Baseball for the Thinking Fan

Login | Register | Feedback

You are here > Home > Hall of Merit > Discussion
Hall of Merit
— A Look at Baseball's All-Time Best

Monday, November 22, 2004

John Beckwith

Another quality shortstop to muddy up the waters for us.

John (You Can Call Me Grandma) Murphy Posted: November 22, 2004 at 02:47 PM | 382 comment(s) Login to Bookmark
  Related News:

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

Page 1 of 4 pages  1 2 3 >  Last ›
   1. TomH Posted: November 22, 2004 at 04:14 PM (#977242)
Any major league comparison? A short-career Dick Allen who could play some shortstop? Howard Johnson with more power but the attitude of Latrell Sprewell?
   2. Chris Cobb Posted: November 22, 2004 at 05:25 PM (#977417)
John Beckwith data

From Holway

1917 .000 for Chi Giants; ss
1918 no data
1919 .188 for Chi Giants; c
1920 .280 for Chi Giants; ut
1921 .355 for Chi Giants; ss
1922 .305 for Chi Am Giants; ut
1923 .330 for Chi Am Giants, 24 2b, 9 3b, 14 hr; 3b (should be, but isn’t, all star)
1924 .382 for Bal Black Sox, 5 hr, 16 hr/550 BA leads league; ss (should be, but isn’t, all-star)
3-8, .375 with 2 HR vs. Phil A’s
1925 .406 for Bal Black Sox, 24 HR lead league, 22 2b, 50 hr/550 ab; ss, Holway all star, MVP
(constructed batting line, 264 ab, 107 hits, 22 2b, 8? 3b, 24 hr, .406 ba, .822 slg)
1926 .311 for Harrisburg Giants, 3 triples; 3b, Holway all star (as dh)
1927 .362 for Harrisburg Giants, .(223 for a few games with Homestead Grays), 9 hr, 18 2b, 16 hr/550 ab; 3b, Holway all star (constructed batting line, 309 ab, 112 hits, 18 2b, 6? 3b, 9 hr, .362 ba, .547 slg)
2-10 vs. major-league All-Stars
1928 .240 for Homestead Grays; ss
2-17 vs. major-league competition
1929 .439 for Homestead Grays, 15 hr, 25 hr/550 (330 ab, 23? 2b, 7? 3b, 15 hr, .439 ba, 144 hits, .684 slg) ; ut
1930 .493 for NY Lincoln Giants, 6 hr, 47 hr/550 ab; 3b, Holway all star, missed time with broken ankle
1-8 in playoff vs. Homestead
1931 .350 for Baltimore & Newark Browns, 16 hr, 7 2b, 53 hr/550 ab); 3b, Holway all-star
5-10 vs. major-league all stars
1932 no data (teams he played for were not in the East-West league)
1933 .391 for NY Black Yankees; 3b
1934 .286 for NY Black Yankees; 3b
1935 no data, listed in Riley as playing for the Grays, but apparently not a regular

Positional data
C – 1 year, 1919
SS – 5 years, 1921, 1924, 1925, 1928 (1917 no ml credit)
3B – 7 years, 1923, 1926, 1927, 1930, 1931, 1933, 1934
ut – 3 years 1920, 1922, 1929

Career data
.352 lifetime avg. 767-2176 according to Holway
mean avg. 1919-34 = .366
19-60 vs. major-league competition (.319)
80 hr in 2198 ab, according to Holway
   3. ronw Posted: November 22, 2004 at 05:36 PM (#977443)
It seems that Holway honored Beckwith with All-Star selections much more often than his contemporary Judy Johnson.

Likewise, Jud Wilson is honored with All-Star selections much more often than his contemporary Ray Dandridge.

Bill James (I think) has Johnson #1 and Dandridge #2 for his all-time Negro League 3B. The HOF also obviously had the same rankings.

Who is right?
   4. Chris Cobb Posted: November 22, 2004 at 06:26 PM (#977558)
On Beckwith vs. Johnson, I think for the purposes of the Hall of Merit Holway is clearly correct.

Beckwith was a _much_ better hitter than Johnson, although he was a poor fielder.

Beckwith does less well than he might with James and the HoF because of "character" issues.

Wilson and Dandridge were not really contemporaries: Wilson started his career a decade before Dandridge; they are not really competing against each other for all-star slots. Dandridge's all-star totals may be diminished because he spent a number of his best seasons playing in Mexico.
   5. sunnyday2 Posted: November 23, 2004 at 02:46 AM (#978487)
My sense right now is that Johnson and Dandridge have been somewhat overrated by history, sort of in the way Bingo DeMoss has been--i.e. somebody had to be "the best" 3B (or 2B) and whoever was chosen from a weak field was naturally overrated. It looks to me that Wilson was better than Johnson and Dandridge, but that's not important right now.

Mainly I just wanted to say WOW! Beckwith was a much more serious dude than I thought. I think it is correct to say that history has buried the guy because of his alleged personality (and I don't say alleged in the sense that I think the image is wrong, just that I wouldn't know).

That and the fact that he didn't have a regular position so nobody could ever say that he was the best 3B or the best whatever. So he kinda got forgotten even by those who didn't actually WANT to bury him.

Anyway, he looks almost like Dobie Moore to me, with obviously weaker defense, but with a slightly longer career. But only slightly given Moore's 7 years in the army and that it looks like Beckwith's first couple and final seasons weren't much. But then, Beckwith's documented career is considerably longer (now it's 15-7 versus 15-14). But right now, for me, I see Beckwith sort of coupled with Dobie who IS on my ballot.

So, again, it's early but I see Rogan near the top and Beckwith near the bottom, but both on the ballot.
   6. Michael Bass Posted: November 23, 2004 at 03:52 AM (#978548)
Definately interested in a Win Shares analysis on him. I think he'll make my ballot, but I'd love some hard numbers to decide for sure and to place him if he does make it.
   7. JoeD has the Imperial March Stuck in His Head Posted: November 23, 2004 at 04:14 AM (#978560)
I agree Michael, I'm really looking forward to the Win Shares estimate on Beckwith.
   8. John (You Can Call Me Grandma) Murphy Posted: November 24, 2004 at 02:26 AM (#980423)
I'm with Joe and Michael here. A Chris Cobb WS estimate analysis of a Negro Leaguer is always appreciated.
   9. Chris Cobb Posted: November 24, 2004 at 03:07 AM (#980442)
I'm glad the estimates are useful. I've started work on Beckwith now; I expect I'll have the estimate ready sometime tomorrow.
   10. KJOK Posted: November 24, 2004 at 05:00 AM (#980489)
Beckwith's 1928 numbers above are definitely way off:

1928 Beckwith
AVE-.343 (NeLgAve .279)
OBP-.370 (NeLgAve .334)
SLG-.486 (NeLgAve .384)
OPS-.856 (NeLgAve .717)
   11. KJOK Posted: November 24, 2004 at 05:06 AM (#980491)
1928 Beckwith Fielding:
Inn - 98


   12. Chris Cobb Posted: November 24, 2004 at 05:42 AM (#980530)
Thanks, KJOK!

1928 did look like an anomalous off-year in Beckwith's records. KJOK's data is much more in line with the rest of Beckwith's record, so I'll use it in place of the Holway data for 1928.
   13. Dr. Chaleeko Posted: November 24, 2004 at 02:49 PM (#980861)
Even with his poor defense, Beckwith's an awfully impressive-looking player. Here's a quick snapshot of what Beckwith's WS might have looked like.

I used his i9s projection and ran it through short-form WS with no career shaping of any sort. I'll leave that task to more expert hands. For his defense, I assumed that he netted 2 WS for every 1000 innings he played, or 2 per 118 games.

So I created three possible scenarios for Beckwith's career, the first using Chris Cobb's standard 5% discount on hitting stats, and the other two using familiar contemporary assessments of minor-league playing level as a benchmark. I'm working from memory about the discount rate for AAA and AA, so if I'm a little off, please let me know, and I'll happily recalculate and repost the data.




Personally, I think this range of possibilities describes an extremely impressive peak and a solid career total that certainly would not keep him from being a serious HOM contender. If this range of possible values was absolutely accurate, he'd probably be in my top three or four this year.

