has implications for the validity and analysis of Negro League statistics. If you run up your numbers against some (but not all) vastly inferior competition, how meaningful is that in translating raw numbers to major league statistics?

It's still meaningful, just not in a quantitative way. You just have to be aware of the context and recongize that the numbers are soft. They're (MLEs as published here on the HoM) still the best translations/interpretations of NgL value that have ever been created by human endeavor.

the newer stats (OWP, batting wins) at bb-ref seem to be finally working "right". Here are the career batting wins (essentially RCAA scaled for league offensive context) marks for our top backloggers in order. Obviously the glove candidates don't do as well, nor should they need to.

RSmith 36
TLeach 9
BJohnson 37
KPuckett 22
GCravath 29
TPerez 28
PRizzuto-
JMcGraw 27
HDuffy 20
GVHaltren 24
KSingleton 36
DConcepcion-
ABelle 34
(FChance 24)

Tautologically, major league quality simply means the quality of the major leagues. But my point, as I think we all understand here, is that if we want to apply a consistent standard of quality across the pre- and post-integration eras, we have to deflate both MLB and MLE numbers from the pre-integration era enough so that runs created on offense would equal RC surrendered by pitchers in a hypothetical integrated league.

RCAA is above average, right? And that's average average, not position average, right? No wonder we can't agree, there's really nuthin' to pick here. Add a few years to Cravath and maybe he's your man. On a rate basis he and Belle would have to be right up there, maybe McGraw. Add in some defense and, well, McGraw is just about the only guy with 20+ RCAA AND a glove. I suppose this says, hey, go ahead, vote for Johnson, Smith and Singleton and maybe Belle and add in McGraw and maybe Cravath and screw the rest. But wow, is that ever hair splitting of the highest order.

But if DanR is right, then Johnson is the one guy on this list who needs to get a nice big, fat, juicy discount for playing in an inferior league--25 to 50 percent of the superstar caliber players on earth at that time weren't allowed to compete with Bob. (Earlier guys like McGraw and Leach wouldn't have seen a whole lot of additional competition regardless.)

If I used this number, which I don't, I'd say Smith, Singleton, Belle and McGraw are your guys.

I thought we just wanted to quantitatively compare 1932 NeL stats to 1932 NL stats so that we could slot players from both leagues on the same HOM ballot. But I thought we had agreed back when NeL candidates were first eligible that the numbers of the NeL stars were inflated due to the lower overall level of play... like they were unfairly stuck in AAA for their entire careers (and indeed they *should* have been promoted).

These adjustments have always been part of the MLE's that were done for NeL players going back 60+ elections. DL, was this just a passing comment noting the injustice of segregated baseball, or were you commenting that the MLE-adjustments that have been done here have been systematically unfair?

I think MLEs for both of the segregated leagues into an integrated league are needed if you only want the top 1%.

I don't think he said the NgL MLEs are unfair, but that the ML stats should also have been "MLE'd" into an integrated context. Meaning, Babe Ruth doesn't hit 60 HR if he has to face Joe Williams et al instead of the bottomest rung of ML pitchers. It might only be 5-6-7-8 pitchers but with about 32 starting pitching slots available in each league that could be 25 games in which they would face vastly tougher pitching. And he'd have to hit ground balls through Dick Lundy instead of Bobby Reeves or a 36 year old Wally Gerber. And maybe in 1933 or 1934 he is replaced on the Yankees roster by Turkey Stearnes or Mule Suttles instead of by George Selkirk in 1935.

Or more to the point, maybe a borderliner like Tony Lazzeri who pretty much fell off a cliff in 1933 (age 30), maybe he isn't able to hang around til 1937 because there's a half dozen more ML caliber 2B around to fill in. He doesn't hang around, he doesn't pad his resume, he doesn't make the HoF.

That's the sort of consideration that wasn't given to integrating ML and NgL records.

Lazzeri was still above average from 1934-36, no?

I see what DL is saying though. Many think we inducted too many 1920s & 1930s guys overall. We kept treating MLB stats the same in 1925 as we did in 1905 despite that fact that there was a slew of star NeL's by then who should have been in MLB displacing white retreads. We did allow for more slots during the NeL era, but it wasn't a large number percentage-wise (the final elect-one year in the late 50s was partly due to the 'contraction' that resulted from the loss of the NeL's). On the contrary, we were inducting players like there was a full third major league.

One thing of note, though is that our MLE translations were for converting NeL numbers to white MLB. If we were to adjust white MLB numbers down, we'd have to adjust our NeL-MLE's down by the same factor.

I support and have tried to implement integrating ML and NL records, FWIW.

Yes, batting wins is not position adjusted. Certainly we should NOT use Batting Wins as a sole HoM source. It's an offense-only nunmber; I'd recommend it as a sanity check if you have hitter A even with hitter B.

