I think that for posterity we put Irvin as an LF, even if it takes a couple of weeks and even if we vote on him here.

the "NL dominance," [in HOM active players] by this measure, takes a little longer than some might think
Al leads until 1951, NL basically catches up, but NL doesn't establish lead until 1960 and doesn't start utterly dominating until 1964-71, when it's an amazing imbalance

1952 NL (12/14/14)
Brooklyn - C Roy Campanella, 2B Jackie Robinson, SS Pee Wee Reese, OF Duke Snider
New York - RP Hoyt Wilhelm, OF Willie Mays*, OF Monte Irvin*
St. Louis - OF(1B) Stan Musial, OF Enos Slaughter
Philadelphia - SP Robin Roberts, OF Richie Ashburn
Chicago
Cincinnati
Boston - SP Warren Spahn, 3B Eddie Mathews
Pittsburgh - OF Ralph Kiner

1952 AL (11/12/16)
New York - C Yogi Berra, 1B Johnny Mize*, OF Mickey Mantle (Charlie Keller 1 G)
Cleveland - SP Bob Lemon, SP Bob Feller, SP Early Wynn, OF Larry Doby (Quincy Trouppe 6 G)
Chicago - SP Billy Pierce, 2B Nellie Fox, OF Minnie Minoso
Philadelphia
Washington
Boston (Ted Williams 2 G, Lou Boudreau 2 G)
St. Louis - RP Satchel Paige
Detroit - SP Hal Newhouser

I think if you look at the actual numbers that these players put up in 1952, you'll see that whereas most if not all of the NL players were putting out peak-like performances, many of the AL players were either at the end of their careers (Mize, Feller, Trouppe, Boudreau, Paige, Newhouser). Not to mention that in hindsight the NL group was far superior overall.

The point is this: You can't just count HOM numbers and form much of a conclusion about the relative league quality. A better method would be to count the number of HOM players who were at, or nearly at, the peak of their careers. A variant of that would be to count the number of A-level HOFers in each league who were at their peak in any given year. These give you a truer indication of how the leagues compared. And if you use this method, the NL's marked superiority beginning in the early 50's jumps out at you far more clearly.

I agree with that point, basically - which is why I specifically said "BY THIS MEASURE, takes a little longer than some might think" and why I listed some of the examples as I did so people could examine that point.

On the other hand, at 27-10, it's not likely it's just going to be all about old-timers hanging on, either.

In the same spirit, listing other lopsided TOTAL HOMers years since 1900 (not listing those right next to others already listed or listed here):

1901 NL (19/20/21)
Pittsburgh - SS-OF(2B) Honus Wagner, OF Fred Clarke
Philadelphia - 1B Hughie Jennings, OF-1B Ed Delahanty, OF Elmer Flick
Brooklyn - 1B Joe Kelley, SS Bill Dahlen, OF Willie Keeler, OF Jimmy Sheckard
St. Louis - SS Bobby Wallace, OF Jesse Burkett
Boston - P Kid Nichols, OF Billy Hamilton
Chicago - P Rube Waddell, 2B Cupid Childs*
New York - P Christy Mathewson, SS George Davis
Cincinnati - 1B Jake Beckley, OF Sam Crawford (Amos Rusie 3 G)

1901 AL (7/7/7)
Chicago - P Clark Griffith
Boston - P Cy Young, 3B Jimmy Collins
Detroit
Philadelphia - P Eddie Plank, 2B Nap Lajoie
Baltimore - P Joe McGinnity, C Roger Bresnahan
Washington
Cleveland
Milwaukee
........................


1939 NL (11/11/11)
Cincinnati (Al Simmons 9 G/2T)
St. Louis - 1B Johnny Mize, OF Joe Medwick, OF Enos Slaughter
Brooklyn
Chicago - C Gabby Hartnett, 2B Billy Herman, 3B Stan Hack
New York - P Carl Hubbell, OF Mel Ott
Pittsburgh - SS Arky Vaughan, OF Paul Waner
Boston - Al Simmons(2T)
Philadelphia

1939 AL (16/17/21)
New York - P Red Ruffing, C Bill Dickey, 2B Joe Gordon, OF Joe DiMaggio, OF Charlie Keller (Lou Gehrig 8 G, Wes Ferrell 3 G)
Boston - P Lefty Grove, 1B Jimmie Foxx, 2B Bobby Doerr, SS Joe Cronin, OF Ted Williams
Cleveland - P Bob Feller, SS Lou Boudreau*, OF Earl Averill(2T)
Chicago - P Ted Lyons, SS Luke Appling
Detroit - 1B Hank Greenberg, 2B Charlie Gehringer, OF Earl Averill(2T) (Hal Newhouser 1 G)
Washington (Early Wynn 3 G)
Philadelphia
St. Louis

..........

1984 NL (13/14/14)
EAST
Chicago - SP Dennis Eckersley**(2T), 2B Ryne Sandberg
New York - 1B Keith Hernandez
St. Louis - SS Ozzie Smith
Philadelphia - SP Steve Carlton, 3B Mike Schmidt
Montreal - C Gary Carter, 1B-OF Pete Rose(2T), OF Andre Dawson, OF Tim Raines
Pittsburgh

WEST
San Diego - RP Rich Gossage, 3B Graig Nettles, OF Tony Gwynn
Atlanta
Houston - SP Nolan Ryan
Los Angeles
Cincinnati - 1B Pete Rose(2T)
San Francisco

1984 AL (23/25/27)
EAST
Detroit - 2B Lou Whitaker, SS Alan Trammell, DH-1B Darrell Evans
Toronto - SP Dave Stieb
New York - SP Phil Niekro, 2B Willie Randolph, OF Dave Winfield
Boston - 3B Wade Boggs, OF Dwight Evans (Dennis Eckersley** 9 G) (ROGER CLEMENS*)
Baltimore - 1B Eddie Murray, SS Cal Ripken (Jim Palmer 5 G)
Cleveland - SP Bert Blyleven
Milwaukee - SP Don Sutton, RP Rollie Fingers, 3B-DH Paul Molitor*(11G), SS(DH) Robin Yount, DH-1B Ted Simmons

1984
AL WEST
Kansas City - SP-RP Bret Saberhagen*, 3B George Brett
California - 1B Rod Carew, 2B(13) Bobby Grich, DH Reggie Jackson
Minnesota
Oakland - 2B Joe Morgan (RICKEY HENDERSON)
Chicago - SP Tom Seaver, C Carlton Fisk
Seattle
Texas

Marc sunnyday posted his prelim (#50).
#80-83:
I challenged him on ranking Roush 12, a gap ahead of the pack, where whoever ends up in ranks 13-14 will feel like overrated. Paul Hines and Geroge Gore are two in that pack. I presented their fielding careers in terms of playing time and their "Rate" in CF according to Clay Davenport.


Marc replied, in part (#85-86)
As to the 19C guys, i.e. where do Hines and Pike and Gore and O'Rourke go relative to one another, after Hamilton at an obvious #1, I dunno, is Hines obvious at #2?

Oh, and as for Roush, I don't know why that's unexpected, he was on my ballot for 50 years. I think you could argue that he was the best NL position player of the '10s. Of course, you could argue that for a lot of guys.


Roush played only four NL seasons in the '10s; six seasons including the Federal League and two of the six were short in playing time ('14, '16). Even if you favor Roush by starting the "'10s" in 1914 or 1916, how does he match Benny Kauff (another CF with two FL and four NL seasons)? Nor is it clear he was better than Hornsby in the league and Groh on his team during 1916-19, his four NL seasons in the decade.

That paragraph essentially repeats #90.
As yet I have only approached Marc's question about the 19th century CFs.
> As to the 19C guys, i.e. where do Hines and Pike and Gore and O'Rourke go relative to one another,
> after Hamilton at an obvious #1, I dunno, is Hines obvious at #2?

Mike Webber was the campaign manager for Roush. 2006-08-25 in the Edd Roush thread he re-posted the all-time rankings of CFs by three others whom he recognized as supporters. He hoped that their rankings of Roush around all-time 15 would contribute to his cause. Marc is one of the three.

Here are the crucial portions of their all-time rankings which highlight where those three supporters of Roush ranked the 19th century centerfielders Hines, Gore, and Hamilton.

==
1.

Eric Chalek 2006-08-23 covered "MLB-only CFs through 2005"
(but I suppose he includes Larry Doby's whole career)

Eric Chalek, I
6 Hines, P
7 Gore, G
8 Snider, D
9 Hamilton, B
10 Duffy, H
11 Doby, L
12 Browning, P
13 Averill, E
14 Griffey, Jr
15 Van Haltren, G
16 Roush, E

Eric Chalek, II "keltnerized"
6 Hamilton, Billy 66
7 hines, paul 62
8 Griffey Jr., Ken 54*
9 Snider, Duke 48
10 gore, george 46
11 roush, edd 43
12 Browning, Pete 43
13 duffy, hugh 42
14 doby, larry 36

2.

Marc Sunnyday 2006-08-23 covered MLB CF through "about 2000" in some integrated fashion
with parenthetical remarks on NeL CF except Stearnes is missing.

>>
[Inner Circle - the familiar six including Charleston]
Lock

[Griffey probably here, accord'g to accomp'g prose]
Snider 249
Puckett 211--aside from the inner circle, TPR has Puckett ahead of everybody but Wynn and Ashburn, and his combined HFM, HFS, BI and GI is pretty good; no real weaknesses in this system

Show Cause Why Not

Dawson 196--a HoMer in my book
Hamilton 195
Browning 187
Hines 186
Roush 184
Carey 176
Wilson 176--e.g. pretty borderline, one-dimensional
Gore 175
(Torriente probably near the top of this group)

Show Cause Why

Doby 173--e.g. added NeL value, that's why [ie, this is without NeL credit]
<<

3.

Andrew Siegel 2006-08-23 covered all CF through Griffey but Stearnes is missing.

[the familiar six]
(7) Hamilton
(8) Snider
(9) Griffey
(10) Torriente
[gap]

(11) Doby
[gap]

(12) Hines
(13) Gore
(14) Averill
(15) Edmonds
(16) Roush

==
Those "gaps" above and below Doby at #11 match Marc's 2008 ladder (sunnyday # )
except Marc now couples Roush with Doby, a gap ahead of the pack, and he lists
Stearnes and doesn't count Griffey in the ten who are a gap ahead of Doby.

Those selections are directly from Edd Roush #45-48 by Mike Webber, indirectly from then-current Jimmy Wynn thread. Mike selected them because he hoped that their support would help revive Roush's case.

Edd Roush 48. Mike Webber Posted: August 25, 2006 at 03:09 PM (#2156710)
I hope Doc, Marc, and Andrew don't mind me reposting their comments here, but hopefully in a couple of election cycles - it is "1984" as I type this - I think Edd being in the top 15 on these lists will be arguements in his favor. Especially since 2 of the three incorporate Negro Leaguers, and one includes active players.

Being the 15th best centerfielder ever is a pretty stong arguement in my book. Still borderline, but on the "in" side of the line.

==
Damn this website! I don't have the highlights quite right in the preceding article. Roush and the author should be bold. At least I have the Hines, Gore, and Hamilton consistently underlined.

No, I don't think Hamilton is the obvious #1 among 19c CFs. He is one plausible choice, Hines is another, not Gore. Strong "careerists" will prefer O'Rourke to both Hs. Pike is not in the discussion at this level. Evidently only one of our three Roush supporters considered Hamilton > Hines obvious (Andrew).

None of them put Roush ahead of Hamilton or Hines and only put him just ahead of Gore (Marc). Let me insert Charleston, Stearnes, and Torriente above Roush where necessary, and delete the active or recent players . . .
Two years ago our three Roush supporters would have ranked him thus in the present CF poll.

Eric Chalek, 19 and 14
Marc Sunnyday, 15 (since then he has passed Dawson, Browning, Hines; they were all in the pack and Roush has left the pack behind a gap)
Andrew Siegel, 15

That is about what I would expect from his best friends.

HOM CenterFielders who are not in the Hall of Fame
How did they fare in our "Group" polls?

Group1 (12)
- 11, Dawson

Group2 (12)
- 10, Wynn

Group3 (21)
- 3, Hines
- 5, Gore
- 21, Browning

Group4 (12)
- 10, Pike
- 12, Oms

Revised Prelim

I'd like to vote now but I really want to wait until I see what Chris comes up with for MLEs and how Dan R translates them to WARP. The downballot slots are close enough that there numbers will have a large impact.

1) Cobb (and I can't stand the guy)
2) Mays
3) Speaker (best defensive CF ever?)
4) Mantle
5) Charleston
6) DiMaggio
7) Stearnes
8) Hamilton
9) Torriente
10) Max Carey (BFF?)
11) Paul Hines
12) Duke Snider
13) Ashburn
(Tommy Leach if he's a CF)
(Reggie Smith)
14) Pete Hill (lots of uncertainty & doubt)
15) Jim O'Rourke (could be better but I think he's behind Hines)
16) Doby
17) Gore
18) Cool Papa Bell (in/out)
19) Alejandro Oms (could move, will be PHoM within 5 elections)
[Dom DiMaggio] (out)
20) Earl Averill
21) Jim Wynn
22) Edd Roush
23) Andre Dawson
24) Lip Pike (career is too short)
25) Pete Browning - mistake
26) Willard Brown - a big time mistake. Offensively he's Andre Dawson. Defensively he was good, indifferent or awful depending on how he woke up that day. Attitude was a big minus. HoM talent/potential but most evidence suggests he never lived up to that talent.

until I see what Chris comes up with for MLEs

I'lll try to get updated MLEs for Bell and Brown done over the weekend, along with FWS for them and for Torriente, who also needs career XBH. Oh, and Doby has only a few seasons: he'll be quick to do.

Oms I doubt I will get to: integrating CWL data and NeL data takes a lot more time. I could do XBH based on current translations, and then Dan R would have enough data to do WAR for Oms if he wants. We'll see. Now that my semester is underway, time is a lot more scarce than it was recently.