That said, I'd prefer Chris Cobb and others to weigh in with their estimates before I finalize my thinking on Beckwith.
   14. Chris Cobb Posted: November 24, 2004 at 02:57 PM (#980867)
John Beckwith Win Share estimates

Basic Principles

1) These estimates are based on actual data whenever possible; I’ve used the Holway data and KJOK’s data posted above; when that data was insufficient to calculate a slugging percentage, I’ve used i9s, adjusted by the same # necessary to bring their projected BA in line with my projection. I’ve used i9s for projected playing time in each season, except in cases where I believe i9s to be inaccurate by a large amount (individual cases explained below).
2) To make the batting win share estimate, I find Beckwith’s MLE batting average and slugging average by prorating his NeL values by .87. I then find his closest major-league match, trying to match obp (using i9s as a guideline for walks) as closely as I can also. Generally, I use NL in odd years and AL in even years. I find the bws for the matched player, and prorate them for Beckwith according to plate appearances, sometimes adjusting by 5 to 10% up or down when the match is obviously imperfect.
3) To make the fielding ws estimate, I calculated defensive innings from plate appearances (8.5 defensive innings/4.2 pa) and assigned Beckwith fws according to my best guess at his overall fielding value. He played third base, shortstop, catcher, and first base. I rated him a C third baseman for his career (3.0 ws/1000 def innings), a D shortstop (3.8 ws) and catcher (3.6) and a B- first baseman (1.7). I then assigned fws to create a gradual decline from an early fielding peak that would match these career averages at each position.

John Beckwith’s win shares

Year (games) BWS + FWS = Total
1919 (18) 0.0 + 0.4 = 0.4
1920 (127) 7.8 + 4.3 = 12.1
1921 (149) 18.1 + 5.0 = 23.1
1922 (131) 9.8 + 4.1 = 14.3
1923 (144) 23.7 + 4.3 = 28.0
1924 (109) 18.8 + 3.5 = 22.3
1925 (140) 29.1 + 4.4 = 33.5
1926 (117) 12.7 + 3.2 = 15.9
1927 (151) 20.1 + 4.1 = 24.2
1928 (128) 13.5 + 4.0 = 10.5
1929 (146) 24.9 + 4.0 = 28.9
1930 (100) 18.3 + 2.6 = 20.9
1931 (149) 24.9 + 3.0 = 27.9
1932 (100) 8.6 + 1.9 = 10.5
1933 (95) 13.9 + 1.3 = 15.2
1934 (54) 2.0 + 0.8 = 2.8
16 (1858) 246 + 51 = 297

Notes on Playing Time and positions
C 1919, 1920
SS 1920, 1924, 1925, 1928, 1929
3B 1920, 1922, 1923, 1926, 1927, 1929, 1930-34
1B 1922, 1923
1919 treated as a token season, not as partial season as in i9s because BA doesn’t justify a half season of play.
1920 treated as full season, not as partial season as in i9s because there’s no indication he wasn’t a full-time player and his BA was good enough to be a starter at key defensive positions. Even playing time split among the 3 positions
1922-3 80-20 3b/1b split
1923 treated as full season, because Holway has stats for a full season. The Riley bio reports that Beckwith “got into trouble with the law and left Chicago” when he had played “less than two full seasons for the American Giants.” If he did leave before the season ended, it was nearly over.
1929 40/60 ss/3b split
1933 added 20 games to the i9s projection. It looks like Beckwith hit pretty well this year, so I decided i9s decline curve was too steep, esp. since Beckwith played in 1935.
   15. Chris Cobb Posted: November 24, 2004 at 03:05 PM (#980875)
Quick analysis and comparison to Dr. Chaleeko's results:

1) i9s boosts Beckwith's home run numbers way too much, more than their usual 5% overvaluing -- by my estimates, they give Beckwith something like 80 hr more than they should. He was an excellent hitter -- his season by season comps are Heilmann, Hornsby, Frisch, Wilson, Foxx in his best years and Frisch, Lazzeri, Cecil Travis in his down years, but he wasn't a Foxx-type home run hitter for most of his career as i9s thinks. So that's the main reason for the lower values here.

2) Despite these lower values, Beckwith looks like a solid candidate, perhaps not for immediate election, but he ought to be making a lot of ballots.

3) His fielding value was higher than Dr. Chaleeko estimates, in my view, because he was playing key defensive positions. Even as a below-average player at those positions, his defensive contributions would be significant.
   16. John (You Can Call Me Grandma) Murphy Posted: November 24, 2004 at 03:11 PM (#980882)
He looks a little better than Sewell and Moore, so he looks like he'll be near the top of my ballot.
   17. Dr. Chaleeko Posted: November 24, 2004 at 03:33 PM (#980912)
Thanks Chris!

Man, I thought the 9s looked a little puffy at the usual 5% discount, but 80 HR is a boat load of SLG!!!
   18. Dr. Chaleeko Posted: November 25, 2004 at 12:48 AM (#981587)
So, just thinking out loud here...we've got a guy here who
-is very good offensive player
-is an infielder
-is a questionable fielder
-had about a 14-15 year career
-finished up around 300 win shares.

Sounds a lot like Laughing Larry Doyle to me, only at a more valuble defensive position.
   19. Chris Cobb Posted: November 25, 2004 at 02:11 AM (#981666)
In my rankings, Beckwith lands squarely between George Sisler and Larry Doyle.
   20. Gadfly Posted: November 25, 2004 at 07:13 PM (#982431)
Notes on John Beckwith:

John Beckwith was born in Louisville in 1900. When he was young, Beckwith, his mother, and his siblings all moved to Chicago. Either his father died or left the family, I've always wanted to know which. In any event, he grew up fatherless.

John Beckwith had an older brother named Stanley who was a pretty good player himself. Stanley was playing for various Chicago teams in the mid teens (and his career has gotten messed up with John's), and he encouraged John to also play ball.

But Beckwith wanted to be a boxer. Beckwith evidently changed his mind after sparring with Sam Langford (Langford, known as the "Boston Tar Baby," was the most savage black boxer of the day and had fought numerous times with the black heavyweight champ, Jack Johnson).

Beckwith, who wasn't stupid, took his brother's advice and became a baseball player instead, but always had a boxer's mentality. John Beckwith joined the Chicago Giants in 1918 and played baseball until 1942.

Notes on his career:

Beckwith was about 6 feet tall and 210 pounds in his prime. He weighed around 190 in the early 1920s and was up around 220 to 230 in the 1930s. He is usually listed at 6 foot 3, but he wasn't that tall.

Beckwith is listed as batting and throwing right-handed. But there are several articles from the 1920s that refer to him as a switch-hitter. However, all the photographic evidence simply supports him being right-handed.

John Beckwith had, shall we say, a unique personality. He wasn't some malcontent like Dick Allen (who is a great comp as a hitter). Beck was basically quiet, but he wasn't going to take crap or disrespect from anybody.

Beckwith, I think, has been slandered by various sources ("pimp, possibly a bootlegger"); but it should be noted that, from 1924 to 1942, Beckwith was usually the manager of his teams.

Beckwith slid down the defensive spectrum during his career. He started as a shortstop and sometimes catcher, was a third baseman during his prime, and was a first baseman and sometimes catcher again in his later years.

In my opinion, Beckwith was the greatest third baseman from the Negro Leagues (with Jud Wilson as number 2). In 1952, the Pittsburgh Courier ran a poll and named the top 3B as Oliver Marcelle, Malarcher, Dandridge, and Wilson. This poll has muddied up the waters ever since.

Two things should be pointed out about this poll. One is that it is evident that the voters considered the question as "Who was the greatest defensive 3B?" Two is that the voters could vote the players in whatever position they wanted.

Beckwith got voted in as the greatest utility player. Wilson was actually listed as one of the greatest 1B and Dandridge as one of the greatest 2B. In other words, the support for Beckwith, Wilson, and Dandridge was all diluted.

Personally, I'd put them 1) Beckwith, 2) Wilson, 3) Dandridge, 4) Marcelle or Johnson.

Beckwith (and Wilson) have both been hurt by the fact that they played various positions and were not considered pure third baseman.

Notes on Beckwith (2):

During the 1920s, John Beckwith played for a lot of different teams: Chicago Giants, American Giants, Harrisburg, Homestead, Baltimore Black Sox, Lincoln Giants.

It should be noted that this wasn't because teams did not want him, but because Beckwith wasn't going to let anyone pay him one cent less than he was worth. Whenever John Beckwith was feeling unappreciated, he would jump the League to play for the independent Homestead Grays, returning when things were worked out to his satisfaction.

Beckwith evidently went through women like he went through teams. In 1929, his wife tried to stab him to death while he was playing in the California Winter League. Beckwith, who was supposed to be on the brink of death, came back to hit 2 Home runs in the deciding game of the season, shortly after being stabbed.

Beckwith evidently divorced that wife because, in the early 1930s, Beckwith married his wife Dorothy. They settled in and lived in Harlem. This is an important fact in Beckwith’s career because, with one exception, Beck never left the New York area after remarrying.

From 1931 until the end of his career, Beckwith mostly just played for New York based teams: the Lincoln Giants, Newark Stars, New York Black Yankees. In 1935, he did begin the season with his old friend, Cum Posey, for the Homestead Grays.