Yes, McGraw looks great by this measure, being the highest 'glove' guy by far.

Yes, Indian Bob had 9 BW in the years 1943-45, which oughta be discounted.

> we'd have to adjust our NeL-MLE's down by the same factor

Or just adjust replacement level up for both sides. That may be the easiest way to approach it. Instead of 80% of positional average, perhaps it is 85% or 90% of positional average. We ought to be able to study this and figure it out based on the NgL MLE's and figuring out how many white players would have been pushed out assuming roster spots were static (no expansion).

Of course after the real integration there was an expansion. If integration had occurred earlier, would expansion also have occurred earlier? I tend to believe expansion happens based on economics regardless of talent pool. There is an argument that integration would have improved the talent level of the PCL enough to let them become a 3rd major league. There's another what-if scenario - the PCL decides to become a major league and welcomes black players in 1930. Would they have been equivalent in talent to the existing majors?

Dizzy
the level of talent among the median player, or among the lesser players was not as good in the Negro Leagues as in the white leagues.

I don't know if this is true, but I suspect it is. It might also matter whether we are talking about the Negro Leagues of 1921 or the Negro Leagues of 1937.

There was some East-West geographical division including the formation of a league mainly in the West before a league in the East followed. At times after 1923 there was again only one Negro League. Riley and others recognize a league mainly in the South for one year in the 1930s, because "most of the good players went there" (paraphrase) after the first National league went out of business. Finally there were some independent clubs of major league quality, at least until sometime in the 1930s (when that was decades past true in white baseball).

Authors of MLEs may use coefficients of translation that are uniform for all of the 1920-1949 "Negro major leagues" (is that universal here?) and zero for independent black professional clubs outside those leagues (that is not common here). But such methods are simplifications for convenience. I suppose everyone agrees that estimates of the relative strength of Negro major leagues and independent clubs are relevant. Does anyone know whether there is good evidence, or consensus, that the quality of league play was significantly higher in those years with only one league in operation?

If it is true for any league where MLEs have been developed, it has implications for the validity and analysis of Negro League statistics. If you run up your numbers against some (but not all) vastly inferior competition, how meaningful is that in translating raw numbers to major league statistics?

Constant coefficients of translation such as 0.82 for slugging average are designed to account for the difference in quality, "vast" or otherwise, between one hypothetical Negro League and one hypothetical contemporary white major league. (For some years Chris Cobb uses one of the actual white major leagues. Or is everyone doing that for some time period?)

Just a quick count in my Negro League player list has 50 players of major league caliber in 1930. I think that's low, there were 14 "big league" NgL teams - 8 in the Negro National League and 6 independents in the east. I'm going to assume you could fill out 4 rosters in the 16 team majors with black players (25%). Now your "worst 3 regulars out of 16" are no longer in the majors, and the replacement level just increased to the average of your "11th-13th best white regulars out of 16". Does that make Joe Sewell and Bill Terry look a lot less attractive?

Indeed. But there are two corrections--one for replacement level (to zero out production that would have been "under water" in an integrated league) and then a multiplier that needs to be applied to the remaining value.

The best way to do this, I think, is to construct a hypothetical integrated league of 16 teams, and then just regress everyone's stats towards league average until the hitters' RC equals the pitchers' RC allowed. If anyone could provide me with the list of players and their MLE's to do this, I could calculate the proper correction factors in no time. It would be a very valuable exercise, as the segregation penalty I use is simply guesswork and not at all quantitatively rigorous.

I don't think we've done MLE's on 100 Negro League players (or even 50) for each year. That's a big task but I think it could be tackled.

I actually don't mind the approach we seem to have taken - treat the black talent as an equivalent league. If you added 4 black teams to the majors the replacement level probably wouldn't change at all. However, we didn't do the regression of the MLB stats to league average assuming all Negro League players above replacement would have been major leaguers (expansion).

Here's another less egregious example from 2006. This is the top half of Sean Gilman's ballot. I think the commentary for Leach and McGraw is adequate, but then it degenerates. Duffy is described as the best borderline OF candidate but is placed below Dale Murphy on the ballot. The Van Haltren and Oms comments are without any substantitve meaning; he could have written "black guy" and "white guy" in lieu of his actual comments, and the ballot would be no worse.
Lets say I want to try to persuade this voter to consider a candidate I support; what is he looking for? OK, so he wants a good peak and good career, but what measure is he using to rank peaks and career? An uberstat? OPS+? All of the above? A heuristic assessment? Again, I'm not arguing that a qualitative ranking system is bad; I'm arguing that, no matter what type of system you use, your comments need to be sufficiently descriptive so that all voters understand a voter's logic, not just the ones that remember his witty posts in the Sam Crawford discussion thread back in 2003.