I don't know if Speaker is the best defensive CF ever by *ability*, but I don't think there's much doubt that he was by *value*. A superlatively gifted CF with the brilliance to play ultra-shallow in the deadball era could simply make more of an impact than the best fielding CF once the lively ball was implemented. In the same sense, the greatest K pitcher of all time was probably Dazzy Vance, but the one whose K ability had the most value was either Ryan, Clemens, or the Unit.

That actually opens up some tricky cross era comparisons. Let's say, for the sake of argument, that Speaker could be +30 runs every on defense (I don't think that he was, but I do remember Dan mentioning that some systems like DRA have CRAZY SD's in early baseball), while a modern center-fielder could only be, say, +15 runs at most. Is it fair to give Speaker the extra credit for those runs? They obviously produced real wins. I guess this may come down to value vs. ability as merit, but either way there could be a "fairness to all eras" consideration here as well. SD adjustment may mitigate the issue to some extent, but I am not sure it would remove it entirely.

I find it, um, interesting that most of us have come down on the "ability" side when discussing DH-vs-no-DH players; we recognize Win Shares undervalues modern AL batters compared to NL. Extending this argument a bit further, if the NL decided to go a 8-man game (take away one OFer or an IFer from the defense), this would increase the value (in runs) for each batter who would now come to the plate 80ish more times a year, without a smidgen change in his ability.

But... more than half our voters seem to edge toward the "value" side of the argument when it comes to park effcts (G Cravath) or changes in the game (Speaker's CF defense?).

Anyone wish to plead "guilty as charged, your honor, but the charge is irrelevant and should be dismissed?"

Yes, there are definitely some crazy outliers. How 'bout a minus-39 for Joe Jackson in 1919? Maybe he was on the payroll for the regular season too! :) (FRAA has him at a much more modest -12). And if Ed Delahanty was really +44 in 1893, or Flick +38 in 1901, or Sheckard +66 over 1902-03 combined, then...well, wow. Clearly, some of this is pitcher BABIP prevention being credited to the fielders. But another chunk of it is that defense mattered more back then. There were many fewer strikeouts and home runs, increasing the number of balls in play and their relative importance in run scoring. Furthermore, errors were much more common, meaning that an extremely sure-handed defender could save many more runs relative to the average in those days than one can today.

As for how to address it: there were still no more wins above replacement to go around in the deadball era than there are today (after accounting for overall standard deviations). If fielding mattered more, that means pitching mattered less--certainly, if you assigned all deadball pitchers the same BABIP and just calculated a defense-independent ERA for them using their K BB and HR rates, they'd all look somewhat similar. The logical conclusion would be to elect more position players and fewer pitchers from that era. However, the flip side is that--indeed perhaps because they had to exert less effort, since they relied so much more on their defense--pitchers threw far more innings in those days than they do now. So although the total pool of WARP is the same, it's being divvied up among fewer players, thus leading to more of them having HoM-worthy statistics as a percentage of the total. I don't think there's any one right way to correct for this; it's just a voter preference issue.

Tom, adjusting for the DH is only a value vs. ability question if you use a 'tarded system like Win Shares. With my approach, every player is compared to a replacement at his position, meaning that the inclusion of value for DH's does not lead to a reduction of value for other position players--it just means that the hitting WARP that were previously going to pitchers who hit now go to DH's. Yes, if they went to an 8-man game, that would increase the value of each position player, but there would also be fewer position players, so the total # of HoM-worthy position players would not change.

Let's say, for the sake of argument, that Speaker could be +30 runs every on defense (I don't think that he was, but I do remember Dan mentioning that some systems like DRA have CRAZY SD's in early baseball), while a modern center-fielder could only be, say, +15 runs at most. Is it fair to give Speaker the extra credit for those runs? They obviously produced real wins.

I think it's almost imperative to give him the 'credit' (it's not EXTRA credit). Otherwise, you end up with possible crazy results, like Speaker was the best player in MLB in 19xx, BUT for HOM purposes isn't even the best player on his own team, etc.

I know I've posted about this before, but I'm really not feeling the preference for Mantle over Speaker, and I'd like to ask every voter who puts the Commerce Comet over the Grey Eagle to reevaluate. Let's compare here, with a war-credited Mays thrown in for good measure. (For Mays, I'm using a DRA/TotalZone hybrid for defense and Dan Fox's EqBRR for baserunning to be as favorable to him as possible. For Mantle, there's no significant difference between those figures and the FRAA/Fielding WS numbers, so I've stuck with the originals).

Speaker



3-year peak: 32.8
7-year prime: 68.7
Career: 145.4
Salary: $502,506,620


Mantle



3-year peak: 35.6
7-year prime: 71.6
Career: 123.7
Salary: $429,078,376


Mays



3-year peak: 31.4
7-year prime: 67.8
Career: 161.4
Salary: $555,926,690

Ranked by 3-year peak: Mantle 35.6, Speaker 32.8, Mays 31.4
By 7-year prime: Mantle 71.6, Speaker 68.7, Mays 67.8
By Career: Mays 161.4, Speaker 145.4, Mantle 123.7
By Salary: Mays $556M, Speaker $502M, Mantle $429M

OK, if you're a pure peak/short prime voter, I get it. But then you should have Mantle #1 or #2 (Cobb's peak was roughly as stratospheric).

What I don't understand are the Mays-Mantle-Speaker placements. The gap in career value between Mantle and Speaker is enormous--Speaker's offensive value above average was the same as Mantle's, and then he tacks on a mere twenty-two wins of defense. Put another way, Mantle and Speaker had roughly equivalent career value through age 33 at 110 WARP2 apiece. After that, Mantle was reduced to three banged-up years, two of them at first base. By contrast, Speaker promptly hit .378/.474/.606, .380/.469/.610, .344/.432/.510, and .389/.478/.578 the next four seasons, while sticking in CF. We're not talking about meaningless late-career stat-padding here. We're talking about an extra half-decade of MVP-caliber performance. I find it hard to swallow that one measly extra win per season over the top three years is sufficient to counteract over Twenty wins after age 33. It just doesn't pass the smell test. And again, if you really think that way, then why do you have Mays above Mantle?

It seems to me that people are implicitly dinging Speaker on a timeline basis. In terms of the ease of domination of their leagues, there is simply no empirical justification for this. The regression-projected standard deviation of Mantle's leagues was 95.7% of the 2005 level, while in Speaker's it was 94.5%. (The late aughts and late teens AL had quite low standard deviations). It's just not true that it was easier to rack up X Win Shares/WARP in Speaker's time than in Mantle's. And in terms of league strength, sure, Speaker didn't have to compete with black players--but with a handful of exceptions, Mantle didn't either. Moreover, Speaker was playing in by far the stronger league during the teens, battling with Cobb and Collins for MVP honors every year, while the NL only had Cravath and George Burns to write home about. By contrast, Mantle was playing in the far weaker league during his time, with the black stars clustered in the NL. So if there's a league strength case to be made, it has to go to Speaker, I think.

Does your ballot rank Mays ahead of Mantle ahead of Speaker? If so, please stick your neck out to justify your placement!

Upon re-reading my rambling stream-of-consciousness post #70 on this thread, I realize that Dan isn't talking to me.

Speaker is an interesting case. He does quite well in many advanced stats, including win shares, yet I believe there's a reluctance to believe he was as good as his stats suggest. There are a few good reasons for this. First, he was not offensively dominant. Very good, even excellent, but not historically great. He rarely led the league in offensive categories, although he was always among the leaders. Mantle, on the other hand, was historically dominant - he didn't just lead the league, year after year, in the most important offensive stats, he utterly dominated his leagues in a few of those years at nearly a Ruth/Williams/Bonds kind of level. It is reasonable based on this to think that Speaker's defensive advantage can't make up for this - keep in mind that defensive statistics are probably more open to interpretation than Dan would suggest.

So this leads to the reasonable assumption that Mantle was a good deal better than Speaker at their respective peaks, and that makes up for Speaker's career advantages, leading to the ratings.

I do not believe, for a second, that the American league of the 1950's and 1960's was inferior in terms of the level of competition than the level of competition during Speaker's time. The fact that the American League was much slower to integrate than the NL during the 1950's and 1960's should not be considered a point in Tris Speaker's favor. It is reasonable to say that the AL was the stronger league in Speaker's time, and the weaker league in Mantle's, but it is a logical fallacy to then assume it means that the AL of 1915 was stronger than the AL of 1955.

I also don't believe it was harder to dominate the AL of 1915 than the AL of 1955, no matter what Dan's analysis tells him. The AL of 1915 had several players putting up statistics dwarfing the rest of the league, including Cobb, Speaker, Collins, and Johnson. The only one putting up numbers even close to Mantle in the 50's AL was Ted Williams, of which nothing needs to be said.

Overall I am confident that Mantle was better than Speaker in terms of peak value, and Speaker was better in terms of career value. I'm guessing a lot of people would agree with me. Putting Mantle over Speaker overall is eminently justifiable by the numbers (although I can see an argument for the opposite), and no one needs to stick their necks out to justify it.

Speaker rarely led the league in offensive categories because he played in the same league as Ty Cobb, Dizzypaco. There is no doubt that Mantle was a superior hitter at his peak, but I do not see why you say it is "reasonable...to think that Speaker's defensive advantage can't make up for this." Of course defensive stats from the teens are less reliable than the ones we have today. But we certainly do know that Speaker was universally regarded as the greatest defensive outfielder the game had ever seen until Willie Mays. Speaker's own fielding stats simply back up what his reputation was already telling us. We also know from modern play-by-play statistics that an otherworldly defensive CF (Andruw Jones or Darin Erstad) most definitely can save over 20 runs a year. And if you add 20 runs of fielding on to Speaker's hitting, you get a peak that's about one win a year below Mantle's--enough to be noticed, but hardly a huge advantage for Mantle.

In terms of overall quality of competition, I'm sure you're right that the 50s AL was tougher than the teens AL, but we have a strict no-timelining policy at the HoM, do we not? The point is that Speaker was playing in the much stronger league of his time, while Mantle was playing in the much weaker league of his time. It's on those grounds that a constitutionally acceptable league strength argument would have to go to Speaker.

Again, the reason why everyone was tearing up the teens AL and only Mantle and Williams did so in the 50's AL is precisely because these were two of the periods of greatest league strength disparity in major league history. You're focusing on Cobb/Speaker/Collins but ignoring the NL, where MVP awards would have gone to the likes of George Burns and Gavvy Cravath. And you're not comparing Mantle to Mays, Aaron, Banks, F. Robinson, Mathews, Musial, etc. There were far more inner circle HoM position players in their prime in the 1950s (Williams, Mays, Musial, Aaron, Mantle, Mathews, Berra if you give a big catcher bonus, and the beginning of F. Robinson) than there were in the teens (Cobb, Speaker, Collins, the end of Wagner and Lajoie, the beginning of Hornsby). If anything, the argument goes the other way!

Yes indeed, Mantle was better than Speaker in terms of peak value, and Speaker was better in terms of career value. But Mantle's advantage in peak is marginal, while Speaker's advantage in career was enormous. In magnitude, it's equivalent to ranking Nellie Fox above Lou Whitaker, or Harlond Clift above Paul Molitor, or Elmer Flick/Joe Kelley over Al Kaline (a difference of about +3 on 3-year peak and 7-year prime, and of -22 on career). It's logically consistent for a pure peak voter, but it's just too far from the reality of winning baseball games to be credible.

From Post 119: First, he was not offensively dominant. Very good, even excellent, but not historically great. He rarely led the league in offensive categories, although he was always among the leaders.

Umm, he rarely led in categories because he was competing with Ty Cobb in the Teens and Babe Ruth in the 20s. Those two just happen to have the led league in the most offensive categories. First and second in black ink? Ruth 161 points and Cobb 150. Despite competing with those two players for his entire career, Speaker is still 46th in career Black Ink. Tied with? Joe Dimaggio.
The only player whose career overlaps Cobb and Ruth who is above Speaker is Gehrig. But if you look at Gehrig's Black Ink it demonstrates how difficult it was to get Ink competing with Ruth/Cobb. Almost all of Gehrig's Black Ink is from 34 to 38. The Black Ink before that was the result of historically great XBH years or RBI totals from driving Ruth in.

Looking at Grey Ink, Speaker is 6th all-time with 346. Most of those are top 5. Speaker was top 3 in AVG 10 times in 16 years. He was top 4 in OBP 13 years out of 16. And top 5 11 years out of 16. I don't think there are many players with that profile. And at this same time, he was a Gold Glove defender most every year.

Oops, I forgot a term. Speaker was top 5 in Slugging 11 years out of 16.

What Dan said.

Black Ink / Grey Ink (place)
Cobb: 150 / 417 (2nd / 1st)
Mantle: 65 / 272 (14th / 17th)
Mays: 57 / 337 (19th / 8th)
Speaker: 34 / 346 (46th / 6th)

Ty Cobb: Just wow. If you cut Cobb in half, his Black Ink would be 2 players tied for 10th (or Lou Gehrig) and his Grey Ink would be 34th and 35th (or Alex Rodriguez and Billy Williams).
The top 6 in Black Ink are so far ahead of the rest: Ruth 161, Cobb 150, Hornsby 125, Williams 122, Musial 116, Wagner 109. Then Brouthers is 7th with 79.

I know Black Ink / Grey Ink have their faults, but it is amazing to see how dominant some of these players were. Alex Rodriguez will tie Gehrig for 10th if he holds on to his SLG lead and can finish first in HR.

CF is strange with how severe the talent disparity can be. After Cobb and Speaker dominate the position for 20 years, Earl Averill comes in and per both Dan WARP and WS is consistently the best CF in the AL. I think 9 times by WARP and a similar number by WS. I am at work without my stats so going by memory. The problem was there wasn't any competition so Averill was an "All-Star" almost by default. In some years, Averill was the best despite having less than 4.0 WARP. He wasn't having great years, but he was the best in the league.

Yeah, I mean, Dizzypaco's argument is no different than dinging Foxx and Greenberg for playing in the same league as Gehrig, or Yount and Trammell for playing in the same league as Ripken, or Jeter for playing in the same league as A-Rod. Sometimes you'll just get random gluts of great players in the same league at the same position, just like you'll get random droughts (like 1950s 1B). They don't say anything about the Merit of the players involved.