Posey wanted Beckwith to be his Josh Gibson, who was then playing for the cross town Pittsburgh Crawfords. Beckwith showed up with his catching gear and demanded that Posey reimburse him for having to purchase it. Posey refused. The relationship ended quickly.

In 1936, Beckwith became the manager of the independent Brooklyn Royal Giants. From 1936 to 1942, Beckwith played and managed the Royal Giants or his own team, the New York Stars or John Beckwith’s Stars. He could still hit. In the late 1930s, it was noted that Beckwith had gotten twelve straight hits as a pinch hitter for his team.

Beckwith died in 1956 of cancer, stilled married to Dorothy and living in Harlem.
   21. Gadfly Posted: November 25, 2004 at 07:36 PM (#982440)
Notes on Beckwith (3):

Beckwith, in his prime, was the best hitter in the Negro Leagues.

One interesting thing to note is that, in 1929, Beckwith finished second to Chino Smith in BA and SA for the American Negro League. During that season, Smith played for the Lincoln Giants in the Catholic Protectory Oval, an all-time bandbox.

Beckwith played for the Lincoln Giants that year too. But only for the last week of the season. He was traded from the Homestead Grays to the Lincoln Giants in early September. He spent most of the year playing for the Grays in Forbes Field, an all-time pitcher's park.

If Beckwith had played the whole year with the Lincoln Giants, he would have lead in everything.

In 1930 and now playing in the CPO full time, John Beckwith was off to a tremendous start, outhitting Smith easily. But then he missed half the season with a leg injury.

From 1931 to 1942, Beckwith never really played with League teams any more. He basically just played where he could 1) stay in NY and 2) be in control.

In a sense, the second half of his career is missing.

So the question is: "What would Beckwith's career have looked like if he had played in the Majors." There is one contemporary player who is an excellent comp for Beckwith. That is Rogers Hornsby. Only compared to Hornsby, Beckwith had less speed, a stronger arm, and more power.

(Hornsby is also an excellent comp for Oscar Charleston as a hitter.)

In his prime, Beckwith would have been hitting 40 HRs a year and batting from .350 to .400. There is no telling what would have happened if he had played in the National League in 1930.

Like Hornsby, the second half of Beckwith's career is messed up by injuries and being a manager. But, if they had paid him well, John Beckwith would have probably just kept playing.

I think, if he had played in the Majors, Beckwith would have still been playing in the late 1930s and hit well over 500 HRs total.

But that's just my opinion. It's also my opinion that Beckwith is one of the most underrated of all the Negro League players.
   22. Chris Cobb Posted: November 25, 2004 at 08:32 PM (#982475)

Thanks for chiming in on Beckwith! I hope you may be able to give us more details about Joe Rogan's army years?

What are the sources on Beckwith's career after 1931 and especially 1935?

I ask because Riley paints a very different picture of his later career:

After his playing skills began to erode, he ddropped out of the lineups of top teams but hung on as manager of a team of dubious quality, the Crescents of White Plains, New York. When he left baseball entirely, he worked briefly as a policeman in New York, but eventually reverted to activities on the other side of law enforcement that involved loose women, dice games, and bootlegging.

(Gotta love the old-fashioned vocabulary Riley trots out from time to time.)

The limited statistical evidence in Holway is ambivalent about the extent to which Beckwith was declining after 1931, so I'm prepared to consider the idea that Beckwith's major-league equivalent play lasted longer than I've credited him for, but I can't do that without more evidence.
   23. Michael Bass Posted: November 26, 2004 at 10:59 PM (#983445)
OK, for me the obvious comparison point is to Moore. I took the Beckwith estimate above, and compared it to the Moore estimate from his thread. The Moore estimate I used was:

- Taking the WS estimates from Post 7.

- Increasing 1924-1926 by 5% for the reasons detailed in post 21

- Crediting 1917-1919 as Chris did in post 35.


That gives Moore a 10 year record. Comparing his 10 years to Beckwith's 10 best years, I get the following Win Shares comparison

Best year: Moore by 2
2nd best year: Moore by 5
3rd best year: Moore by 5
4th best year: Moore by 1
5th best year: Moore by 2
6th best year: Moore by 1
7th best year: Moore by 1
8th best year: Even
9th best year: Even
10th best year: Beckwith by 1


Beckwith has additional years of 14, 12, 11, 11, and 3 Win Shares (for this exercise, not counting his 1919, where he was credited 0.4 WS).

To me, Beckwith's additional years do not overcome Moore's peak advantage (which admittedly is slight, and well within the margin of error considering the guesswork we're doing here). For a more career oriented voter, they might. As it is, he joins the NL glut at the bottom of my ballot (5 of the 6 players from 9 through 14 are Negro Leaguers), at #12, between Redding and Poles.
   24. Gadfly Posted: November 27, 2004 at 06:23 PM (#983950)
Chris Cobb-

My sources for Beckwith's later career are 1) various newspapers, especially the Pittsburgh Courier and New York Age, 2) various ballplayer interviews, 3) Beckwith's Obituary, and 4) Al Fennar, who was a teammate and friend of Beckwith.

There is an interview with Mr. Fennar in one of Brent Kelley's Negro League interview books and I talked to Mr. Fennar myself about two years before he passed away (a very nice man).

The following is an outline of Beckwith's 1930s playing career (this is off the top of my head, without checking, so there may be some small mistakes like calling the Newark Browns the Stars in my previous entry):

1931: Beckwith, in a fight with John Henry Lloyd to be manager of the Lincoln Giants, loses and joins the Baltimore Black Sox. Later that season, he bcomes the manager and star for the Newark Browns, a new team formed around him.

1932: Newark Browns, a member of Cum Posey's East West League.

1933-34: New York Black Yankees. The Black Yankees are an independent, non-league team run by James Semler for Nat Strong, who controls semi-pro baseball in New York. The Black Yankees are a percentage team, playing and then splitting the proceeds.

1935: Homestead Grays. After Beckwith left Homestead, I am pretty sure that he joined the Brooklyn Royal Giants but have never been able to confirm it.

1936-1938: Manager and star of the Brooklyn Royal Giants. In 1938, Beckwith was the first manager of Lyman Bostock Sr.

1940-1941: Manager and star of a team variously called Beckwith's Stars or New York Stars. Beckwith's obituary says that Beckwith stopped playing baseball in 1942. I assume that he ran his own team from 1939 to 1942.

As for the Crescents of White Plains, New York, this team is the Brooklyn Royal Giants.
   25. Gadfly Posted: November 27, 2004 at 06:37 PM (#983963)
There are several factors that are biasing the evaluation of Beckwith in this thread.

Factor 1 would be that Beckwith played from 1922 to 1923 for the American Giants in one of the worst hitters parks of all time.

Factor 2 would be that Beckwith spent half of 1924 playing for the independent Homestead Grays.

Factor 3 would be that Beckwith played all of 1928 for the independent Homestead Grays (and was credited with hitting 60 home runs in 147 games).

Factor 4 would be that Beckwith played, with few exceptions, for mostly independent teams from 1931 to 1938.

As for Beckwith's eroding hitting skills, there is really only one piece of evidence to support this. In 1935, the Negro National League released batting statisitcs for the first half. Beckwith, in under 100 abs, batted just under .200.

But as I said before, he was in Pittsburgh, arguing with Cum Posey (the Homestead Grays were in terrible financial shape at that time) over money, and evidently unhappy.

From 1931 to 1934, there is really no evidence that Beckwith's batting had slipped. In fact, the limited evidence seems to show that he was still one whale of a hitter.
   26. Gadfly Posted: November 27, 2004 at 06:45 PM (#983971)
As for comparing Beckwith to Dobie Moore, it just doesn't work.

Dobie Moore was a bigger stronger slower version of Frankie Frisch playing shortstop.

John Beckwith was a bigger stronger slower version of Rogers Hornsby playing third base.

Dobie Moore wasn't even the best hitter on his own team (Oscar Johnson, Bullet Rogan).

Beckwith was the greatest Black hitter of his time (Well, him or Oscar Charleston), and would have been slamming out 50 HRs with a near .400 BA at the peak of his career.

Dobie Moore would have been, in the best year of his life, hitting half of that in power.
   27. Gadfly Posted: November 27, 2004 at 07:05 PM (#983988)
One more note-

The evaluations above seem to assume that only time spent with teams in the Negro Leagues should count as Major League service time.

However, Negro League ball was not quite as black and white in this regard as organized baseball's Major and Minor Leagues.

There were always independent teams during the heyday of the Negro Leagues that had Major League quality Black Players on them.

The Brooklyn Royal Giants were one example. Another example would be the St. Louis Giants before World War 2 with Luke Easter, Sam Jethroe, and Herb Bracken.