Well, you could have just asked. . . .

I don't have a "system" I just look at all the available evidence (mainly WARP1 and Win Shares, but also all the fascinating stuff in the discussion threads (like Dan R.'s WARP). I try to balance peak and career and keep in mind the particular weaknesses and blindspots of each stat.

I don't update my comments as often as I should, not a unique problem around here (I think the Van Haltren comment is 60 elections old). And as there's a number of candidates on my ballot I think are similar (borderline outfielders with good but not great peaks and medium length careers) my comments about them are too similar. If I rank one over the other, that's because I think the higher ranking one has a slightly better peak and/or a slightly better career. Sorry for assuming that was obvious.

'zop, I'm just wondering, what did I get classified as?

4-F?

I don't have a "system" I just look at all the available evidence (mainly WARP1 and Win Shares, but also all the fascinating stuff in the discussion threads (like Dan R.'s WARP). I try to balance peak and career and keep in mind the particular weaknesses and blindspots of each stat.

See, that's all I'm asking for! If you put that above your ballot, then everyone knows "what you're doing", and the ballot doesn't seem arbitrary...

...I didn't mean to pick on yours, I just had 6 ballots where I couldn't figure out what the motivation was and I picked out 2 of them as examples. You're right; the ballot comment thing is endemic to everyone.

I cannot figure why the Tim Raines thread is not taking posts, will look into it.

test

I had no problem posting on the Raines thread, Joe.

Am I the only guy who isn't quite as impressed with Raines as I expected?

Love the 1983-87, and the SB pct as well.

But after 1987, he only totaled 600 PA 3 times (4 if you count 1994, but who's to say he'd have been healthy anyway?). He's a mostly forgettable part-timer after age 35. His last top 10 in OBP (his bread and butter) came in 1989.

He may grab No. 1 for me because virtually everyone I like has gotten elected. And I'm not saying he's not a HOMer.
But I can see if people like someone else better.

So what am I missing?

Howie, no, I said that on my ballot. Raines is +124 WS over the position median each year. Lundy was +130 though of course that's hypothetical. Trammell was +147. Belle and Puckett +108 and +107, Ed Williamson +109. So +124 is pretty damn good, but I think you could make a good argument to have somebody else #1, if you're a peak voter, or a big glove man, e.g. I had him #1, just for the record.

if you're a peak voter
Jeez, I see Raines as a peak candidate, not a career guy. Given the really low stdev of performance in the early 80's, thats a monster peak. We've elected guys with peaks like that and barely any shoulder season.

I mean, he has both. The '83-'87 is a *superlative* peak--only 34 MLB position players since 1893 have accumulated more value over five straight years, and all of them are in the HoM or will be (Williams, Ruth, Wagner, Bonds, Mantle, Hornsby, Cobb, Morgan, Speaker, A-Rod, Collins, Vaughan, Mays, Schmidt, Musial, Gehrig, Jennings, Lajoie, Foxx, Aaron, Ripken, Yount, Banks, Ott, Boggs, Santo, JRobinson, Pujols, Bagwell, Gehringer, Groh, Delahanty, and Baker). He's a substantially better pure-peak, best-five-years candidate than pure-peak, best-five-years inductees like Kiner.

Then he has two more years as an All-Star starter-type player (1989 and 1992), two more as a definite All-Star (1981 and 1993), three more as a borderline All-Star (1988, 1990, 1991), and finally four more seasons with the value of an above-average regular (1982, 1994, 1995, 1997). So that makes him a compelling prime/career candidate. I have only 35 MLB position players since 1893 with more career value: Cobb Bonds Wagner Ruth Speaker Mays Aaron Musial Collins Williams Mantle Hornsby Lajoie Morgan Ott Henderson Schmidt Gehrig FRobinson Ripken Foxx Vaughan Mathews Yount Larkin Ozzie Kaline Trammell Boggs A-Rod Gehringer Waner Brett Rose and Crawford.

Top 35 in peak, Top 35 in career, should be a unanimous #1.

Just to follow up on posts 280-282 . . . I guess it's just me who can't post to the Raines thread. I cleared my cookies and it doesn't work either at home or at work. Very, very strange. Jim is looking into it. Obviously, from this post, I can post to other threads.

John (You Can Call Me Grandma) Murphy Posted: January 11, 2002 at 11:35 AM (#509524)

The only caveat that I would use concerning evaluating a pitcher by ERA+ is that you have to take into account the standard deviation of the league.

What a gem of an insight from all the way back in 2002. John, you sure you don't want my spreadsheet full of standard deviation-adjusted numbers? It's not too late to revise your ballot...:)

What a gem of an insight from all the way back in 2002. John, you sure you don't want my spreadsheet full of standard deviation-adjusted numbers? It's not too late to revise your ballot...:)

:-)

I actually have always used standard deviation since the Eighties to the present, Dan, though obviously my analysis still differs in some respects with yours.