The correct approach to dealing with this is, as I've said some umpteen zillion times,

1. Assess the overall ease of domination of a league based on its intrinsic characteristics (like run scoring and years since expansion), rather than on how specific players actually happened to perform in it. This enables you to distinguish between star gluts (the 50s NL) and droughts (the teens NL) and leagues that were actually easy to dominate (the 1998 NL) or hard to (the AL around 1905).

2. After adjusting for ease of domination, compare players to their positional peers by looking at the broad bottom of the distribution (I look at the bottom 3/8), which is much less susceptible to random fluctuations than the overall average (like RCAP) or particularly the top.

Question: if Reggie Smith gets elected to the HoM in 2009 (which seems quite possible), which positional pool would he belong in? CF or RF? The total games played are roughly similar. He was primarily a CF in Boston, and that's presumably when he defensive value was highest. On the other hand, for those couple of years in Los Angeles when his offensive rate stats went through the roof, he was in RF. I don't know what he played in Japan.

The possibility of him being elected very soon makes me wonder about his placement on the appropriate list.

I did forget to mention Lloyd as an inner circle HoM position player whose prime was in the teens.

Part of the Ink phenomenon is just expansion--it's twice as easy to lead an league with eight teams than one with 16...

I thought I would look at players who spent most of their careers with 24 teams or more for the B.I. and G.I.

Here are the hitters in the top 30 of Black Ink who are 1969 and later:

11. Schmidt - 74
12. Bonds - 69
13. Rodriguez - 68
15. Rose - 64
20. Gwynn - 57
22. Yastrzemski - 55
28. Henderson - 50

Here are the hitters in top 30 of Grey Ink who are 1969 and later:

1. Cobb - 417
...
5. Anson - 358
...
10. Hornsby - 329

14. Bonds - 289
24. Rose - 239
30. Schmidt - 224
(Alex Rodriguez is tied for 35th with 208. He should be in the top 30 after this season.)

Here are the pitchers in the top 30 of Black Ink who are 1969 and later:

1. Walter Johnson - 150
2. Alexander - 126
3. Grove - 108
4. Spahn - 101
5. Young - 100

6. Clemens - 100
7. Randy Johnson - 96
10. Maddux - 85
11. Ryan - 84
16. Carlton - 66
21. Seaver - 57
22. Pedro Martinez - 55

Here are the pitchers in the top 30 of Grey Ink who are 1969 and later:

1. Young - 472
...
5. Alexander - 339

6. Maddux - 333
8. Clemens - 314
10. Seaver - 296
12. Carlton - 282
13. Johnson - 277
17. Perry - 253
19. Ryan - 251
23. Mussina - 244
24. Sutton - 240
25. Blyleven - 237
(Pedro Martinez is currently 31st)

Simple check on ease of domination:

Tris Speaker was 3 times the best hitter in his league (by BBref's batting wins). Four times he was 2nd. Two of those four were Ruth and Cobb years, so if you wanna say "but he played with Ruth and Cobb", you could argue he was five times the best. Of couse maybe a Negro Leaguer might have beaten him if they had the chance.

Mantle had 9. And some of them weren't 'squeak over the line' years, they were blow-em-all-away seasons.

I don't have the stats with me right now; I'll get them in a day or two; but Speaker also had a huge home park advantage, more than typical for his teammates. Yes, real value came from this, but in a neutral setting... well, I guess I'd say if I were managing a pre-1920 ball team, I'd take Speaker, but in the live ball era, I have the Mick higher.

TomH:


1. How many times do I have to say that you simply cannot compare Speaker and Mantle just to their AL peers, because Speaker played in perhaps the strongest AL (relative to its contemporaneous NL) ever, while Mantle played in perhaps the weakest AL (relative to its contemporaneous NL) ever? Please, pretty pretty please, can we never ever mention Speaker and Mantle's ranks in the American League ever again?? Can I get these comparisons banned? I BESEECH YOU ALL, if we are going to talk about ranks, use the major leagues. Mays and Aaron and F. Robinson and Musial and Banks and Mathews were all playing at the same time as Mantle, you know. You can't reward Mantle because his league was slow to integrate.

2. Of course Mantle was a better pure hitter at his peak than Speaker was! I have never suggested otherwise. The peak gap on offense ALONE is indeed significant.

3. But again, as I seem to still have to insist at every opportunity, catching the ball also forms part of this sport! There is no question that Speaker is one of the two greatest defensive center fielders of all time. He was universally regarded as such, and the quantitative evidence at our disposal (be it FRAA, FWS, or DRA) completely supports that reputation. Moreover, modern play-by-play defensive stats certainly can tell us the overall magnitude of just how much of a difference a superlative center fielder can and cannot make on a team's bottom line of runs and wins. I am not giving Speaker any kooky +35 seasons here or anything--I only have him at +10 a year, peaking at +24. Guys put up +24 seasons in CF all the time these days, and the best CF of our day (Andruw, Erstad) did so routinely at their peaks. Do we really think that the man universally acclaimed as a top-two all time fielding CF never would have broken +20? Well, once you tack on a few +20 defensive seasons to Speaker's hitting, then the gap in peak between him and Mantle narrows substantially. Mantle still does retain a small peak/prime edge.

4. My point was never that Speaker's peak was actually equal to Mantle's. It wasn't. I am only saying that the gap between their peaks is rather small, while the gap in career value is immense, and that you have to be an almost religiously devoted pure peak voter to not see the forest for the trees on this one. Again, it's the same as putting Harlond Clift over Paul Molitor. No, Molitor never put up a season like Clift's '37 or '38. But really, come on.

5. Synthesizing all this, here are Speaker and Mantle's major--not American--league ranks in total value (hitting plus baserunning plus fielding) for every season they were top 10, year-by-year, along with the portion coming from each type of win above average, and the players who topped them for the years they ranked #2 through #5:

Speaker




Mantle



Looks a little different when you include those NL players too, dunnit? Speaker and Mantle were each the best player in baseball three times, the second-best player twice, and the third-best player once. After that, it's all Speaker: he has four fourth-place finishes to Mantle's three, they each have one fifth-place showing, and then Speaker has another six top-ten placements to Mantle's one. If anything, if you're talking about rank in the majors, Speaker was more dominant than Mantle! (He wasn't, of course; this is a gimmicky approach, but it's the one that TomH is advocating).

6. I have no idea about Speaker's home/away splits. But he split his career between two very different parks: one was short to left field, the other was short to right. If he retained a substantial home-field advantage throughout his career, that just suggests he was a great enough hitter to adapt to his surroundings and take advantage of them, rather than a guy who just had the dumb luck to play in a park suited for his skills.

7. Well, Speaker played pre-1920, and Mantle played in the live ball era, so you just have to evaluate which one was more valuable within his own context (hint: Speaker!). :)

Brock,
Here is one
Wt Ht Bat Throw
155 72 B L Bell (Riley 150# 71)
200 72 L L Charleston (Riley 190#)
215 73 L R Hill
190 69 L L Oms
175 71 L L Stearnes (Riley 72)
190 69 L L Torriente

Generally I have entered Riley's weight and height data where available. Perhaps the Bell, Charleston, and Stearnes data (only 2 of 6 match Riley) matches the NBHOFM website. Those are the older HOFers who were in this database "originally".

Which point still stands? Mine is that after giving Speaker reasonable credit for his defense (i.e., the best center fielder of his generation, average +10 peak +25, rather than anything extreme like average +20 peak +35), Speaker was the best, second-best, and third-best player in baseball just as often as Mantle was. Speaker was the best in the major leagues in '12, '14, and '16, the second-best in '17 and '23, and the third-best in '13, while Mantle was the best in '56, '57, and '61, the second-best in '55 and '58, and the third-best in '60. After that, of course, it's all Speaker due to his much greater number of star-caliber seasons.

Black/Hispanic HOMers during Mantle's career, by league by year, listed below

just a compilation, but this does implicitly address "leading AL" issues with Mantle, re the competition. Um, check out 1962 for instance.

Basically Mantle arrived in 1951 to what already was a lesser pool of black/Hispanic stars.
And from 1952-65 - Mantle's prime, heck almost his whole career - the NL added Ernie Banks, Hank Aaron, Roberto Clemente, Frank Robinson, Willie McCovey, Billy Williams, Bob Gibson, Juan Marichal, Willie Stargell, Jimmy Wynn, Richie Allen, Joe Morgan, and Ferguson Jenkins.

The AL added - no such players of that level of quality.

13 young black/Hispanic future HOMers join the party in a span of 14 years - and Mantle didn't have to compete against any of them. No insult intended to Mantle, but it's noteworthy.

* for part-time

1951 NL - Willie Mays, Monte Irvin, Roy Campanella, Jackie Robinson
1951 AL - Larry Doby, Minnie Minoso, Satchel Paige*

1952 NL - Willie Mays*, Monte Irvin*, Roy Campanella, Jackie Robinson
1952 AL - Larry Doby, Minnie Minoso, Satchel Paige (Quincy Trouppe 6 G)

1953 NL - Monte Irvin, Roy Campanella, Jackie Robinson, Ernie Banks*(10 G)
1953 AL - Larry Doby, Minnie Minoso, Satchel Paige

1954 NL - Willie Mays, Monte Irvin, Roy Campanella, Jackie Robinson, Ernie Banks, Hank Aaron
1954 AL - Larry Doby, Minnie Minoso

1955 NL - Willie Mays, Monte Irvin*, Roy Campanella, Jackie Robinson, Ernie Banks, Hank Aaron, Roberto Clemente
1955 AL - Larry Doby, Minnie Minoso

1956 NL - Willie Mays, Monte Irvin, Roy Campanella, Jackie Robinson, Ernie Banks, Hank Aaron, Roberto Clemente, Frank Robinson
1956 AL - Larry Doby, Minnie Minoso

1957 NL - Willie Mays, Roy Campanella, Ernie Banks, Hank Aaron, Roberto Clemente, Frank Robinson
1957 AL - Larry Doby, Minnie Minoso

1958 NL - Willie Mays, Ernie Banks, Hank Aaron, Roberto Clemente, Frank Robinson
1958 AL - Larry Doby, Minnie Minoso

1959 NL - Willie Mays, Ernie Banks, Hank Aaron, Roberto Clemente, Frank Robinson, Willie McCovey*, Billy Williams*, Bob Gibson*
1959 AL - Larry Doby*, Minnie Minoso

1960 NL - Willie Mays, Ernie Banks, Hank Aaron, Roberto Clemente, Frank Robinson, Willie McCovey*, Billy Williams*, Bob Gibson*, Juan Marichal*
1960 AL - Minnie Minoso

1961 NL - Willie Mays, Ernie Banks, Hank Aaron, Roberto Clemente, Frank Robinson, Willie McCovey, Billy Williams, Bob Gibson, Juan Marichal
1961 AL - Minnie Minoso

1962 NL - Willie Mays, Ernie Banks, Hank Aaron, Roberto Clemente, Frank Robinson, Willie McCovey*, Billy Williams, Bob Gibson, Juan Marichal, Minnie Minoso*, Willie Stargell (9 G)
1962 AL - NONE

1963 NL - Willie Mays, Ernie Banks, Hank Aaron, Roberto Clemente, Frank Robinson, Willie McCovey, Billy Williams, Bob Gibson, Juan Marichal, Willie Stargell, Jimmy Wynn, Richie Allen (8 G), Joe Morgan (7 G)
1963 AL - Minnie Minoso

1964 NL - Willie Mays, Ernie Banks, Hank Aaron, Roberto Clemente, Frank Robinson, Willie McCovey, Billy Williams, Bob Gibson, Juan Marichal, Willie Stargell, Jimmy Wynn*, Richie Allen, Joe Morgan (10 G)
1964 AL - Minnie Minoso (5 G)

1965 NL - Willie Mays, Ernie Banks, Hank Aaron, Roberto Clemente, Frank Robinson, Willie McCovey, Billy Williams, Bob Gibson, Juan Marichal, Willie Stargell, Jimmy Wynn, Richie Allen, Joe Morgan, Ferguson Jenkins (7 G)
1965 AL - Satchel Paige (1 G)

1966 NL - Willie Mays, Ernie Banks, Hank Aaron, Roberto Clemente, Willie McCovey, Billy Williams, Bob Gibson, Juan Marichal, Willie Stargell, Jimmy Wynn, Richie Allen, Joe Morgan, Ferguson Jenkins
1966 AL - Frank Robinson

1967 NL - Willie Mays, Ernie Banks, Hank Aaron, Roberto Clemente, Willie McCovey, Billy Williams, Bob Gibson, Juan Marichal, Willie Stargell, Jimmy Wynn, Richie Allen, Joe Morgan, Ferguson Jenkins
1967 AL - Frank Robinson, Rod Carew, Reggie Jackson*

1968 NL - Willie Mays, Ernie Banks, Hank Aaron, Roberto Clemente, Willie McCovey, Billy Williams, Bob Gibson, Juan Marichal, Willie Stargell, Jimmy Wynn, Richie Allen, Joe Morgan (6 G), Ferguson Jenkins
1968 AL - Frank Robinson, Rod Carew, Reggie Jackson

Dan - you make some good points. But. Yes, the AL was slow to integrate in 1955. So was the AL in 1920!! Let's run your chart again, taking away all of the dark-skinned players from the major leagues for each CFer:

Speaker - oh, his ranks remain the same :) As I'm feeling Olympian, he gets three golds, two silvers, one bronze.

Mantle - Six golds, two silvers, one bronze.

Yes, defense matters. It makes their primes closer.

sorry, my counting skillz aren't very good this morning. It's seven golds for Mantle-in-MLB-sans-black-players, not 6. Every .. single .. year from 1955 to 1962, excluding 1959. Babe Ruth would be pleased with such a record.