From 1931 to 1942, John Beckwith wished to continue playing in the New York area. His options were 1) the Black Yankees, 2) the Brooklyn/Newark Eagles, 3) the New York Cubans, and 4) the Brooklyn Royal Giants.

The Black Yankees were ok in 1933 and 1934 since they were a true co-op team. But after that, the team was controlled more and more by James Semler, who was one cheap bastard.

The Eagles were controlled by the Manleys, who while not as cheap as Semler, were also cheap and also overly controlling.

The New Cubans basically only wanted to hire Cubans because, well after all, they would play cheap.

On the other hand, the Brooklyn Royal Giants were a true co-op team that split the gate and did not travel because they weren't in a league.

Throughout the 1930s, the Royal Giants were filled with older Negro League veterans. That Beckwith played for this team doesn't detract from his career one bit.

This didn't change until 1939 when Bill Leuschner, Nat Strong's successor, regained control of the Royal Giants and hired Dick Lundy to manage the team.

Leuschner was also one cheap bastard.
   28. Gadfly Posted: November 27, 2004 at 11:37 PM (#984222)
Jeez, you can't say b*st*rd or shorthand s*n of a
b*tch on this thread? That's a little extreme.

Chris Cobb:
You asked me to weigh in on Bullet Rogan and I went and read the Bullet Rogan thread.

From reading it, it's apparent that no one has read or owns Rogan's wonderful (but very biased) biography by Phil Dixon "Bullet Joe and the Monarchs."

Rogan was born in 1893, not 1889 (he advanced his age to join the army). Rogan played with his hometown Kansas City Giants in 1911, the 24th infantry from 1912 to 1914, and the 25th Infantry Wreckers from 1915 to 1919.

Rogan get out of the Army in 1917 and did play for the All Nations and the Kansas City Giants before being redrafted for the duration of World War One. He finally got his Negro League career going in June of 1920 at the age of 26 (almost 27).

Both Rogan and Beckwith would have been first ballot Hall of Fame players in my estimation if given the chance to play in the Majors.

Rogan was basically a combination of Zack Wheat and Dazzy Vance, a Hall of Fame hitter and Hall of Fame pitcher. I think he would have been in the Majors from 1915 to 1935 or so.

Beckwith was a cross between Rogers Hornsby and Lou Gehrig playing Third base and would have played in the Majors from 1919 to about 1938 or so.

I've got to admit that, if I had to chose between them, I'd pick Rogan. Great as Beckwith was at bat, Rogan was like two Hall of Famers rolled into one and is definitely in the top 10 Negro Leaguers of all time.

But John Beckwith is not far behind.
   29. Mark Shirk (jsch) Posted: November 29, 2004 at 06:30 PM (#986568)
Here is my thinking on Beckwith, tell me if I am really missing something.

Right now I have beckwith behind Dobie Moore. Neither are what you would call 'career' guys, at least as far as I know. As par Michael Bass's post above, it seems that Moore had the higher peak. Also, I am not convinced that Beckwith would have played anything other than 1B in the Major Leagues. There are virtually no records of his fielding beign outstanding, and for a man his size he probably would have been over at the first sack early in his career, even if he could actually have handled 3B. So as a slugging first baseman without the peak of Dobie Moore, he is just off my ballot.

Though this may force me to take a look at Sisler again. I bumped Gorgeous George up a few slots last week, but if he wasn't any better than Beckwith with the bat maytbe they should be closer.
   30. Chris Cobb Posted: November 30, 2004 at 03:39 AM (#987429)
Been meaning to respond this for a while. Busy day . . .

As I posted on the ballot discussion thread, I think treating Beckwith as someone who would have been a first baseman in MLB is a mistake. I argued cautiously there, but I think I now would make a stronger case. Consider:

1) Beckwith may not have had a strong defensive reputation, but he played one third of his career at _shortstop_. He was marginal there, but he did have substantial playing time even farther up the defensive spectrum than third base.

2) Third base was still a more important defensive position than second base at this time, though the switch was coming.

3) The quality of play in the Negro Leagues was about at the level of the high minors. I've had a devil of a time finding positional information about minor league players on (or off) line, but it doesn't seem to me that we see a whole lot of players who were shortstops in AAA being converted to first basemen the minute they get to the majors. Maybe things were different in the 1920s and 1930s, but I think we need some evidence. I tried to track down some minor-league positional info for third-basemen who were rapidly converted to first base (or outfield in the majors) to see if any of them had been shortstops in the minors, but I couldn't find data.

Does anyone have a good way to find out info about players like

Tony Perez
Chipper Jones
Jim Thome
Bobby Bonilla
Harmon Killebrew (well he didn't have a minor-league career, now did he? but before that?)
Bob Horner (not much minor-league play there, either, but what about college?)

Jimmy Foxx played a little third base early in his career, but I don't know that he played in the minors much to speak of, either.

To make Beckwith a first baseman is to equate him with this kind of player. Unless some of these guys were shortstops along the way, fairly close to the majors, I just don't see downgrading Beckwith.
   31. Michael Bass Posted: November 30, 2004 at 03:48 AM (#987443)
Not mentioned, but what about Kevin Mitchell? He played a fair amount of SS for the Mets, but was quickly moved to a permanent corner OF position.

Chipper, of course, was a SS at some point. I do not know for sure when that ended.
   32. DavidFoss Posted: November 30, 2004 at 04:26 AM (#987489)
Harmon Killebrew (well he didn't have a minor-league career, now did he? but before that?)

Killebrew played parts of 56-58 in the minors... after his bonus-baby time was up. All I can find of his numbers are a snapshot of an old baseball card on Ebay here

Recent minor league hitting numbers can be found at the Baseball Cube... the site has some popups, unfortunatedly, but Bonilla's page with minors numbers is

Let me know if anyone finds a good site for minor league numbers.
   33. KJOK Posted: November 30, 2004 at 06:40 AM (#987611)
Jones definitely played SS at least thru AA, and possibly some of AAA.

Killebrew played 2B almost until reaching the majors?

Horner played SS at Arizona St.
   34. Kelly in SD Posted: November 30, 2004 at 08:54 AM (#987750)
Regarding minor league positions for various players: numbers are from Daguerreotypes 1981 and Stats All-Time Handbook.

It looks like Killebrew started out at 2nd and/or 3rd in high school and then during his bonus baby time in the majors. By the time he was 22 he was splitting time between 3rd and OF.
Foxx was signed by Mack as a catcher, but Mickey Cochrane was a rookie that same year, so that didn't last long.

1954 3 games at 2b. 18 yrs old.
1955 3 at 2nd, 23 at 3rd with above avg range and a .935 FPct vs. .950 LgPct.
1956 4 at 2nd, 20 at at 3rd with above avg range and FPct.
Also, this year he played 70 games at Charlotte in the Sally League where he played 3rd.
He led the Sally League in Assists and errors.
1957 1 at 2nd, 7 at 3rd
For Chattanooga of the Southern League, he played 142 games and was listed only at 3rd.
1958 9 at 3rd
with Indianapolis of the American Association, 38 games and listed at 3rd.
with Chattanooga, 3rd and OF in 86 games.

Jimmie Foxx
1924 - C for Easton at age 16
1925 - C for Providence and the Athletics. 1 gmae. Cochrane was a rookie this year also. He caught 134 games for the Athletics.
1926 - C for Athletics. Foxx catches 12, Cochrane 120
1927 - 1st for Athletics
   35. Dr. Chaleeko Posted: November 30, 2004 at 01:45 PM (#987827)
IIRC, Jones's SS career was at least in part derailed by breaking his leg during spring training of the 1994 season. He played 8 games at short in 1993 before missing 94 with the fracture.

Also, Tim Raines and Danny Tartabull came up as second basemen and were quickly moved to corner outfield slots.
   36. andrew siegel Posted: November 30, 2004 at 02:39 PM (#987862)
What about Kevin Mitchell and Gary Sheffield, both of whom came up as potential SS, quickly washed out at that position, played some 3B poorly, and quickly settled in as medicore corner OF's? There is a type here: big strong guys who are good overall athletes and surprisingly quick for their size when very young but who bulk up and stiffen up quickly as they age into their mid-20's. I may be wrong on this, but I picture Beckwith as such a guy.
   37. Dr. Chaleeko Posted: November 30, 2004 at 03:02 PM (#987896)
A couple more switchers....

Ben Chapman came up as a MI/3B and went to the outfield where he played nearly equally in all three OF positions.

Rudy York took the Foxx route of quickly going C to 1B.

Jeff Bagwell was, of course, a minor league 3B very soon before moving to 1B, and I think the same is also true for Mark McGwire. Albert Pujols didn't last long at 3B. I don't know if any of them were ever MIs.