What's your methodology? I thought I was the only one who cared about them.

One of the first things I looked at when we began discussing 19th century players was the standard deviation of OPS+ throughout baseball history as a less-than-perfect reflection of "league quality". The data file was posted to the Yahoo groups site many moons ago and I refer to it frequently in my own voting.

What's your methodology? I thought I was the only one who cared about them.

I just try to compare each player to his era using SD, that's all, Dan. If you remember, that's one reason why I wasn't as crazy as some about the pitchers from the Seventies. Of course, I felt there was a logical reason for their longevity; regarding Richie Ashburn, I didn't blow him off because he happened to have 3 great center fielders competing against him (though I tried to find a middle ground with him).

I was just asking whether it was a rigorous quantitative measurement or more of touch-and-feel thing. Also, if you use actual SD's, then you run into the problem of being misled by random star gluts or droughts (what sunnyday cautions about). That's why I did a bunch of regressions to find out what actually causes changes in SD, and use those equations' predicted SD to make adjustments, rather than use real ones. (For example, the 1941 AL wasn't particularly easy to dominate, it just happened to have three insane yaers by Williams, DiMaggio, and Travis, while the 1962 AL was extremely easy to dominate but no one took advantage). Also, if you do it in terms of decades/eras rather than yera-by-year, you lose the individual seasonal fluctuations (expansion, run scoring etc.) that can make a big difference.

If you're interested in the SD question, you should check out the StDevs and Rep Levels.xls sheet in my .zip file in the Yahoo group.

What is the logical reason for the longevity of 1970's pitchers?

I was just asking whether it was a rigorous quantitative measurement or more of touch-and-feel thing.

Somewhere in the middle, Dan. :-) It's not as rigorous as I would like due to time constraints, so I have to do some eyeball analysis at certain times.

What is the logical reason for the longevity of 1970's pitchers?

The pitchers who thrived tended to have lighter workloads during their formative careers, which happened to be in the pitcher-friendly portion of the Sixties. I honestly don't think that would have been the case if they had started their careers in the Seventies with the higher offensive levels. With that said, I still supported all of the inductees except for Sutton, so it wasn't as if they weren't of HoM quality.

If you're interested in the SD question, you should check out the StDevs and Rep Levels.xls sheet in my .zip file in the Yahoo group.

I actually have and have made a couple of decisions based on them. I appreciate your work there.

Well, this is too late to influence half the ballots or more, but I like this comparison so I'm going to post it one more time. Joe Medwick and Bob Johnson were exact contemporaries, so era comparisons aren't an issue, and I don't know that either league was considered significantly better in this era.

If you take off 1932 and 1946-8, when Medwick was playing but was essentially useless (12 WS in those 4 years), by WARP1/WARP3/WS you have Johnson at 100.7/94.3/287, and Medwick at 95.8/95.2/300. So they're more or less even under WARP, while under Win Shares Medwick has an advantage of 1 WS/year. But Medwick was playing mostly on very good teams, and more importantly Johnson was playing on terrible ones, which has been shown to distort WS because those teams play a harder schedule than the rest of the league. Johnson's OPS+ is higher, 138 to 134, although cutting off the dross would probably bump Medwick's up a couple of points. I think it is fairly clear they are essentially even on career value.

OTOH, there is no argument that Medwick had a higher peak. His 1935-37 was significantly better than anything Johnson did (all over 10 WARP/30 WS, while Johnson's highs were 9.7 and 31, the latter in a war year). On the other hand, Medwick's 1943 and 1945 were significantly worse than Johnson's lowest, and that was during the war! I would also argue that Johnson was remarkably consistent, never having a below-average season, and this should mitigate, to an extent, Medwick's peak advantage. But Medwick has an advantage - he should be ahead of Johnson. I just don't think it's that huge.

And the other element, that may or may not make up for Medwick's peak advantage, is Johnson's minor league credit. Even if you just give him 1 year of credit, he would have a clear edge in career value on Medwick. The inarguable fact is that Johnson was able to contribute as a major leaguer until he was 38, while Medwick was essentially done by 32.

If Medwick is ahead of Johnson, it's not by very much. If you think Joe Medwick is a Hall of Meriter (and plenty of people did), then you really should have Bob Johnson in as well. I do understand if you think neither one should be in because the 1930s are overrepresented - I disagree, but I'm not debating that point. But I do feel that they should either both be in or both be out, or Joe Medwick has to be right on the edge of your in/out line.

(Now, having said all that, if you're only going to vote for one OF besides Raines, it should be Reggie Smith. I just don't have a good comparison for him.)

Medwick was a mistake.

I'm not crazy about either one, but if I had to choose one, I would take Johnson.