I just did a quick league strength study for the 50s and 60s, and my preliminary results are absolutely jawdropping. I looked at every position player who switched leagues between 1951 and 1968 (Mantle's career), a sample of 249 players. I took their rate in each season of batting wins per year, baserunning wins per year, and fielding wins per year, and added 8.7 to turn them into wins created per year. I then weighted each player by the harmonic mean of his playing time in the two seasons before and after the switch, giving 87 full seasons' worth of sample (where a player who plays every game in both the year before and after the switch is counted as 1.00). Finally, I took the ratio of their weighted performances before and after the switch. The ratio for batting wins was 1.092, for baserunning wins it was 1.001, and for fielding wins it was 1.007. This is, astonishingly, on par with the gap between the major leagues of 1944 and those of 1942/46. Notable examples:

Frank Robinson went from a 151 OPS+ in the 1965 NL to a 198 in the 1966 AL
Pete Runnels went from a 130 OPS+ in the 1963 AL to an 88 in the 1964 NL
George Strickland went from a 67 OPS+ in the 1951-52 NL to a 98 in the 1952-53 AL
Harvey Kuenn went from a 118 OPS+ in the 1960 AL to an 86 in the 1961 NL
Earl Torgeson went from a 96 OPS+ in the 1954-55 NL to a 120 in the 1955-56 AL
Dick Stuart went from an 81 OPS+ in the 1962 NL to a 126 in the 1963 AL, and then back from a 118 in the 1964 AL to a 101 in the 1965 NL
Moose Skowron went from a 115 OPS+ in the 1962 AL to a 60 in the 1963 NL, and then back to a 108 in the 1964 AL
Tommie Agee went from a 105 OPS+ in the 1967 AL to a 69 in the 1968 NL
Frank Howard went from a 111 OPS+ in the 1964 NL to a 138 in the 1965 AL
Roy Sievers went from a 143 OPS+ in the 1961 AL to a 117 in the 1962 NL
Lee Thomas went from a 128 OPS+ in the 1965 AL to a 70 in the 1966 NL

...and the list goes on and on. Overall, 152 of the 249 players and 53 of the 87 seasons' worth of playing time sampled (both 61%) improved upon switching from the NL to the AL or deteriorated upon switching from the AL to the NL, while the remaining 39% did the opposite.

A 1.09 conversion factor is so dramatic it boggles the mind. Could any of the group's other number crunchers possibly take the time to do their own study here and make sure I'm not entirely off the reservation? This would have tremendous implications for my PHoM, and for my understanding of the game's history in general. It suggests, for example, that Mantle's 223 OPS+ in the 1957 AL would have only been a 200 in the 1957 NL, or that Aaron's 181 OPS+ in the 1959 NL would have been a 202 in the 1959 AL. It would be great to have someone else do a reality check on this.

TomH, this is precisely why I impose a segregation penalty on pre-1947 players. You will get no argument from me on this. But the study I just completed suggests it should be something on the order of 20 points of OPS+ a year, not the mere tiebreaker I've been employing so far.

And wouldn't the segregation penalty be even bigger for pre-1947 leagues, since the integration of the NL presumably had at least a trickle down effect on the quality of the AL, e.g. forcing out sub-par white ballplayers? This is, indeed, pretty big if true. Mays becomes the solid #1 at CF, and Bonds jumps past Williams on the all-time list, and possibly even Ruth.

Interestingly, I just repeated the study on 1910-19 and found no difference in league strength, although the sample was much less robust (82 players, 28 full seasons' worth of play). For every 1917-18 Lee Magee, there was a 1917-18 Dave Shean.

Also, just reverse-engineering BP's all-time adjustments, they have a conversion factor of 1.054 between the AL and the NL for 1959, which fits with my preconceptions a lot better than the extreme 1.09 figure produced by my study.

Of course it would, Blackadder. It's just hard to put a hard number on it, especially since the ratio of black baseball talent to white baseball talent may not have been static throughout the 1890-1947 period. The real way to do it would be to get MLE's for every single player in the Negro Leagues, have better black players bump worse white players out of the league until you reach the maximum talent level over 16 teams, and then reduce everyone's wins above that new replacement level at the same ratio until total league runs scored equaled total league runs allowed. But you'd need complete NgL MLE's for that, and I don't think we'll ever get those, will we?

But you'd need complete NgL MLE's for that, and I don't think we'll ever get those, will we?

If the HoF would release its data in encyclopedic form, we'd have the evidentiary basis for it. If they would release the data in database/spreadsheet format, it would be possible for someone with database management skills to crunch the numbers.

I don't see it as unimaginable that we will have MLEs for all NeL players, 1920-47. But I'd guess we are at least 10 years away from having it.

Chris, any chance you could use your well-established NgL MLE methodology to work out a conversion factor between the NL and AL for the 1951-68 period to double-check mine?

addendum to post 136 - what about "almost" HOMers, stars of this sort debuting 1951-68 and getting a lot of support in our 2008 election?

- Reggie Smith, our No. 1 holdover, would list for AL in 1966 with 6 G. He's full-time ALer in 1967-68, adding to the about-time improvement in this area in the late 1960s.
- Luis Tiant, our No. 9 holdover, would fill in 1964-68 for AL and be the only such pitcher other than Satchel Paige.
- Tony Perez, No. 14 of the holdovers, gets one back for NL with 12 G in 1964 and then 1965-68.
- Don Newcombe, No. 20, would already be on the NL list at 1951, and add 1954-60 (also AL in 1960, where the only black HOMer he'd face on the mound would be Minnie Minoso).
- Lou Brock, No. 23, plays 4 NL games in 1961 and then adds to the 1962-68 NL pile.
- Elston Howard, No. 28, would be a huge AL addition as he'd list from 1955-68. But he was only on 12 of 50 ballots in 2008 voting, with a high of No. 5.
- Bobby Bonds, No. 30, would provide another NL player in 1968.

And so it goes from there, pretty even near the bottom, at least at a glance.
Newcombe and Howard would be the only 1950s additions, and they are long shots at this point.
And Tiant and Perez cancel each other out for 1963-65 as well.

Point being, the 1951-65 lopsidedness is not skewed by too many "just missed" AL stars compared to NL....

Dick Cramer's original study, quite the extensive mastrpiece, (published in the 1980 BRJ, and The Hidden Game by Pamler/Thorn) showed for 1957 and 1959 that an AL hitter would have lost 18 and 16 pts of SLG respectively if his league were NL quality (not that his SLG would have gone DOWN if he had moved to the NL, since the NL was a better run-scoring league); I don't have the OBP equivalent, but assuming similar OBP diffs that would mean about 7 runs (.7 wins) a year per hitter difference.

Howie,
If you go further down the list, black or hispanic
AL: Campaneris, Oliva, Aparicio
NL: Cepeda

I asked about #139 over in "Dan Rosenheck's WARP Data", to check my understanding of his study. The league quality thread would have been a better destination.
(sigh)


At SABR34 (2004) I talked to Dick Cramer about his approach. He was hoping or planning to redo it with more accurate measures available today. I think he told me next year that he had done so, with similar results. Whatever the results it would be good to have that available, published online.

I was getting something closer to two wins here, which is just implausibly large. Thanks for the reality check. I'd love to see the nitty-gritty of his approach, and how he's reading the data differently than I am.

Part of the Ink phenomenon is just expansion--it's twice as easy to lead an league with eight teams than one with 16...

I don't particularly buy into this. There might be twice as many players in a 16 team league than an 8, but there aren't necessarily twice as many guys who have the ability/potential/whatever to actually lead the league in something. How many guys there are who have that sort of potential, it seems to me, is more reflective of at least 2 other things than it is of the raw number of guys in the league.

1. The talent pool. How many guys there are who have the ability to lead the league is going to reflect the number of guys on the right end of the talent distribution, which is going to reflect, among other things, the overall size of the talent pool. Guys, not teams.

2. And it's also going to be dependent on where on the right end of the talent distribution the very best players happen to fall. IOW how far to the right end of the distribution a guy has to be to be a candidate to lead the league depends on where the outermost points are located. Or to put it another way, it probably would have been no harder for Babe Ruth to lead a league that had 16 or 32 or 100 teams. Barry Bonds would have led 1,000 team league in walks, OBA and OPS, because the addition of another 970 teams would have added exactly zero players with the potential to OPS 1.000.

OTOH if there is nobody in a league that is any better than Carlos Quentin, then, sure, it's going to be hard to lead that league. It might in practice be harder to lead a weak league. But that's just because it's harder for Carlos Quentin than it is for Barry Bonds. IOW it's not because there are 10 Carlos Quentin's, it's because there's no Barry Bonds.

>>
Wt Ht Bat Throw
155 72 B L Bell (Riley 150# 71)
200 72 L L Charleston (Riley 190#)
215 73 L R Hill
190 69 L L Oms
175 71 L L Stearnes (Riley 72)
190 69 L L Torriente

Generally I have entered Riley's weight and height data where available. Perhaps the Bell, Charleston, and Stearnes data (only 2 of 6 match Riley) matches the NBHOFM website. Those are the older HOFers who were in this database "originally".
<<

Larry Lester covers the early Hall of Famers in The Negro League Book (SABR 1994).
145-170 72 Bell
180-230 73 Charleston

Robert Peterson (1970)
Bell: 135 pounds in 1922 when he pitched semipro in East St Louis and left that team for the St Louis Stars
Charleston: mentions the weight gain but doesn't say anything specific
Torriente: 190 70

--
By the way, according to the same sources
Wt Ht
190 74 Dihigo (Riley 75) (Lester 190-225 75.5) (Peterson 190 73 "his full size")

Just a further response to Dizzypaco's comment that

I also don't believe it was harder to dominate the AL of 1915 than the AL of 1955, no matter what Dan's analysis tells him. The AL of 1915 had several players putting up statistics dwarfing the rest of the league, including Cobb, Speaker, Collins, and Johnson. The only one putting up numbers even close to Mantle in the 50's AL was Ted Williams, of which nothing needs to be said.

I understand that the nuances of standard deviation calculations, particularly when you are trying to distinguish between actual ease of domination and observed deviations from the mean, can be a bit daunting. So let's take a much less technical approach: following Joe Dimino's method for starting pitcher innings translations, let's look at the average of the #6 to #15 finishers in OPS+ in the major leagues over rolling nine-year periods during the entire 16-team era. Given the large sample sizes (90 player-seasons), this should serve as a reasonable proxy for ease of offensive domination: it intentionally ignores the absolute top outliers, who are likely to be once-in-a-generation players that say little about the overall ease of domination of the league, but focuses on the middle of the leaderboards, which should provide a sense of how far above average we can expect most of the best players in the league to be. Here are the results:

1901-1909: 143
1906-1914: 143
1911-1919: 142
1916-1924: 142
1921-1929: 146
1926-1934: 148
1931-1939: 147
1936-1944: 148
1941-1949: 144
1946-1954: 143
1951-1959: 146

This should make pretty clear why I say the deadball era was not easy to dominate on the whole. If it were, wouldn't you expect guys who ranked around 5th in their league in OPS+ to have a higher average OPS+ circa 1915 than circa 1955? Well, in fact, it's the other way around. What you had in the deadball era was NOT a high standard deviation overall but rather very high kurtosis ("fat tails"): a few outliers named Lajoie, Wagner, Cobb, Collins, and Speaker ripping the league to shreds, while the other players were actually clustered quite close to average. Once you got past the two giants of the aughts or the three of the teens, you were down into the dregs pretty quickly.

The top position players in the weak early-mid teens NL were Wheat, George Burns, and Gavvy Cravath, and in the second half of the decade you added the pre-peak Hornsby and Groh. The AL was certainly stacked, with Speaker/Cobb/Collins throughout the decade, and Baker and Jackson tearing things up as well around 1910. But the #4/#5 OPS+ finishers in the strong league were still guys like Fournier, Strunk, and Veach, with pre-peak Sisler coming along at the end. In short, the game was extremely top-heavy. Now look at how deep the 1950's were: you had T. Williams, Mantle, Mays, Musial, Aaron, F. Robinson, Mathews, Berra, Kaline, J. Robinson, Banks, Snider, Kiner, etc. If the deadball era was so easy to dominate, why were only a handful of players able to take advantage of it? Shouldn't that rising tide have lifted all boats (as it certainly did in, say, the Steroid Era NL)?

Now, maybe there was some systematic reason why kurtosis was so high in the deadball era--a small handful of great players utterly obliterating the league, with the rest of the All-Stars left in the dust. But I certainly can't think of one. It seems to me that the only valid explanation is indeed a random "star glut," which is why I don't discount the achievements of that period's giants more in my WARP.

Paul - Thanks so much for the NgL height and weight data. I'd done a web search, but that particular info was hard to find. I think I found the NgL Hall site, but it didn't have reliable data. For example, the only source for Bell had him at 6' 2" and 140 pounds. That's unreasonably thin for anyone over the age of 18, and may well come from the same source as the rookie data you found. The data you provided, while suggesting that Bell was very thin, doesn't have him down as anorexic. What I'm doing here is trying to see if top center fielders are, in general, smaller than other top ballplayers (yes, although it's a generalization), and how many of them have what Bill James calls "Kirby Puckett" bodies - short and stocky but still very fast (a high percentage). I also want to know if the very very best defensive center fielders have the KP bodies (no) or larger (yes, in general, Curt Flood being on the borderline). This allows me to get an extra indicator on those guys we have for whom very little defense data is available and we have to go on reputation. I got the idea from living in St. Louis and watching Mel Gray play football, and then from being from the same high school (Webster Groves) as the current and presumably eternal (the distance is no longer run) holder of the 100-yard dash record, Ivory Crockett. Both were about 5' 7" tall, weighed more than one would expect for a speedster of that height, and were blindingly fast anyway. There appears to be a "sweet spot" for speed in the human body makeup between about 5' 6" and 5' 8", weighing about 170-190 pounds. But those guys, while as fast as anyone, don't seem to be the truly dominant fielders at center. I'll post up the data later today. Again, thanks! - Brock

Another approach to calculating a segregation penalty would be to go backwards. If I remove all the black players from the majors in, say, the 1970-80 period, and fill in their playing time with replacement players (lowering the replacement level a little in the process), how much do I have to improve the remaining white players' average OPS+ to get a league OPS+ of 100? The two problems with this method are 1. it assumes that the ratio of black to white talent was the same from 1893-1947 as it was from 1970-80, which may be false and 2. I don't have a spreadsheet with every player's race, and I'm not about to look up 500 photos to put one together myself. Does anyone have any ideas to help me with this logistical challenge?