And last but not least, Cory Snyder was a SS on Team USA if I remember my 1985 Topps set correctly. He played a little there his first couple years in MLB as well as a little third, then went to the OF for the rest of his career.

Which of these guys is not like the other???? ; )
   38. andrew siegel Posted: December 01, 2004 at 04:13 PM (#989408)
Baseball is a funny game. Because the failure rate at the plate is so high, there is an immense opportunity for a truly outstanding hitter to rack up huge values over replacement level, or even huge values over star level, with the bat. Because of that fact, if you are a good enough hitter, literally nothing else matters. There are in the history of the game a set of guys who are total ########, drains on their team's energy, awful fielders, etc., who were still great players solely b/c/ of their bat. Fort those guys, as long as they stay out of jail, show up 90% of the time, keep enough friends or admirers to keep getting jobs, and throw no more than a handful of games, they are all-time elite players.

In general, I am a supporter of inducting those guys. Lip Pike (who might fit the bill) in in my personal HoM. Dick Allen (the standardbearer for the group) will be. Garry Sheffield (who belonged among these guys at least in his early days) almost certainly will be as well. Albert Belle would have been if he hadn't lost too many seasons to immaturiy on the frontend and injury on the back end.

Having reviewed the available statistical and anecdotal evidence, I now think that John Beckwith was that kind of hitter. Therefore, I am inclined to write off all the rest of his crap. He'll join my ballot somewhere around 7th next week.
   39. Chris Cobb Posted: December 01, 2004 at 05:55 PM (#989631)
A few thoughts in response to Tom H's questions on the main Negro-League thread re reliability of the data and the difference between Beckwith's numbers and his reputation:

1) The data for Beckwith are very extensive, by Negro-League standards. They give us a picture of Beckwith's value that is probably more reliable than we've had for any Negro-League position player who has so far been a serious candidate, with the possible exception of Torriente because of all the data from his Cuban play. But Beckwith is better documented than Grant, Johnson, Rube Foster, Hill, Monroe, Poles, or even Dobie Moore. There are still holes in the data because of Beckwith's play outside the organized leagues, but those holes don't change the picture of Beckwith's ability.

2) There is no conflict between the data and his reputation, in that his reputation admits what the data shows: Beckwith was a great hitter. His overall ranking in the reputation game is brought down by factors that would not appear in the data even if it were better: surliness, unreliability, indifferent fielding. I don't think that there's anything in his reputaton that suggests that his numbers as a hitter shouldn't be taken at face value.

In sum: Beckwith's reputation and his placement in subjective rankngs of Negro League players does not provide evidence that the numbers are untrustworthy. Rather, it shows that people, despite Beckwith's talents with the bat, ranked him lower. We must decide how much stock to place in claims that Beckwith's temperament negated his batting value.

On this point, Andrew Siegel has said everything that I would say, and better.
   40. Chris Cobb Posted: December 01, 2004 at 06:07 PM (#989652)
A second, more concrete response to TomH:

Having mulled over gadfly's additional biographical information, here's how I am incline to revise Beckwith's numbers.

1) In years during Beckwith's prime when he jumped to the Homestead Grays outside the league, I will project MLE credit for his time with the Grays, if a cross-check of the i9s projections shows that they docked his playing time in those years. I haven't had a chance to do that yet, but I will. The Grays were a major independent team; it is documented that they did well in competition with other black teams and with ML teams during these years. Joe Williams got credit for this play, and so should Beckwith, if he hasn't.

2) In the last years of Beckwith's career with major teams, I will not follow the steep decline pattern projected on i9s; I will project him as declining slowly from his peak, but not at the end of his abilities.

3) For the undocumented years after 1935 when Beckley was with the Bkn Royal Giants and other teams, I will not give MLE credit. Unless some record of competition against the top black teams is found, I don't have enough information to go on to award credit, given that it's possible that he was not good enough to play on a major team anymore.

There's also the Kid Nichols case to consider. Nichols left the majors to pitch for his minor league team by choice. He was still good enough to pitch in the majors, and he proved it by coming back two years later. I don't think anybody gave him credit for those seasons (not that he needed it). Beckwith might have had the ability to play for a top team, but he chose not to do so. We give credit to players who are involuntarily trapped in the minors when they have the ability to be starters in the majors, but not to players who choose not to play at the highest available level of competition. As gadfly notes, circumstances were somewhat different in black baseball than in white baseball, but I think the principle still holds.

I am planning to revisit my MLEs for Beckwith, then, and I expect his value in the revised MLEs will rise in light of gadfly's biographical info. But it's not going to be the huge rise he would get if I projected him as a major-leaguer for another 5 seasons, which I think is what gadfly would argue ought to be done.
   41. John (You Can Call Me Grandma) Murphy Posted: December 01, 2004 at 06:29 PM (#989696)
I don't think anybody gave him credit for those seasons (not that he needed it).

I gave him credit for those years, Chris.
   42. Mark Shirk (jsch) Posted: December 01, 2004 at 09:47 PM (#990128)
I guess my worry is that in the Majors, Beckwith may have been sheparded to either 1B or the OF simply becauseof his size, whereas in teh NEgro Leagues, they would put anyone anywhere so long as they could play because of the 'dearth' of talent.

I am pretty certain that ARod would have been a CF if his debut had come in 1965 or 1975, as men that big just didnt' play SS. Could the same phenomenon have applied to Beckwith? Should this be counted against him,there may be others we are subconsciously downgrading because of this? Am I way off the beaten path here?
   43. John (You Can Call Me Grandma) Murphy Posted: December 01, 2004 at 10:25 PM (#990161)
If Beckwith had played 1B or the OF, his offensive production would have improved since he wouldn't have taken the same beating at those positions. He would have had a longer extended peak, too.
   44. Gary A Posted: December 08, 2004 at 02:47 PM (#1005917)
From Patrick Rock's NNL yearbook:

John Beckwith 1923
Chicago American Giants

HR-8 (led team; team as a whole hit 24)
RBI-77 (led team)

Beckwith's primary position in '23 was first base; he also played at third. His home park, Schorling's Park (formerly West Side Park, the home field of the "Hitless Wonder" White Sox), was the most extreme pitcher's park in Negro League history.
   45. Mark Shirk (jsch) Posted: December 08, 2004 at 02:52 PM (#1005929)
That is possible John, but you still would have to admit that a player with Beckwith's abilities is more valuable at 3B than 1B.
   46. John (You Can Call Me Grandma) Murphy Posted: December 08, 2004 at 05:09 PM (#1006203)
That is possible John, but you still would have to admit that a player with Beckwith's abilities is more valuable at 3B than 1B.

No disagreement there, jschmeagol. It's always better to move a player farther right on the defensive spectrum if you can.
   47. Gary A Posted: December 16, 2004 at 05:09 AM (#1023842)
Some incomplete data for Beckwith, 1934:


I have him in one early May game for the Newark Dodgers, the rest of the year for the New York Black Yankees. He played third base in four games, and appeared as just a pinch-hitter in five.
   48. Gary A Posted: December 16, 2004 at 05:13 AM (#1023847)
John Beckwith 1921
Chicago Giants NNL

G-34 (team 40)
AVE-.396 (NeL .263)
OBA-.434 (NeL .324)
SLG-.597 (NeL .361)

Joe Green's Chicago Giants were a travelling team, so these stats are achieved in basically a neutral context. Fielding stats to come in a few days.
   49. Chris Cobb Posted: December 26, 2004 at 06:42 PM (#1041339)
Here are revised win-share MLEs for John Beckwith. Based on information from Gary A. and gadfly, I have changed my projections for four seasons -- 1921, 1922, 1924, and 1928. An explanation of each of the changes follows.

John Beckwith's Win Shares

Year (games) BWS + FWS = Total
1919 (18) 0.0 + 0.4 = 0.4
1920 (127)7.8 + 4.3 = 12.1
1921 (149) 22.7 + 5.0 = 27.7 [was 23.1]
1922 (131) 15.0 + 4.1 = 19.1 [was 14.3]
1923 (144) 23.7 + 4.3 = 28.0
1924 (146) 25.2 + 4.7 = 29.9 [was 22.3]
1925 (140) 29.1 + 4.4 = 33.5
1926 (117) 12.7 + 3.2 = 15.9
1927 (151) 20.1 + 4.1 = 24.2
1928 (135) 16.7 + 4.0 = 20.7 [was 17.5]
1929 (146) 24.9 + 4.0 = 28.9
1930 (100) 18.3 + 2.6 = 20.9
1931 (149) 24.9 + 3.0 = 27.9
1932 (100) 8.6 + 1.9 = 10.5
1933 (95) 13.9 + 1.3 = 15.2
1934 (54) 2.0 + 0.8 = 2.8
16 (1902) 265.6 + 52.1 = 317.7

Quick Explanation of changes:

1921 -- Data posted in 48 above show Beckwith's production as better than the more limited data in Holway. More importantly, data shows that offensive levels in NeL at this time were lower than in the majors, so I have adjusted for that difference.