Mark Armour, who leads the SABR Bioproject, added a race variable to his database. I don't know how many values ("races") but one purpose was to identify all players who would have been unacceptable before 1947. Scope, major leagues 1947-1986 or so. That was a lot of work, more than looking at photos. It was not a Bioproject project, so the data may not be available to SABR members. On the other hand he may have put it in the public domain.

> 1. it assumes that the ratio of black to white talent was the same from 1893-1947 as it was from 1970-80, which may be false

get serious!
I am skeptical about 1908-1927 vs 1928-1947.
1908-1947 may be practically all you need but let me be discouraging about the leap from 1970s to any segregation period.

Willie Mays comment -

Mays' ranks in OPS+ thru his prime, chronologically:

1 1 4 1 1 / 3 3 / 3 3 2 1 1

The "/" are when he switched parks (twice). It's been often siad Mays adjusted his swing to the various dimensions of his home place (and the prevailing winds!). Slight trough in the middle of his career may reflect this. He adjusted well tho!

Looking at the defensive stats shows this also; his ##s by Win Shares and WARP actually show a brilliant early career, lull in mid-career, and then amazing numebrs ofr a mid-30s CFer.

Like DiMaggio, Mays was a truly all-around skilled player who was even better than the numbers he put up.

As a Larry King impersonator once said on Saturday Night Live, "Willie Mays was a heckuva ballplayer."

I don't see it as unimaginable that we will have MLEs for all NeL players, 1920-47. But I'd guess we are at least 10 years away from having it.

10 years is a bit pessimistic! ;>)

The data is in a database, and could theoretically be released tomorrow. All that's needed is for the HOF to give up on someone putting an encyclopedia deal in their laps, and allowing the data to be made public.

Hi. Here's the height and weight data I promised for the HoM center fielders plus a few selected others, sorted by height then weight.

Billy Hamilton 5 06 165
Hack Wilson 5 06 190
Lip Pike 5 08 158
Jim O'Rourke 5 08 185
Kirby Puckett 5 08 210
Curt Flood 5 09 165
Dom Dimaggio 5 09 168
Jimmy Wynn 5 09 170
Earl Averill 5 09 172
Paul Hines 5 09 173
Alejandro Oms 5 09 190
Cristobal Torriente 5 09 190
Richie Ashburn 5 10 170
Edd Roush 5 11 170
Max Carey 5 11 170
Turkey Stearnes 5 11 175
Willie Mays 5 11 180
Vince Dimaggio 5 11 183
Tris Speaker 5 11 193
George Gore 5 11 195
Terry Moore 5 11 195
Mickey Mantle 5 11 198
Willard Brown 5 11 200
Cool Papa Bell 6 0 155
Pete Browning 6 0 180
Darren Lewis 6 0 189
Duke Snider 6 0 190
Oscar Charleston 6 0 195
Ty Cobb 6 1 175
Larry Doby 6 1 182
Pete Hill 6 1 215
Joe DiMaggio 6 2 193
Andre Dawson 6 3 195

The non-HoMers, and the reasons for including them, are:

Curt Flood, because the current state of the competition for best defensive outfielder ever seems to be down to Speaker, Mays, and Flood (SMF). Hack Wilson because he's sort of the poster boy for short stocky guys who can run but who aren't exactly prime defenders. The two "other" DiMaggio brothers and Terry Moore because everyone seems to agree that they are just one step below the SMF trio. Darren Lewis because I wanted a more modern top glove, and Lewis played 11 years in the bigs despite being helpless to hit the pitching.

As you can see, I was on to a small something here. There really is a cluster of guys at 5' 8" and 5' 9", and another at 5' 11" and 6' even, with Richie Ashburn standing alone in between. But it's a small something, not a large one, and reporting accuracy is not exactly the strong point of height and weight. Still, it does seem to be there. There does seem to be a "sweet spot" for speed down there. I doubt that the HoM has any other position where that high a percentage of the players are under 5' 10" tall, much less as stocky as this set is.

And yes, there is a small correlation between being a bit taller and being a superglove, especially among the short stocky guys. Flood and Dom are the best gloves under 5' 10", and they are also the two thinnest.

The main reason I wanted to do this was to get some sort of handle on the defensive capabilities of the Negro Leaguers and, especially, the latins. All of those guys have hot glove reps, because, well, they were the best athletes in their leagues. But Ohms and Torriente look like there's a limit to how good they may have been. It's not anything like proof, but even indicators are few and far between here, and this is one.

The guy who is most helped here is Cool Papa Bell. He has always had a truly hot glove rep, but he also had a hot bat rep and his MLEs indicate that he may have had too many press-friendly pals pumping that rep up. However, he is just the right body type to be a truly hot glove, and there are no good defensive MLEs that I know of for the NgL. Also, one post in his HoM thread shows that, for one year, he had a very large percentage of his team's outfield putouts. That's only one year, and who knows how good his corner outfielders were, but it is one indicator, and we have to fish here. This is another one. So we have his rep, the one year, and the body size and shape, and no indicators that he was not a superglove. Someone from the NgL ought to have been, so I'm nominating Cool Papa.

Anyway, that's it. This is nothing big, but there may be something there if you have to make a hard decision between two guys. Thanks for reading. - Brock

Ty Cobb - as I've explained w.r.t Hornsby, D Allen, etc., I'd be wary as a GM of taking Cobb over Speaker on my team; who knows how long I could keep the team together? Cobb was almost traded for 50c on the dollar early in his career because of this; but the other team turned them down :)

Mantle / Speaker

below are bb-refs neutralized career stats for both. Below Spekaer's line is a 'reduced Speaker', cutting off his first 2 and last 2 yrs which didn't add much to his numebrs.

player G . AB . . R . . .H . . 2B .3B HR . RBI . . BB SB . Avg OBP SLG BtW
MM 2468 8566 1857 2710 388 078 599 1671 1956 172 .316 .442 .590 85.3

TS2962 10844 2018 3736 832 240 119 1636 1463 481 .345 .428 .498 87.5
TS' 2739 9972 1919 3486 767 230 114 1526 1400 463 .350 .435 .500 86.8

BtW are bbref's (Pete Palmer) batting wins.

Offense only, Mantle has a clear lead on peak, Speaker a slight lead in career.

Speaker has large home-park advantages. I chop a mere win or so for that.

League strength: no, I do not timeline. But I DO take into account integration. I do not wish to penalize Mantle, playing in a slowly-integrated AL, over a guy who didn't have to play against Oscar C, etc; let's just pretend the best Negro League stars went to the 1920 NL. And ya know what? BP agrees with me.

Defense: I'll give Speaker 90% of his defense as measured by DRA, etc. Maybe worth 14 wins more than the Mick over his career.

Does ANYONE take into account post-season here? Why do I hear so little about it?? Here are basic stats in World Series play:

player .G . ab . r . h hr rbi bb
Mantle 65 230 42 59 18 40 43
Speakr 20 .72 12 22 .0 .3 11

Now, I KNOW r and rbi are team-dependent, and poor tools to use for career (or even season) evals. But in a short-series context, actual runs scored and runs batted in are important! Tris Speaker drove in 3 runs in 20 WS games. One of those was indeed a truly key hit; game 7 of the 1912 series. Of course, he had already effectively popped out, but Matty and Chief Wilson goofed and the ball dropped foul. So he should get as much clutch credit for that as Jeter's 'Jeffrey Maier home run' in 1996.

I give Mantle about 1-1.5 wins of extra credit for his WS play. And I count WS play (pre-division) as 6-8 times the amount of regular season. That's 9 wins. If you wish to disagree, go ahead... but I get the feeling around here that we're almost ignoring it, which frankly is a lot more bogus than electing Nellie Fox :)

Mantle-Speaker-Cobb-Willie-Oscar are all close. I can see arguments for any order.

Good point, I tend to ignore post-season play mainly because the information isn't easily sorted. However, I agree that there is no justification for ignoring this information. It's usually our best source of info on Negro League players but for MLB players we collectively toss it out the window.

Quoting TomH
ballot:
4 speaker
Docked a bit for lesser ability to play post-live ball

It seems to me that he adjusted well, or didn't adjust but showed that his skills and habits were well-suited to the "lively ball". Do you mean literally less ability to play in the '20s?

just above:
League strength: no, I do not timeline. But I DO take into account integration. I do not wish to penalize Mantle, playing in a slowly-integrated AL, over a guy who didn't have to play against Oscar C, etc; let's just pretend the best Negro League stars went to the 1920 NL. And ya know what? BP agrees with me.

agrees that NL > AL in the 1920s as in the 1950s?

TomH, thanks very much for going through your reasoning in such detail. Responses:

1. Palmer's batting wins are in perfect agreement with mine that Speaker and Mantle had the same offensive value above average over their careers. (That gives Speaker more offensive value over replacement, of course).

2. Given how different Fenway and League Park were, if Speaker had big home park advantages in both places, that's a credit to his adaptability, just in the way that you are crediting Mays for adapting to park changes. Moreover, Mantle himself had a substantial home-field advantage as a switch hitter batting mostly from the left side in the old Yankee Stadium: he had a 1.031 OPS at home and a .966 away starting in '56. If anything, I think Mantle should be docked for a favorable environment more than Speaker, but one win is not worth fighting over.

3. BP *is* using an explicit timeline, so that should be irrelevant for our purposes. You know I agree with you about penalizing for segregation. But since neither Speaker nor Mantle were competing against numerous black players, the segregation factor shouldn't apply to the comparison, right? We should just be able to look at their actual value, since the non-timelined quality of competition is the same. And on actual value, it's Speaker running away, no?

4. 90% of DRA would be 25-26 wins of marginal defense to Speaker, since DRA has Speaker at +265 and Mantle at -20. I'm actually using more like 80% of DRA, which is 22 wins. But none of these numbers are close to 14! Positing only 14 wins of defense is regressing the fielding gap between the two by *half*. Given that the reputational evidence is completely in line with the stats (Speaker one of the top two fielding CF ever, Mantle average), 14 wins seems really, really unfair to the Gray Eagle.

5. Postseason: Speaker was never a big RBI guy; I imagine he hit too high in the lineup for that. He was certainly a big Runs guy, with a bunch of 2nd and 3rd place showings in the tough AL. 12 R in 20 G certainly seems like a reasonable showing to me. Speaker and Mantle's OPS's in the World Series were both 8% lower than in the regular season; I certainly don't think it's fair to suggest that Speaker was a choke artist while Mantle was Mr. Clutch. Mantle simply GOT to the postseason much more than Speaker, thanks to the likes of Berra, Ford, Rizzuto, Maris etc. But if you want to give credit for that, that's your prerogative.

Could you walk me through your math step by step?

Regular season offense: Equal (85-87 Palmer batting wins apiece)
Fielding: Huge advantage to Speaker (you say 14 wins, I say 22)
Park: You say -1 to Speaker, I say if anything -2 to Mantle, but really it's irrelevant
Integration: Equal, since neither played against meaningful numbers of blacks. If you want to give a few wins to Mantle for Doby and Miñoso, that's fine by me.
Postseason: You say +9 for Mantle. Does Speaker get ZERO wins for his .856 World Series OPS? He obviously contributed SOMETHING to those three rings....

So in the very, very, very most favorable scenario for the Commerce Comet, he just breaks even with the Gray Eagle (0 - 14 + 1 + 2-3 + 9). If you give Speaker anything close to the defensive credit his reputation would suggest (which would be at least a +18-+19 spread), or any credit at all for his own postseason play, it's still Speaker comfortably...no?

Paul Wendt, I think TomH is suggesting that it's not fair to ding Mantle for not playing against blacks but not Speaker (which of course is correct), so they should both be compared exclusively to their AL peers. (I still disagree with this, because it's not Speaker's fault that Cobb, Collins, Jackson, Baker, and Crawford were all in the AL, while the NL only had Wheat/Cravath/Burns).

I meant BP agrees that AL 50s = to NL 20s.

I gave Mantle 1.0-1.5 post-season wins (= 9 regular season wins) over Speaker. He would have about 12 wins over someone who had played poorly in post-season.

I don't use exclusively DRA for defense. WARP, WS, reputation, etc.

And there is the "peak" thing too.

I appreciate the analysis, Dan. Without WS heroics, Mantle would indeed be behind on my ballot. I will go back and do a play-by-play study [before Sunday :) ]of their WS at-bats to see if this large clutch bonus is deserved.

WARP has the fielding gap between Speaker and Mantle at 23.0 wins. Fielding WS (after correcting for its compressed standard deviation) has the fielding gap between Speaker and Mantle at 22.5 wins. DRA has it at 32.5 wins, but that should be regressed a good bit, due to the system's high stdev for the deadball era; probably something in the mid-20s would be appropriate. Reputation has Speaker as either the best or second-best defensive center fielder of all time, and Mantle as average. We know from modern PBP metrics that the absolute best defensive CF save 20-25 runs a year at their peaks in today's game, so if you just make up, say, 3 years at +25, 3 years at +20, 3 years at +15, and 3 years at +10 for his prime, and then league average in his decline phase, that gets you to +210, or 21 wins. No matter which approach you take, it's the same answer. I just don't think there's any legitimate justification for placing the Speaker/Mantle fielding gap at much below 20 wins. And after crediting Speaker appropriately for his fielding, Mantle's peak advantage becomes marginal rather than imposing.

I know I am beating a dead horse here, that you are all probably absolutely sick to death of hearing me repeat the same three points. But I just feel *really* strongly that Speaker is getting a raw deal from our electorate, just because he had the misfortune to share a league with two other inner circle immortals, and because a huge part of his greatness (fielding) doesn't show up in everyone's favorite baseball-reference statistic of OPS+. I'll drop it, I promise--for now. :)

On a related note, Tom mentioned Speaker's RBI. I am *stunned* to see that Mickey Mantle only had four 100-RBI seasons (compared to nine straight 100-R seasons). Was he hitting second in the lineup for many years or something?

Mantle walked a lot. And he did have 5 more 92 RBI years. Plus, some yrs like 1960; (40 HR, 94 RBI) hitting behind the guy who had a .302 OBP when not hitting a HR.

I checked the 1912 W.S.
Speaker overall playded well. Made on error in the field, but also had two DP in CF, including one of his patented unassisted ones, doubling off a runner on 2nd. Gosh, I woulda luvved to see some of those. I think Speaker and Ozzie would be my favorites to watch on defense.

Speaker had 34 PAs in the Series; and came up with a total of 25 runners on base. That is HUGE. He hit third, andthe guys in front of him did great. 12 men on first, 8 on second, and 5 guys on third. He drove in 2; one with a groudnout, one after he was 'retired' but the popup was dropped.
With no one on: 1 W, 14 AB, 6 H, 11 TB, 0 RBI, 1253 OPS
With runners on: 3 W, 16 AB, 3 H, 3 TB, 2 RBI, 452 OPS

They did win the Series, and he did get the game-tying hit in the 7th, so I won't ding him much at all. But he gets no bonus credit for .300 avg and gold glove.

sorry, that's a 503 OPS with runners on.

and Mantle as average

Everything I have read has suggested that Mantle in his prime had a reputation as being an above average defensive center fielder, with excellent speed and a strong arm.

On a related note, Tom mentioned Speaker's RBI. I am *stunned* to see that Mickey Mantle only had four 100-RBI seasons (compared to nine straight 100-R seasons). Was he hitting second in the lineup for many years or something?

The fact that Mantle had only four 100 RBI seasons is something I thought was pretty well known. He batted either third or fourth most of his career, but often had guys like Bobby Richardson and Tony Kubek hitting in front of him. He did just as well with runners on base and runners in scoring position as in other situations. His lack of 100 RBI seasons resulted from a combination of the guys batting in front of him, the number of walks he drew, the context in which he played, and the injuries later in his career.

Just quickly paging through my biography of Speaker here...

In Game 2 of the 1912 WS, with Boston trailing 6-5 in the 10th inning against Mathewson, Speaker tripled, and despite being tripped by the shortstop, came home and kicked the ball loose from the catcher to score, keeping the Sox alive before the game was called. Without that hit, the Giants would have won the game, and the Series with it, because there would have been no Game 8. It was a championship-defining play of fearless baserunning as well as clutch hitting.

Of course, as you mention, he did indeed have the game-tying hit in the 10th inning of that decisive Game 8, only after the Giants let his popup drop.

Moreover, in the 1915 WS, he singlehandedly won Game 2 for the Red Sox with his glove. In the bottom of the 9th, Boston had a 2-1 lead. Dode Paskert drove a pitch to right-center that seemed destined for the temporary bleachers that had been installed. Speaker raced back and literally snatched the ball out of the fans' laps, taking a homer away from Paskert and ending the game. (My biography says there was a man on second, which would have meant Paskert's blast would have won the game, but Retrosheet says the bags were empty, in which case it would have gone to extras. At any rate, it was one of the greatest and clutchest defensive plays in postseason history).

Speaker also tripled and scored in one-run games in Game 1 of the 1912 WS and Game 3 of the 1915 WS, and hit .320 in the 1920 Series. He may not have been Reggie Jackson, but he definitely left his mark on the postseason with iconic plays for the ages that won multiple championships for his teams--at the plate, in the field, and on the basepaths. I think giving a nine-win margin to Mantle over Speaker for postseason performance is patently unfair. You're discounting the fact that they had similar OPS+ in the World Series by saying that Mantle's hits were clutcher, but Speaker had more than his share of heroics as well.

His lack of 100 RBI seasons resulted from a combination of the guys batting in front of him

In 1961 the New York Yankees hit a [then] record 240 home runs.
Their primary leadoff man played in 162 games and had 706 PAs

he scored 80 runs.
Their primary #2 batter had 658 Pas, he scored 84 runs

That's why Mantle batting forth had "only" 128 ribbies to go with his 54 Homers- even though he hit .364/.490/.907 with RISP that year.

Mantle's splits for his career (not incl his first 5 years though) have him hitting better with men on than bases empty, yet despite his power he had only 4 100 Ribbie seasons.
He batted with men on in 46.7% of his PAs, with RISP in 24%.

By contrast RBI man Joe Carter batted with men on in 49.9% of his PAs and with RISP
30% of the time.

That's a big difference, on average 80.1% of the teammates Mantle drove in cam,e from RISP at bats- if he batted with RISP 30% of the time rather than 24% he might have had about 150-175 more career RBIs 8-10 more a year, 4-5 more 100 ribbie seasons.

well, Dan, like I said, I'm going to look at PBP data, and I've only gotten thru 1912 so far. Then I'll decide if it's patently unfair. So far it seems you're spoon feeding me more fan-boy than data, altho I do appreciate the '15 WS defensive gem anecdote.

OK, I just couldn't let this idea that Speaker somehow underperformed in the postseason linger any longer, and as you can imagine I don't take kindly to accusations of providing "more fan-boy than data." So I've actually taken the time to calculate his Win Probability Added in the World Series, using the win expectancy calculator at winexp.walkoffbalk.com and the play-by-play records of all 20 of his World Series games.

Speaker's offensive WPA (batting plus SB/CS) in the World Series was .854. That is almost entirely attributable to the two key clutch hits mentioned earlier: the triple (stretched into an inside-the-park HR due to Speaker's obliterating Art Wilson at the plate) in the 10th inning of Game 2 in the 1912 WS, and the RBI single (which also moved the eventual winning run to third base with one out) in the 10th inning of the decisive Game 8 of the 1912 WS (yes, after the Giants allowed a foul popup to drop). Those two plays accounted for .496 and .457 of positive win expectancy, respectively.

Note that just by extrapolating from his raw offensive line of .320/.398/.458, we'd only expect Speaker to have .504 of positive win expectancy over 20 games. This means that far from being a choke artist, Speaker was in fact Mr. Clutch: the timing of his hits in the World Series meant they were 70% more important in determining the outcome of the games than they would have been had they been distributed randomly.

And this, of course, is before counting Speaker's fielding. Judging by the description in my Speaker biography, the Paskert drive at the end of Game 2 was 100% likely to be a home run if it were not caught, and it sounds like there was no more than (completely making up numbers) a 5-10% likelihood that an average center fielder would be able to corral it. (The book cites George Sullivan's Picture History of the Boston Red Sox, John Foster's report in The New York Times, and N.J. Flatley's account in the Boston American as its sources for the difficulty of the catch). If the ball had gone over the fence, the game would have been tied with two outs in the bottom of the ninth, giving Philadelphia a 55.2% chance to win (according to the win expectancy database); if it were caught, the game would be over. So let's be generous and say an average CF would have grabbed it 15% of the time. That means that 85% of the time Philadelphia would have a 55.2% chance to win, and 15% of the time they'd have an 0% chance to win, for a weighted average of 46.9%. Since Speaker did catch the ball, that's a further .469 of positive win expectancy from that play alone.

The two double plays you mention, Tom, actually didn't matter much at all, as they occured in very low-leverage situations, providing just .008 of positive win expectancy for the Sox. But my Speaker biography mentions three additional "nice running catches" at different points in his World Series career, although it fails to provide specifics as to which situations they occurred in. So I'll just resort to a brute-force approach: assume that an average CF makes a "nice running catch" on 35% of the balls that it is possible to make them on, and that they happened in situations with a leverage of 1.00. An average hit to the OF is worth .072 wins (.056 for the hit, .016 in that run environment for avoiding an out), while an out is worth -.01 wins. So an average CF will cost the opposing team .01 wins 35% of the time those balls are hit, and surrender .072 wins 65% of the time, for a weighted average of .043 wins for the offense per ball hit on which it is possible to make a "nice running catch." Therefore, each time Speaker made a "nice running catch," he cost the opposing team .01 wins, for a total of .053 wins above an average CF per "nice running catch." Add 3*.053 to the .469 from the Paskert catch and the .008 from the double plays, and you get .636 of positive win expectancy for Speaker on defense, assuming that on all plays besides those six he was league-average.

.636 of positive WPA on defense plus .854 of positive WPA on offense is a total of +1.49 WPA in 20 World Series games, or .0745 per game. Project that out to a 162-game season, and you get a +12.07 WPA, which is the level of Barry Bonds in 2001 and 2004 (11.63 and 12.63). Yes, you heard me right: Tris Speaker improved his teams' chances to win in World Series play at the same rate that Barry Bonds improved his teams' chances to win in the regular season in 2001 and 2004. Postseason performance should be a very strong argument in favor of Speaker, not against him. Is this data good enough for you, TomH? :)

OK, leave '04 out of it, he was better that year. Bonds in '01 was +11.63 WPA in 153 games, so +12.31 per 162, which is close enough to Speaker's +12.07 in World Series play. If anyone wants the spreadsheet with the WPA for all 78 of Speaker's World Series PA and the calculations for his fielding, I'd be happy to provide it.

Dan, isn't there an argument that CF wasn't as important defensively in the 1910s and 1920s as it is today.

Does your data show anything like this? Where does CF show up on the 1910-1930 defensive spectrum?

I'm not arguing Speaker's worthiness or anything, just checking on a theory.

That argument is correct. As you can see from the above chart, replacement CF were 1.0 wins below average per 162 games during Speaker's career, and 1.2 wins below average per 162 in Mantle's (which results in a three-win edge to Mantle for intrinsic positional value that is already baked into my WARP). CF was in a clump with 1B and 2B at the height of the deadball era, with LF not substantially deeper; C, SS, and 3B were the demanding defensive positions. By Mantle's day, CF had clearly distinguished itself from the corner outfield positions (and 1B of course) and ran about even with 3B, followed by C, then 2B, and then SS. But it wasn't until the Steroid Era that CF *really* "broke loose" and became as scarce as 2B again. This info is of course available in the StDevs and Rep Levels.xls file in my WARP archive.

Sigh. I feel like an idiot. I should have worked this out about 20 years ago and published the results in the BBBA. But I didn't, so here I am, asking for help. As we all know, there were more errors in early baseball than there are now. But there's no assignment of those errors to the hitters who hit into them, just the defenders. And there's no caught stealing data for a lot of the 19th century, either. Therefore, all the player measures that I know of evaluate the early hitters as though errors and stolen bases did not exist, because there's no data.

So far, that's OK by me. You can't do what you don't have the data to do. But then people want to make era adjustments to the hitter measures based on actual runs scored in those leagues. That's not so OK, if you're comparing an early guy to a later one. The early guy gets docked for actual runs scored, but the measures only can see what amounts to the most simple runs created formula: AB x OBP x SLG (there are other algebraically equivalent ways of stating the formula, but this one works with easy to find data).

I've always figured that there just weren't enough error-driven and running-driven runs to make a whole lot of difference, so no adjustment was really needed. But I was comparing Billy Hamilton to Joe DiMaggio (a thankless task) and I took a look at the data from their times. The difference isn't small. It's very large, maybe huge. I have a chart at the end of this post that shows just how large the discrepancy is. Maybe a quarter of all the runs scored in Billy Hamilton's leagues are unaccounted for by simple Runs Created, whereas the percentage for Joe D. is about 4%.

And I've never seen anyone try to deal with this. That's what scares me. I've known there was a difference there for maybe 30 years, and I never studied it, but surely someone has dealt with something so straightforward. So what I'm asking for is help finding where it's dealt with. Do all era comparisons really need to reinflate the 1890s hitters, not to mention the earlier ones, because there's an error and running adjustment that isn't made? Surely someone else has spotted this. Can anyone help steer me to a study where this is dealt with? Because otherwise, right now, I have Hamilton ranked ahead of DiMaggio solely on this basis. Without it, Joe is clearly ahead, no matter how big a Billy Hamilton fan I am.

Um, thanks. - Brock

Here's the chart. What it shows are the year, for both Billy and Joe's careers, the simple Runs Created per actual Game for their leagues, the actual runs scored per actual game, and the ratio of actual to Created R/G. As you can see, the ratios are large in Billy's time, although dropping steadily, and small in Joe's time, and not moving much because there's not much room left to move. The actual R/G are always higher because there are always some error-driven and running-driven runs out there, but even by Joe's time, there aren't many. That's the problem. If you adjust for that, Billy Hamilton actually overtakes Joe DiMaggio, even though I give 3 years of war credit. Thanks again.

Hamilton
Year RC/G Actual R/G Ratio of Actual to Created R/G
1888 3.24 5.19 1.60
1889 4.12 6.07 1.47
1890 3.93 5.58 1.42
1891 3.94 5.54 1.41
1892 3.61 5.10 1.41
1893 4.92 6.57 1.34
1894 5.97 7.38 1.24
1895 5.15 6.58 1.28
1896 4.83 6.04 1.25
1897 4.80 5.88 1.22
1898 3.94 4.96 1.26
1899 4.30 5.24 1.22
1900 4.29 5.21 1.21
1901 3.89 4.63 1.19

DiMaggio
Year RC/G Actual R/G Ratio of Actual to Created R/G
1936 5.42 5.67 1.05
1937 5.13 5.23 1.02
1938 5.15 5.37 1.04
1939 4.97 5.21 1.05
1940 4.86 4.97 1.02
1941 4.62 4.74 1.03
1942 4.04 4.26 1.05
1946 4.05 4.06 1.00
1947 4.11 4.14 1.01
1948 4.56 4.73 1.04
1949 4.52 4.67 1.03
1950 4.92 5.04 1.02
1951 4.48 4.63 1.03

Why does this matter at all, bjhanke? Run estimators are mathematical equations that predict the number of runs a team will score based on its offensive component stats. They can either do a good job of this, or not such a good job. Clearly, your AB*OBP*SLG model is a pretty good match for the 1946 AL, and quite an atrocious one for all of Hamilton's leagues. But that says nothing about the comparative Merit of Hamilton and DiMaggio, and everything about your run estimator: if your equation is giving you a 60% error, then get yourself a new equation!