1922 -- Data from 1923 show that Chi Am Giants' home park was an _extreme_ pitcher's park, so I have revised estimates to account for larger park factor, and also for lower offensive levels in NeL. We have league-wide offensive-level data for 1921 and 1928. In 1921, the NeL's offense was about 10% lower. In 1928, it was about 2% lower. I hope we'll have data of this sort for more seasons soon, but in its absence, I am estimating that the NeL gradually approached the majors in offensive level, reaching parity in 1929, when the NeL experienced a big surge in offense. (for 1923, there is no change, incidentally, because the lower numbers provided by Gary A are offset by the league offense adjustment and the park adjustment.)

1924 -- gadfly's information shows that Beckwith played a full season this year, but part of it was not accounted for by the i9s projections because it was with the independent Homestead Grays. I've raised Beckwith's games played from 109 to 146.

1928 -- A small change this year taking account of the 2% difference in league offense, regressing Beckwith's stats (for only 19 games) towards his career mean, and boosting his games played estimate from 128 to 135. This difference may look larger than it is because of a typo in the original presentation. For 1928 in post 14 above, you'll see it reads 13.5 + 4.0 = 10.5 The batting win shares and fielding win shares are correct (for that estimate model) so the total should have been 17.5. This error did not affect Beckwith's career total.
   50. John (You Can Call Me Grandma) Murphy Posted: December 26, 2004 at 07:13 PM (#1041347)
Thanks, Chris! I would have him at #1 on my ballot regardless, but maybe Beckwith will start to make his "inexorable" climb to the top now. :-)
   51. karlmagnus Posted: December 27, 2004 at 07:08 PM (#1042379)
Nowhere near the top of my ballot; if he can only get to 317WS, after he's been given the benefit of every possible doubt, including the IMHO artificial WS benefit of being a shortstop, he's not ballot-worthy.

Chris's conversions are enormously useful, and help greatly in ranking the Negro league players among each other, but 317 converted Negro League WS is really about 260-270 after appropriate discounting for NL over-enthusiasm. Anyway, wasn't Ben Taylor higher?
   52. karlmagnus Posted: December 27, 2004 at 07:11 PM (#1042387)
Yes, Taylor was higher -- 325.6, which converts to about 280. Still off-ballot, but close to it.
   53. John (You Can Call Me Grandma) Murphy Posted: December 27, 2004 at 07:15 PM (#1042392)
but 317 converted Negro League WS is really about 260-270 after appropriate discounting for NL over-enthusiasm.

No, 317 converted Negro League WS is really 317 major league WS. I know you don't by the conversions yourself, but 317 WS from a shortsop with his peak is at least ballot worthy, IMO.

As for Taylor, he was a pitcher at the beginning of his career, so he received many more WS for those seasons than if he had at another position (overrating him somewhat).
   54. karlmagnus Posted: December 27, 2004 at 08:33 PM (#1042531)
Sorry, the damn machine ate my post explaining my methodology. Try again.

If we accept the conversions as written, we will end up with 25-30 Negro league HOMers. I don't buy it; on the basis of the population there should be about 12. There are thus 2 possibilities: (i) African Americans, being at the bottom of the economic totem pole, were more likely to choose a baseball career, or (ii) there is a considerable element of doubt in the conversions, and doubts are being resolved in favor of the player, by about 15%.

I don't buy (i) by the 10s-20s-30s. MLB was clearly attractive by then to Columbia graduates such as Lou Gehrig; the money was good and the career was stable. This is in distinction to the situation in the 1890s, when someone like Leever with a middle class alternative would reasonably choose it, because of MLB's instability and relatively unexciting wage rates. If baseball is attractive to 95-98% of the general population, it can't be significantly more attractive to African-Americans, particularly as the Negro leagues were less stable and less lucrative than MLB. (to be continued)
   55. karlmagnus Posted: December 27, 2004 at 08:45 PM (#1042562)
(continuation) (ii) People who spend the large amount of time, effort and skill to get up the curve on the Negro Leagues are obvously attracted to them and sympathetic to them; it's human nature. In an ideal world, as well as Chris' conversions, I would have available those of an intelligent baseball-savvy moderate racist, say Ty Cobb. Since Ty hasn't left me his figures, I have to take Chris's and discount them by some factor that makes the Negro leaguers fit within the overall system, while preserving Chris's very valuable ranking within the NL. Hence Beckwith's 315 becomes 15% less, or 268, and Taylor's 325.6 becomes 277. I believe this most accurately places the NL players among their ML brthren.

I then have to worry about WS overrating shortstops, which it clearly does, and undervaluing pitchers and catchers, who tend to have shorter careers. This leads me to put Taylor just off my ballot and Beckwith well off. It doesn't alter the fact that Charleston, even after appropriate discounting, is probably #1 in 1943 (Beckley and he appear to be close, but I want to see Chris' conversion, which I would expect to come to 400+WS, in which case Charleston wins). Williams, too, was elect-me (but #2 to Alexander) in 1936. The best are still the best, but the next level need to be looked at skeptically.

The HOF has 17 NL members, at the absolute top limit of what it should have, well more than the African-American share of the population. Since we, who have looked more closely at the early African American players, have elected some that the HOF haven't, we should also leave some out that the HOF have elected. The ones we leave out will be players from the 1930s NL, which the HOF has over-represented.
   56. Michael Bass Posted: December 27, 2004 at 09:43 PM (#1042686)
Or iii, your assumptions about population proportions are ludicrous, as shown by essentially every major sport today, none of which conform to any population proportion.

There's no point arguing against you on this point, because you've obviously made up your mind against all evidence to the contrary (see: Leever, Sam). But despite this, I will simply point this out: As far as I know, there were no serious avenues for the black athlete other than baseball. White athletes had access to various other sports. This is just one of many reasons why using population data, of all things, to set a ceiling on representation is silly beyond words.
   57. John (You Can Call Me Grandma) Murphy Posted: December 27, 2004 at 09:47 PM (#1042689)
If we accept the conversions as written, we will end up with 25-30 Negro league HOMers. I don't buy it; on the basis of the population there should be about 12.

karlmagnus, I count 18 HOF African-American or dark-skinned Latino players that played a significant amount of games during the fifties and sixties (I'm not even including guys like Reggie Jackson or Rod Carew). How do you account for that? I'm not trying to be a wiseguy; I really am curious.
   58. sunnyday2 Posted: December 27, 2004 at 09:57 PM (#1042708)
There have been many discussions over the "years" of the significance of the absolute size of any population pool. The bottom line (IMO) is that any player getting any consideration here is an outlier, an anamoly. There is no reason to believe that outliers would be a representative sample of the whole. That is why I don't believe there should be twice (or whatever the number is) as many HoMers from the 1960s as the 1890s, which one might argue from the size of the American population as a whole (or whatever the number is).

I also don't believe for a nano-second that the portion of America's total male population consisting of African-Americans is much of a guide for how many NeLers should be in the HoM. I mean, what percentage of the basketball HoF should be black? 10-12-15%? And you can't argue that it should be the same as the percentage of NBA players who are black, that has nothing to do with the total pool and is nowhere near analogous to the question before us. It seems to me that it is a devastating example to the idea that the HoM should be no more than 10-12% black for the period 1920-1946.

Similarly, how do you explain that the one single obviously greatest American track athlete of the Golden Age was also black--Jesse Owens. Or that the one single obviously greatest boxer of the era was black--Joe Louis. And I can assure you that Charles "Tarzan" Cooper was one of the two greatest basketball players in America in the 1930s(John Wooden wasn't bad). The very first national professional championship tournament sponsored by the Chicago Tribune in 1939 was won by Cooper and the New York Rens, and the second such tournament the following year was won by the Harlem Globetrotters.

So, having disposed of that question...what about MLEs for NeLers? Well, I happen to believe that Chris' conversions are a bit generous. I generally discount his conversions another 10-15%. But on the other hand, I do not believe that he is (I know that he is not) merely extrapolating NeL stats out to 162 games as if they should be taken at 100% value. He has made it clear what his conversion (discount) rates are. They are reasonable, I just choose to discount a little more deeply. Of course, as a peak/prime voter the additional discount is not usually fatal to a highly qualified black player whereas for a career voter another 15% discount might indeed be fatal for most such players.

I do not have Pete Hill in my PHoM but still have Dobie Moore, Beckwith and Bill Monroe on my active consideration set of players who could become PHoMers some day. As for players who ARE in the HoF, I am at the present time inclined to question Judy Johnson and Ray Dandridge's qualifications. I think they have benefited way too much from being "the best 3B," as if the HoF or HoM should have to elect "the (two) best ML 3B" of 1920-1946, whether that would be Traynor and Hack or whomever, or that we should have to have a 1B between ABC and FGG.