There are plenty of run estimators that do just fine for the 1890's. Just doing a quick spot-check on the 1894 NL, BP's EQR has a root mean square error of 51.9, or 5.3% of an average team's runs; (shameless plug!) my own modified version of BaseRuns that I use for my WARP has a RMSE of 40.0, or 4.1% of an average team's runs. (By contrast, in the modern game, RMSE's run about 2.9% of an average team's runs). If you plug Hamilton's numbers into a run estimator that is remotely capable of handling the leagues he played in, you will have no trouble comparing him to DiMaggio or anyone else (hint: DiMaggio's better). You might even look at the "batting wins above average" stat (particularly the one adjusted for the high standard deviation of the 1890's NL) available in my WARP spreadsheet posted to the Yahoo group, which is derived from that deadly-accurate BaseRuns estimator. :)

Tris Speaker improved his teams' chances to win in World Series play at the same rate that Barry Bonds improved his teams' chances to win in the regular season in 2001 and 2004. Postseason performance should be a very strong argument in favor of Speaker

Fine WPA analysis, Dan. Much appreciated!

But having said that...
1) most of us feel that WPA has its limitations (some woudl say it a lot stronger than that!!)
2) Your credit to Tris's fielding seems a bit generous. I haven't check thru the data to see his FPct yet; did you check for errors as well?
3) My 'fan-boy' comment was too strong, I apologize. But it DOES seem to me.. and please tell me if I'm wrong; that from your previous few days of posts that you are as much interested in spinning Speaker's accomplishments positively as you are in making sure we get it "right". I assure you, I'm not anti-Speaker; he's been waay underrated as a player, particularly if you compare him to Mantle. I just callz em as I seez em, and I need to dig thru the WS data more before I potnetially change my call. I will say you've given me reason to prioritize it before balotting closes, because I'd like to give as good an input as I can for this fine project.

Yes, please send your Speaker WPA spreadsheet to my home address. Thanks!

Mick and Tris home /road splits, courtesy of Bill Deane:

home AVG OBP SLG road AVG OBP SLG
M M .305 .430 .569 ------ .291 .415 .545
T S. .365 .449 .540 ------ .325 .408 .462

This does not take into account park effects. I don't have all of those, but perusing year-by-year seems to give Speaker about a 103 for his career (slight advnatge ot hitters), while Mantle about a 96. In contrats, Speaker created 36% (WOW!) more runs per game at home thruout his career. Don;'t know why. Myabe the official socrer was too kind? Mayeb he adjsuted well? Maybe his game was well-suited for his park(s)? Part of his large advantage was he hit forty-six percent more doubles at home. I have no idea if those parks were exrtremely doubles friendly.

All in all, in a nuetral setting, I'd say Speaker would have been a little less valuable than he was, compared to Mantle.

a. As far as I can see it, there are only two logically coherent approaches to take. Either you treat every plate appearance the same--in which case you basically just look at OPS+ or some slightly more advanced measure of it like BWAA--or you take into account the context of each plate appearance, in which case you use Win Probability Added. What other method is there? I personally don't tend to use WPA, but since you were the one talking about impact on a team's bottom line in the playoffs, WPA is obviously the way to address that. What is NOT fair is to somehow penalize Speaker for having a low RBI rate in the playoffs, as you did in an earlier post, when a more granular and detailed analysis shows that the timing of his hits actually provided substantial additional benefit to his teams in the postseason.

b. I did not check for errors, no; I merely assumed that Speaker was a league-average fielder (and thus had a league-average error rate) on all plays besides the four mentioned in my biography and the two double plays you noted. However, inspired by your question, I have just done so. Speaker made two errors in World Series play. They were:

1. In Game 7 of the 1912 WS, with the Giants up 10-4 in the top of the ninth and a runner on first base and none out, Art Wilson hit a single. Speaker made a bad throw to third base, which allowed the runner to score (making it 11-4), and Wilson moved up to second on the throw. The Red Sox already had literally an 0% chance to win the game trailing 10-4 with two additional runners on second and third with one out in the ninth, so Speaker's error did not cost them any WPA at all (for all intents and purposes, they had already lost the game).

2. In Game 8 of the 1912 WS, with the game tied in the tenth inning, the Giants had a runner on second with one out. Merkle singled, driving in Murray from second. Speaker's throw apparently missed the cutoff man, allowing Merkle to advance from first to second, and he was charged with an error. If he had hit the cutoff man, the Giants would have been up by one with one out and a man on first in the top of the tenth, giving them an 82.1% chance to win; Speaker's bad throw moved that runner up to second, increasing the Giants' chances of winning to 83.4%. Speaker's error thus cost the Red Sox .834-.821 = .013 of WPA.

However, remember that WPA is compared to a baseline of average. How much WPA would an average center fielder have cost the team by making errors in Speaker's opportunities? Well, Speaker had 53 PO + A + E in World Series play. The league average fielding percentage for outfielders in 1912, 1915, and 1920 was .957, so an average CF would have made (1-.957)*53 = 2.28 errors given Speaker's chances. Now, we don't know when those errors would have occurred or what type they would have been, so we have to assume they occurred in situations with a leverage of 1.00 and that they were league-average errors (of which I believe 60% allow a runner to reach base and 40% are advancement errors). An error allowing a runner to reach base is worth .082 wins, while an advancement error is worth .018 wins, so an average error is worth (.082*.6)+(.018*.4) = .0564 wins, times 2.28 errors, means that a league-average fielder would have surrendered 2.28*.0564 = .129 wins to the opponent on errors in Speaker's 53 chances. Since Speaker only gave up .013 wins to his teams' opponents, we can add a further .116 wins above average to his defensive WPA, bringing it to .75, plus the .854 on offense make +1.604 of positive WPA in 20 games. That is the same rate as a league-average fielder with a 239 OPS+.

c. I don't think Speaker's accomplishments need to be "spun" positively; they stand for themselves. All I am doing is drawing attention to them, particularly in terms of how they compare to the accomplishments of the other center fielders we are trying to rank him against. What have I said in any of these posts that seems to you to be stretching or exaggerating Speaker's record? If I were a Speaker "fan-boy" I'd have him ahead of Cobb or Mays, which I don't. I just think he absolutely leaves Mickey Mantle in the dust--the gap between Speaker and Mantle is far larger than the gap between Speaker and Mays or Speaker and Cobb--and I am doing everything I can to bring the rest of the group on board with that notion!

I'll zip you over the WPA sheet now.

bjhanke
Can anyone help steer me to a study where this is dealt with? Because otherwise, right now, I have Hamilton ranked ahead of DiMaggio solely on this basis. Without it, Joe is clearly ahead, no matter how big a Billy Hamilton fan I am.
. . . The actual R/G are always higher because there are always some error-driven and running-driven runs out there, but even by Joe's time, there aren't many. That's the problem. If you adjust for that, Billy Hamilton actually overtakes Joe DiMaggio, even though I give 3 years of war credit.
Several players from Billy Hamilton's time should show up as better creators of raw runs than Joe DiMaggio, perhaps his Phillie teammates Delahanty and Thompson rather than Hamilton.

But raw run creation shouldn't be important in overall comparison of players across teams (because there are park and teammate effects), much less across historical time. So the study of 1890s run creation should influence player comparisons --within team, across team, and across historical time all at once-- only insofar as it attributes error-driven and running-driven runs creation to different players in proportions that differ from the status quo attribution (eg, that by your "AB * OBP * SLG").

Here, much of the Harry Stovey discussion has been at that level, with sparse data and no systematic data. Did Stovey reach base on errors, advance on errors, and generally run the bases better than his teammates? --better in greater proportion than he bettered them at reaching base safely and batting for bases.

Fenway had a short left field fence, while League Park had a short right field fence. My biography of Speaker makes *explicit* reference to Speaker changing his swing when he moved from Fenway to League Park, seeking to pull the ball rather than inside-out it as he did in Boston. If anything, Speaker deserves extra credit for being able to modify his hitting style to his circumstances.

It's Mantle, not Speaker, who should be dinged for component park effects if we're going to be penalizing people, since he was hitting LH most of the time in the old Yankee Stadium. During his career, Yankee Stadium had a 96.7 PF, meaning he "should" have had a .409 OBP and .537 SLG at home given his .415/.545 on the road. In fact, he was .430/.569, meaning the component park effect was worth 40 points of OPS to him. I'd need data on how much of a boost the average LH power hitter got in the old Yankee Stadium to see whether Mantle took "extra" advantage of his park or whether he merely got the same benefit that all other players of his style and handedness did.

Regarding batter adaptation here is a discussion of Jimmy Wynn by "AG2004" at BaseBall-Fever.
Adaptation by Jimmy WYnn?

Dan says, "Clearly, your AB*OBP*SLG model is a pretty good match for the 1946 AL, and quite an atrocious one for all of Hamilton's leagues. But that says nothing about the comparative Merit of Hamilton and DiMaggio, and everything about your run estimator: if your equation is giving you a 60% error, then get yourself a new equation!

There are plenty of run estimators that do just fine for the 1890's. Just doing a quick spot-check on the 1894 NL, BP's EQR has a root mean square error of 51.9, or 5.3% of an average team's runs; (shameless plug!) my own modified version of BaseRuns that I use for my WARP has a RMSE of 40.0, or 4.1% of an average team's runs. (By contrast, in the modern game, RMSE's run about 2.9% of an average team's runs)."

Thanks! That was exactly what I needed. Yes, exactly, I needed a new equation. I just didn't know where to find one. So, um, can you tell me how to get to yours? "my WARP estimator posted to the Yahoo group" isn't enough info for me, because I don't know which Yahoo group you're talking about. I'll be happy to look yours up and use it. It's my fault that I'm about 10 years out of date on advanced methods, and don't know them, but all I can do about that, if I can't find one myself, is ask. I apologize for this, as it's a matter of timing. If I'd gotten into the HoM earlier, I would have had time to ramp up on the methods in use here. But I got here just in time to start filling out ballots, which takes up all the time I could have spent getting ramped up. The ballots have deadlines. - Brock

PS - If it helps, I agree with your Tris Speaker argument. As I understand it, you're saying that Speaker, no matter what his raw numbers in the postseason look like, managed to be in the right place at the right time, both on offense and defense. In a large-scale analysis, like a year or a career, I tend to try to avoid that kind of detail because it's so often a matter of luck, but in a short sample size like a postseason, it's completely relevant.

Paul says, "But raw run creation shouldn't be important in overall comparison of players across teams (because there are park and teammate effects), much less across historical time. So the study of 1890s run creation should influence player comparisons --within team, across team, and across historical time all at once-- only insofar as it attributes error-driven and running-driven runs creation to different players in proportions that differ from the status quo attribution (eg, that by your "AB * OBP * SLG")."

That's what I thought at first, but here's a made up example of what I see happening. I may be wrong, and would be happy to learn that I am, as it would save me a lot of work.

Let's say you have a run estimator that accounts for all of the runs in a Joe league. Let's say there are 8000 of them, and that happens to be an absolutely average number for a league in history. Let's say that Joe gets credit for 1% of those runs, or 80 runs. That stands, because Joe's in an average-scoring league. Now let's say that Billy is in a league where 10000 runs are scored, but your estimator can only account for 8000 of them. Let's also say that Billy, too, gets 1% of the runs, or 80. So far, all is equal. But then, as I understand it, the methods want to multiply Billy's runs by 8000/10000, as an adjustment for the league's run rates. That's 80 x .8, or 64 runs. That ain't right. What you want is for Billy to get credit for 1% of the 10000 runs, or 100, and then multiply by .8, to get him back to 80. The methods I've seen all basically do the wrong one, as I understand them. As I said, if I've got this wrong, please feel free to let me know where I went awry. I really needed about 6 months of ramp-up time on methods to do this HoM thing without asking for help all the time. Sorry about that. I'm trying to do the best ballots I can. - Brock

That's 80 x .8, or 64 runs. That ain't right. What you want is for Billy to get credit for 1% of the 10000 runs, or 100, and then multiply by .8, to get him back to 80. The methods I've seen all basically do the wrong one, as I understand them.

Wrong for what purpose? Those other people attribute league runs to particular players as a means to some end, I presume, and I don't believe that that is an overall player rating or an overall batter-baserunner rating.

I think you have in mind that Joe and Billy play in leagues of equal size, so that their shares of league runs 1% and 1% are directly comparable. If you are using player runs directly as some rating, and you are arriving at player runs via calculation of runs shares, I think you can use the run shares directly. Joe's rating = Billy's rating = 0.01; the unit is share of 8-team league batting and baserunning. Right?

It's Mantle, not Speaker, who should be dinged for component park effects if we're going to be penalizing people, since he was hitting LH most of the time in the old Yankee Stadium. During his career, Yankee Stadium had a 96.7 PF, meaning he "should" have had a .409 OBP and .537 SLG at home given his .415/.545 on the road. In fact, he was .430/.569, meaning the component park effect was worth 40 points of OPS to him.

Don't we need to add in the 'typical' home field advantage to convert to what he "should" have hit in Yankee stadium?

Brock, Bill James had a number of run estimators for pre-1920 in the original (but unfortunately not the "New") historical abtstract. Not sure how well they worked for 1890s, but I'm quite certain they were gobs better than the basic RC.