In sum, reasonable people can disagree about this. But when you (karl) say that 17 is "the absolute top limit of what it should have," you really ought to add "IMO." Not having gotten through the process yet, right now I am guessing that the HoF has 3-5 of the wrong NeLers in, and 5-7 deserving NeLers out, for a net gain of 2-3. Considering the achievements of black athletes wherever and whenever they had the chance, as far back as Jack Johnson, an a priori that the HoF hit the highest possible justifiable number--that's a hypothesis, not a fact.
   59. karlmagnus Posted: December 27, 2004 at 10:00 PM (#1042711)
For about 20 years after 1947, MLB was the one area of American life that really was integrated -- in other words it had a special attraction to African American players compared to other possible careers. Hence it got more than its fair share of African American players. After 1970 or so, as other areas of American life opened up that was no longer true, and so the percentage of African American players in MLB declined somewhat.
   60. karlmagnus Posted: December 27, 2004 at 10:05 PM (#1042719)
(58) You're right "at the absolute top limit of" should read "well towards the upper limit of, on stochastic considerations." I don't mean to imply there should be a firm upper bound.
   61. Mark Shirk (jsch) Posted: December 27, 2004 at 11:22 PM (#1042810)
I thought the number or percentage of black players peaked in the late 79's?

And not to be a prick to sunnyday2, but I would argue that we should have nearly twice as many players from the 1960's in the HOM as from the 1890's. But taht is because we are talking about nearly twice as many teams.

Also, maybe it is just my Beckley hating ways, but how could he even be in contention with Charleston for the first spot of your 1943 ballot? I would put Oscar ahead of Hornsby who was everyone's number two.
   62. sunnyday2 Posted: December 28, 2004 at 12:36 AM (#1042863)
Two points in reply to gollum ;-) (no offense intended, I just can't spell schmeagol [or did I?])...

First a minor (i.e. micro but important) point: I've read references to the one-league 1890s enough to wonder if everybody knows that from 1892 to 1899 it was a 12-team league? I mean was the level of play in the one-league '90s that much better than in the two-league years just before and after. Yes there were fewer teams, but 4 not 8.

More specific to #61, I think you are confusing the number of teams with the player pool. MLB went from 24 teams in '90 to 16 in '91 to 12 in '92 and to 8 in '99 and '00, all from a player pool the size of which did not change hardly at all. Do we want twice as many HoMers in '90 as in '92, and twice as many in '01 as '00? I mean, if you want twice as many HoMers in the 1960s based on the number of teams alone, then the number of teams is necessary and sufficient to dictate your distribution and then why not the same outcomes in '90 vs. '91 vs. '92, etc? second, more macro, point: Because the argument for more HoMers in recent years (which I don't support) has always been based on the size of the player pool, not the number of teams.

That is not to say you couldn't make that argument--but if it is the number of teams, again, that is decisive, then 1) you're stuck with that methodology in '90-'91-'92 and 2) then you are talking about raw value. More teams play more games and earn more WS, and more WS means more value which means more HoMers. Now I suppose you can make this argument AND employ a timeline, but a pure value argument based on the number of teams makes 30 adjWS in 1885 just as valuable as 30 WS in 1985.

But most people who argue for more HoMers from recent years are timeliners who argue from skill and ability, not from value, and the fact is that the number of teams has nothing to do with skill and ability. Now I would also argue that the size of the player pool is vastly overrated as a determinant of skill and ability, though it is quite true that a larger pool is likely to have more skill and ability. But MLB players are outliers on the bell curve of skill and ability and I have always argued that outliers are more randomly distributed and not strictly subject to the influence of the size of the pool.

But back to my main point. It is illogical to me to argue for more HoMers from the 1960s vs. the 1890s on the basis of the number of teams, and only in part because you only have 1.33X more teams (not 2X), but especially because anybody who wants 2X more HoMers from the 1960s is clearly timelining which pretty much negates any argument strictly from value.

I hope that made sense. It did to me.
   63. DavidFoss Posted: December 28, 2004 at 12:38 AM (#1042866)
You know, I never really give much thought to quotas. I just try to rank the best eligible... when a player gets inducted its deleted from my list. Its out of sight, out of mind at that point. (That's what impresses me about guys who keep PHOM's. Its quite a bit more work to keep tracking an inductee... especially after their names disappear from our "yearly" debates.)

So far, with all the backlog years we've had in the past 15 years or so, I certainly can't have any qualms with the NeL-ers who've been inducted so far. I mean, it was a bit surprising that Pete Hill flew in so easily, but he had several backlog years following his induction to get elected. After all, the same could be said for Red Faber... surprising that he flew in with such little support, but then again there were two more backlog elections in the next three years waiting for him if he didn't make it.

The fun will be in the next ten to fifteen years or so... when ballots are strong enough that the 3rd place candidate appears on almost all the ballots... not these backlog elections where making it to 35 of 50 ballots almost insures induction. I plan on taking it week by week.
   64. Mark Shirk (jsch) Posted: December 28, 2004 at 02:52 AM (#1043001)

1. Oddly enough my screen name isn't really taken from Lord of the Rings. It was just a named I used to forge doctor's slips in High School. I think the spelling is from there though, I dont' know.

2. I was being sort of facetious, but werent' there at least 18 teams in the 1960's? In 1961 the Angels came into being, the Mets in '62, and didn't the Senators start somewhere in there?
   65. jimd Posted: December 28, 2004 at 03:30 AM (#1043033)
1961: The AL expands. The Senators (still owned by the Griffith family) are allowed to move to Minnesota, and one of the expansion teams is a replacement team (politically expedient). The other team is in Los Angeles.

1962: The NL expands, adding teams in Houston, and a replacement team in New York.

1969: Both leagues expand again: the NL adds teams in Montreal and San Diego, the AL adds teams in Seattle and Kansas City (a replacement team for the A's which had moved to Oakland). The Seattle team lasts one year before being purchased by Bud Selig (or a group including him) and moved to Milwaukee.
   66. Chris Cobb Posted: December 28, 2004 at 04:01 AM (#1043077)
I don't have a whole lot more to say about the conversions I make. I've said that the base translation rate is .87, which puts the overall level of competition in the Negro Leagues between AA and AAA, I think, though I am not sure of Davenport's exact current translations.

I based that rate on a study of the batting averages of players with significant playing time in both the Negro Leagues and the Major Leagues. Anybody can go find the same players I used and duplicate the study if they have a mind to.

Because Karlmagnus's interpretation of my numbers includes some consideration of my interests in creating them, I will comment briefly on them. Karlmagnus's less-than-sensible idea that my numbers ought to be compared to those of an intelligent racist (supposing such an oxymoron could be found at the present time) seems to suggest that I have some particularly favorable interest in the Negro league players.

I don't. I am not a Negro-League historian. I had never heard of Frank Grant, Home Run Johnson, Pete Hill, or John Beckwith before this project started. I started working on the conversions because I wanted to find a way to rank the Negro-Leaguers fairly, and I found it an interesting analytical problem. I have learned a lot about the Negro Leagues that has increased my admiration for these players and my sense of the difficulties they overcame, but I don't see how it would do anything to honor them to attempt to represent them as having achieved more than they did as ballplayers.

My win-share projections may be too high, and they may be too low. I welcome others'analyses of them. Application of the win-shares' short form to pitching data, for example (can't remember who did that off the top of my head, sorry), led me to rethink my work on pitchers. I remind everyone that KJOK's translations for Negro-League hitters are available on the yahoogroups site. I think our ranking of the Negro-Leaguers would benefit from more statistical analysis.

However, I think that Karlmagnus's argument that my projections ought to be adjusted downwards by 15% because I am friendly to Negro-League candidates and therefore give "the benefit of the doubt" to the players in every case has nothing to do with statistical analysis, and it's not at all supported by the statistical record.

Applying a .85 adjustment on top of a .87 adjustment would indicate that the level of competition in the Negro Leagues was worth about .74 of major-league competition. A .400 hitter in the Negro Leagues would be about a .300 hitter in the majors by this conversion ratio. If you examine the records of players who played in both leagues, it will be immediately obvious that the difference in quality of competition was not nearly this large. It's not a small error; it's a large error.
   67. Chris Cobb Posted: December 28, 2004 at 04:07 AM (#1043086)
A few remarks on Beckwith:

1) Deciding on his fielding value is difficult.

2) He was a great hitter. Among Negro-League hitters to 1940, I would take Josh Gibson, Oscar Charleston, Turkey Stearnes, and Mule Suttles ahead of Beckwith, but that's it.