Well, using BaseRuns can be a bit complicated, particularly because you have to customize it for each league-season. If all you are looking for is a reliable measure of offensive performance going back to 1893 (that's my cutoff date), then my batting (and baserunning) wins above average should do just fine. Go to groups.yahoo.com and search for hallofmerit, and sign up for the group. Then if you go to "files" you can download the Rosenheck WARP.zip archive, which has two files. The first, Rosenheck WARP Results.csv, gives the following information for every position player-season with over 50 PA since 1893: batting wins above average, baserunning wins above average, fielding wins above average, wins above average a replacement player at the same position would have accumulated in the same playing time, and the sum of the first three minus the fourth which is wins above replacement, in both a "raw" version and one that is adjusted for the standard deviation/ease of domination of the league, to facilitate cross-era comparisons. The second, StDevs and Rep Levels.xls, has two sheets: one shows both the actual standard deviation and the one projected by my regression equation (which is the one used to make the adjustments in the prior sheet) for every league-season since 1893; the other traces positional replacement levels from 1893 to the present to give a visual depiction of the evolution of the defensive spectrum. If you are interested in learning more about my system, there is a thread merely 580 posts long at http://www.baseballthinkfactory.org/files/hall_of_merit/discussion/dan_rosenhecks_warp_data/, which includes a step-by-step example of how the numbers are calculated at post #491. If you *really* want to do your own run estimation for the 1890s, I can show you how, but it's rather convoluted.

Well, I don't tend to pay much attention to the postseason in general in my rankings, but TomH said that the reason he had Mantle over Speaker was because of the gap in World Series value. I countered that Mantle didn't do any better in the WS than Speaker; they both lost 8% off their OPS moving from the regular season to the postseason. TomH then said that he discounted Speaker's postseason OPS because he had very few RBI. I replied that just looking at the percentage of runners driven in is a terribly poor way to assess the contextual value of offensive events--it's essentially just an extremely ineffective brute-force approach to guesstimating leverage index, when in fact we have the play-by-play records and can determine actual Win Probability Added for Speaker's entire postseason career. That analysis showed that far from his World Series hitting being less valuable than simple OPS would make it appear, as Tom was arguing due to Speaker's low RBI total, the reality was quite the opposite: that Speaker's World Series hitting was MORE valuable than simple OPS would make it appear, since he had the decisive hits in both Game 2 and Game 8 of the 1912 WS. I then further added that TomH was not taking appropriate account of Speaker's defensive value in World Series play, where we know that he made at least one catch which completely determined the outcome of a game (Game 2 of the 1915 WS), and that furthermore we have knowledge of three other "nice running catches" he made, and that beyond that, his errors were much less costly than those that would have been made by an average center fielder. When you add up all those factors, Speaker's impact on his teams' won-lost record in World Series play was the same as a league-average fielder with a 240 OPS+ whose hits were distributed randomly.

Now, to be clear, this isn't why *I* have Speaker above Mantle--I put Speaker ahead because he was just overwhelmingly more valuable in regular season play, with the same offensive contribution as Mantle and a mere 22 wins better in the field. But I went to all this trouble to point out that TomH is not applying his own criteria consistently, since a complete analysis of Speaker's postseason contribution to his team's championships shows he was a true World Series hero.

TomH, that's completely right, but the same would also apply to Speaker, so that factor washes out. On this point, where I really think you are being contradictory is that you are lauding Mays for his ability to adjust his defensive approach to maximize effectiveness in three different home parks, while dinging Speaker for adjusting his offensive approach to maximize effectiveness in two home parks that could not have been more differently shaped.

bjhanke, I don't see why you would waste your time with an estimator that is "missing" 20% of the runs scored in the league! Why not just use a run estimator that works, and then stop worrying about how to make adjustments to try to fix the broken ones?

I completey disgaree that I am being contradictory; I am remarkably consistent in my position of value versus ability.

If you vote 100% for value, you take Gavy Cravath's home-inflated big fly totals and count 'em all. Which is one legit way to go.

If you vote some percent of "ability", you attempt to put Cravath in a more neutral park and imagine what he might have done. Yes, it's murky. But it's legit.

Joe DiMaggio was hurt - a LOT - by his home park. Some people give him bonus credit because he would have been better in almost ANY other place thank Yankee Stadium, 1940s. Fine with me if you do or don't.

Willie Mays moved into parks which initially hurt his stats. He adjusted, tho, (hitting more to RC), so he did not lose value for too long. He likley would have been more valuable in most other neutral parks. I give him a bit of creidt for that; not much tho.

Tris Speaker may have taken more advantage of his home parks than any other top 25 player in major league history. The OPS shows this; the facts are not up for debate. Now, Speaker MAY have been so good at adapting that he DOES deserve 100% credit for his ability to thrive both at Fenway and League parks, with their different dimensions. But I choose to move about 10% of this uber-bonus-value into the 'he was real fortunate' category. Slag me if you think this is improper.. but it sure ain't contradictory.

At this point, I really ought to move to other things besides Speaker/Mantle comparisons.

Interestingly, I think the real conclusion to draw is that Speaker was a skilled enough hitter to take extra advantage of any quirky park--for which he deserves credit--but that he would not have been able to make use of that ability had he played in one of the perfectly neutral, symmetrical, cookie-cutter stadiums that prevailed in the 1970's.

Paul says, "Wrong for what purpose? Those other people attribute league runs to particular players as a means to some end, I presume, and I don't believe that that is an overall player rating or an overall batter-baserunner rating."

Oh, good grief. I don't believe I've done this. Thanks, Paul. I don't know how I missed this, but I did. It took your wording to wake me up. When I saw your comment, my first response was that yes, as far as I know, the purpose is to make overall player rankings, all of which are wrong without this adjustment. Then it occurred to me to check that out, just to make sure. And you know what? I've missed this for at least 6 years, and presumably more. It shows up in the first place I looked, in Win Shares, page 92. Bill is going through his Runs Created method in detail and there it is:

"F3 Take the ACTUAL RUNS SCORED BY THE TEAM (emphasis mine)
F4 divide F3 by F2 (F2 is the team Runs Created)"

I completely missed when that step got added into RC, and it is exactly the adjustment that I was talking about missing. It wasn't there in the original version of RC, but then, the original was only being used on modern players. For all I know, it goes all the way back to the first Historical. For all I know, Pete Palmer was doing this when he first started to compare eras, back I don't know when. For all I know, everyone's been doing this, or the equivalent, for a decade or more. And I completely missed it. I am really sorry to have wasted everyone's time on this, which is simply my blindness. All I can guess is that I've "known" how RC was computed since the 1980s (the original A, B, and C component formula is algebraically equivalent to AB x OBP x SLG, which is where I got that formula), and just assumed that all later versions, like Bill's 13 in the old Historical, differ only in the data set that is available for the A, B, and C components. In general (I checked the Win Shares formula this time) that's true, but the era adjustment shows up in the part above about actual team runs scored.

Again, sorry about that tempest in a teapot. Please go back to your normal ballot making. Forget I ever posted this nonsense up, and please forgive me for wasting your time. - Brock

Oh, yeah, Dan. Thanks for pointing me to the yahoo group, and for the VERY thorough directions as to what, exactly, is where. I will sign on to the group ASAP and go through your work. ASAP may mean after this HoM ballotting is done, but I'll do this because your comments and analyses seem to be really really good. Also, you're exposing your entire method, which is a huge thing for me. I basically refuse to work with methods that aren't exposed in detail, because I'm being asked to take someone's work on faith. I'm not the best mathematician in the world, but I'm usually good enough to examine a sabermetric method. So I check them out. This, BTW, is why I keep coming back to Bill James. His work may not be the absolute best at any given time, but at least you can figure out what he's doing and why. He exposes his methods, unless the Red Sox won't let him or something. That's an absolute for me. And yes, this is my way of paying you a compliment for exposing your work to criticism. I can't imagine how much time you spent going through a 500-post thread defending your work against all comers. And I appreciate it. Hopefully, at least a few of the comments were real help, and your work is the better for them. That would be the payoff for the Complete Ethics of exposing your work. Thanks, Brock

Dan says, "but that he (Speaker) would not have been able to make use of that ability had he played in one of the perfectly neutral, symmetrical, cookie-cutter stadiums that prevailed in the 1970's."

Oh, I don't know. I think he could have figured out that there is a lot of real estate between the infield and the outfield, if you can just hit flat line drives, and that the astroturf makes hard ground balls zing right through the infield, and that you can steal bases with impunity if you're fast and alert. I also imagine he would have figured out fast that he was going to have to back up on defense just a little, and sacrifice some singles and all the DPs to prevent triples. Of course, Mantle, who wasn't an idiot, would have figured out all this, too. But Speaker's non-uppercut swing might have taken less time to adjust than Mantle's homer-generator would have. But then again, Mantle probably had enough power to get the ball out of even the giant cookie-cutters. Hard to say who would have prospered most or fastest.

Turf outfield vs. a few old-fashioned drains.

Mantle post-season analysis:

(...more accurately, as much analysis as I'm gonna make time for)

It is well known that Mantle holds a significant amount of World Series
batting records. If you need to, check out the postseason hitting
leaders page of bbref, and behold, it's all there.

It was based on this that I had been giving the Mick a large boost in my
rankings. 42 runs scored and 40 RBI in 200+ ABs, with 18 big flies, is
certainly impressive.

But as I check the numbers... he really was not very clutch with his
runs produced. In his first 5 WS (one aborted by injury), he drove in 14
runs while hitting 8 dingers (and 16 other hits). 14 is not as many as he "should" have. He did this while coming to bat in the 3 hole (occasionally leadoff!) of a lineup that put just over .5 runners on base on average per PA (57 runners in 104
non-intentional walk PAs).

Eleven of his total 40 RBI were claimed in the famous 1960 series, which of
course the Yanks lost on Maz's HR. Many of those 11 were in Yankee
blowout wins.

Mantle did help his teams win some World Series, and he did perform well
against some fine NL pitching staffs. But I will lessen the post-season
bonus I had given him, and submit a revised ballot.

Tom -

If it helps any, Mantle played extremely well, and quite clutch, in the 1964 series, when he was a 33-year-old alcoholic with bad knees. He hit a homer in the ninth, off closer Barney Schultz, to win one game 2-1. I'll say that's clutch. In another Yankee win, the score was 1-1 in the 6th, I think, when Maris and Mantle went back to back to break the game open. Sure, Maris got there first, by one batter, but that's another clutch hit. And in game seven, when it was all on the line, the Cards got ahead 6-0, behind Bob Gibson, no less. Mantle halved the deficit with a three-run shot. That was it for the Yankee offense until the 9th, when Gibson, exhausted but finishing the game on grit, gave up homers to - no, really - Clete Boyer and Phil Linz. In other words, with the Series on the line, Mantle was pretty much the whole game 7 Yankee offense. They didn't win, but that was hardly the fault of the aforementioned 33-year-old alcoholic with bad knees.

Oh, and thanks for reminding me of Bill James' 13 Runs Created formulas in the old Historical Abstract. I assume that's what you were talking about when you posted up in my hour of embarrassing need.

- Brock

hey Brock, if that's the worst example of 'hour of embarrassing need' you ever have, you're in for a fine life.

By the way, I wrote an article for a SABR pub 2 years ago (I'll sned a pdf if anyone wants it), analyzing the best defensive CFers ever; at least, using the metrics available for the entirety of MLB (lacking the pbp data we have for the modern guys).

Speaker and Mays did indeed come out as virtually tied for top 2.

#3 would be debatable. Depending on how you give War credit, the ratings for best consecutive 8 years ("prime") included Andruw Jones and Dom DiMaggio as the top 2. For career marks, DiMaggio was 3rd and Curt Flood fourth.

My guess is that Dominic D was the 3rd best ballhawk ever to play the game.

Just a quick note to everyone the election has been extended through 9/21. Thanks!

Quoting Brock's ballot

from Ty Cobb:
I have a theory on the early AL's quick rise to prominence. I believe the AL was the first league to really mine the Deep South for players. It's like the NL with black guys in the 1950s. Cobb, Speaker, Eddie Cicotte, and a few others appear to be unmatched in terms of origin by anyone in the NL at the time. It would explain why the AL was so successful so quickly. As opposed to other leagues competing with the NL, the AL did not just try to raid NL players; they looked in an area of the country that was unused by the NL. So my question is, as always, has anyone actually studied this in any systematic detail? I'd love to know, but know perfectly well that I will never spend the time to find out unless someone comes up with a database with a column for state of birth. Thanks!

The database that Sean Lahman provides for download at The Baseball Archive includes a column for state of birth in its master table. That database and the player, team, league pages at baseball-reference both use the same tables of data (see baseball-databank.org if interested in more than the workable database).

As far as I know, no one has studied geographical origins within the USA systematically and published the findings.
Bob Caruthers was born in the South, moved to Chicago as a schoolboy. Billy Nash and Steve Brodie were southerners: from Virginia, neither the new state of West Virginia (Glasscock) nor the US Capital region (Hines). Ned Garvin and Noodles Hahn came to the NL from Texas and Nashville TN in the late 1890s. That is at least four star players. Johnny Dobbs was not a star but a good player. The Southern League was well-established by the late 1890s and so was the southern trip for NL teams, either spring training at a fixed site or playing practice games on tour. I believe that in 1890 as opposed to 1900 there was not much contact between mlb people and the interior of the South.


19. Pete Browning
. . .
The only reason I can think of for playing Browning in center field 40% of the time is that your corner outfielders can't run. In the AA, this may have been true. The right fielder may have been largely the change pitcher, so it would only take an immobile left fielder to make Browning into possibly the worst center field glove ever. Essentially, I think of Pete Browning as what would happen if you took David Ortiz or Prince Fielder and gave him speed and then put him in center field. You'd find out pretty quick that it is indeed possible to pile up an F defensive ranking in center. It's just that no one will play a glove that bad at that spot. Except for Pete Browning's managers.

In Louisville Pete Browning's rightfield partner was Jimmy Wolf (at bb-ref), one of the few 10-year men in the AA.


21. Earl Averill
Sort of a very lite Yaz. Wildly inconsistent during his good phase, which is short. But the good years are pretty good. Not much showing in other seasons.

Inconsistent in what respects? I think of him as one of the most consistent players, based on his playing time and his batting. Recently we have learned that Davenport (FRAA and Rate) and someone else cited by DanR see him as an inconsistent fielder with a couple of good seasons in the middle of his major league tenure that seem "out of context".


25. Lip Pike
Played 425 games of high-quality league baseball. But there are only 5 seasons of any real worth. If you doubled his games played, to adjust for short schedules, he'd still end up about here. Just a very, very short career.

Five seasons? So you begin counting in 1871. Why not seven seasons, 1871-77?