3) He's at #3 on my preliminary ballot at the moment.

One remark on Oscar Charleston: I've started work on him. I don't have firm numbers yet, but if he earns less than 500 career win shares, I will be somewhat surprised.
   68. jimd Posted: December 28, 2004 at 07:30 PM (#1043765)
This was posted on the League Quality page, though it may also be relevant to this discussion.

The Davenport adjustments are not simply a percentage. They are more complex, having an adjustment involving the league replacement levels as well.

The following table shows 3 Federal League CF'ers from 1915:
 W-1 W-2 Delt  AdjGm Name
13.2 9.4 (3.8) 136.3 Kauff
 7.0 3.4 (3.6) 145.9 Roush
 4.4 1.0 (3.4) 154.3 Oakes

As you can see, the amount of value lost going from WARP-1 to WARP-2 is fairly constant (though not completely). Kauff loses more absolute value, showing that there is also a percentage involved, but Oakes loses almost all of his value, presumably based on the notion that he was very close to AL/NL replacement level.

So some Federal League value is removed purely because it has no Major League value, because it is sub-replacement value. The residue from this adjustment is apparently then modified by applying a percentage.


Presumably, the Negro League replacement level was lower than the Major League replacement level. So a certain amount of Negro League value each season (proportional to playing time) has no Major League value (or 8-9 Win Shares, assuming full-time play), and the surplus above that level would then be discounted by some amount.
   69. karlmagnus Posted: December 28, 2004 at 07:45 PM (#1043787)
jimd, that has to make sense. What it means is that, as well as multiplying by 0.87 to convert NL Win Shares into ML, you also have to subtract a certain modest number per annum, perhaps 3-5. That would get you closer to my rule of thumb of discounting an additional 15%; relative to my rule of thumb it would remove WS from very long career players, the Negro league Beckleys, while penalizing less the NL Jenningses. The distribution changes a bit but the overall effect, of claiming that 15-18 rather than 25-30 NL players were HOM-worthy, remains the same.
   70. Michael Bass Posted: December 28, 2004 at 07:54 PM (#1043800), karl.

As jimd's example shows, the vast majority of the discount of the FL comes from the initial subtraction. The subsequent percentage discount is very, very small in comparison to this subtraction. There's no other way to explain the relatively similar deltas of Kauff and Oates.

You certainly don't take out a chunk in subtraction then take out a large percentage via discount.

A simple regression analysis of Jim's 3 player example:

Adjusted WARP = .957 * Raw WARP - 3.25

As you can see, after the subtraction takes place, the actual percentage adjustment is only 4.3.
   71. John (You Can Call Me Grandma) Murphy Posted: December 28, 2004 at 07:59 PM (#1043803)
The distribution changes a bit but the overall effect, of claiming that 15-18 rather than 25-30 NL players were HOM-worthy, remains the same.

Why are you reducing the equivilents even more? IOW, if Chris hadn't downgraded Beckwith's numbers because the Negro Leagues weren't as strong as the majors, he would have many more WS than Chris' 317 number for him (350?, 400?). Obviously, that would be wrong, but why should Beckwith or any other Negro Leaguer be hit with a double-penalty.

If I'm missing something here, please let me know.
   72. Michael Bass Posted: December 28, 2004 at 08:02 PM (#1043808)
You are correct, John.

jimd's analysis, if one accepts it, is going to be far more forgiving to star players in an inferior environment than a straight value% discount.

As I understand it (And I may be wrong), Chris is currently doing a straight % discount. If he were to adopt this system, Beckwith, and all the "star" NLers we are considering, would likely see their Win Shares go *up*, not down, at least those with high peak numbers.
   73. Michael Bass Posted: December 28, 2004 at 08:08 PM (#1043812)
Followup post on my own post....

That last paragraph may be an overstatement, given how Chris caluclated his discount. It may well be the case that a straight % discount is more appropriate for the star players (and honestly, we're not all that concerned with the non-star NL players).

Still, it is absolutely not the case that jimd's post shows that we need to be docking NLers doubly.
   74. jimd Posted: December 28, 2004 at 08:48 PM (#1043881)
Thought experiment: Let's assume that John McGraw was able to successfully integrate the majors around 1905, or, even better, that segregation never happened. Let's also assume (more controversially) that the percentage of black players mirrors their proportion in the general population as modified by geography. This is a null hypothesis, and yields an MLB that is about 15% black in the 1920's/1930's. What might this look like?

Our expectation is that the bottom 2-3 players at each position would be replaced by the top 2-3 Negro League players at that position. There could be a large variance in this; it's statistically reasonable that some position might have no black players, and another might have 7 (within 3 standard deviations), depending on the relative talent depth of MLB and the NeLs. Based on a typical team having 14 regulars (7 position players + 2 catchers + 5 pitchers), that's 224 regulars for 16 teams and 20-50 black regulars (again 3 standard deviations).

I have no idea where to set the replacement level for Negro League baseball relative to MLB. I have no idea whether the null hypothesis outlined above has any real merit. But it's a starting point.
   75. sunnyday2 Posted: December 29, 2004 at 03:54 PM (#1044806)
I understand that Chris has already adjusted NeL performance both to ML season length but also to MLE (i.e. applying a competition discount). But I have to agree with karl here. I don't think .87 is enough of a discount.

Chris points out that using his .87 and an additional .85 = .74. Whether this is too much or not enough or just right is a matter of opinion, I think.

And I also want to point out that this is not really a double discount. It merely uses the expedient of adopting Chris' MLEs because they are the most complete and most thoughtful numbers that we have, and because we understand how Chris derived them. Because we understand them, we can then decide whether he has been too generous, too difficult, etc., and apply our own adjustments, as Chris has invited all of us to do. I mention this because I think there was some confusion over whether Chris has already applied a competition discount. Yes, he has, but not enough, IMO.

Chris gave the e.g. of a .400 hitter who drops to .300 with a .87 plus .85 discount. I'm not entirely convinced this is unreasonable, but more to the point: My additional .85 is used in the context of a peak/prime ballot where we are often dealing with .450-.475 and similar seasonal BAs. In this case .87 gets you to .390 to .415 which for most NeLers is not a realistic MLE. The additional .85 gets you to .333-.350, which is what I am more comfortable allowing for Beckwith, Dobie Moore and other great NeL hitters.

Part of the problem here is not really competition discount, however, it is in fact small samples. If Dobie Moore hit .465 one year, it was only in about (what?) 50 games? So clearly his MLE *without a competition discount* would have to be adjusted toward the mean anyway. So say that if he played 150 NeL games his average would probably come down to .420 or less anyway. Then apply .87 and you've got .365. That is plausible. So I guess what I'm saying is I use the additional .85 as shorthand for that.

Either way, in Moore and Beckwith's cases you've got a (at least sometime) middle IFer hitting .333-.365 at their peak with some power. This is the profile that gets these two guys on my ballot. Given Moore's defensive advantage and Beckwith's longer career, they come out about even on my ballot but Moore clearly ahead if you account for his army years. I will say again that Moore is more Joe Cronin or Ernie Banks than Joe Sewell. Beckwith OTOH has no obvious comp among MLers. Maybe Frankie Frisch is as close as it gets though Frisch was better than Beckwith by a wider margin than Cronin was better (if at all) than Moore.
   76. Chris Cobb Posted: December 29, 2004 at 09:15 PM (#1045548)
Chris points out that using his .87 and an additional .85 = .74. Whether this is too much or not enough or just right is a matter of opinion, I think.

It's a matter of opinion within the range of about .82 to .92. Conversion factors above or below that are simply not supported by the statistical evidence. I'm traveling for the holidays, so I don't have all the data with me, but when I return I'll post it.

Part of the problem here is not really competition discount, however, it is in fact small samples. If Dobie Moore hit .465 one year, it was only in about (what?) 50 games?

If you are discounting for short seasons, you are still double-discounting. My conversions take account of short seasons, though much of the data we are dealing with for current candidates is drawn from 80-100 game seasons, not 50-game seasons. In seasons with extreme values(high and low) based on less than 80 games, I regress the player's performance towards their career means. If I were good with statistics and using spreadsheets for all of this, I'd have a formula that I could give you, but I don't. If I didn't use such regressions, players like Beckwith and Moore would have 40-50 win shares seasons. _With_ regression to the mean, Oscar Charleston will probably still have a 40-win-share season or two.

It may be the case that I'm not regressing enough. Anyone with a good knowledge of regression to the mean who could explain to me in layman's terms how to handle properly regression to the mean would be doing me a great service!
   77. sunnyday2 Posted: December 29, 2004 at 09:26 PM (#1045560)
Chris, well, I didn't understand that you were already adjusting toward the mean for the short seasons. I look forward to more discussion as to how/why .82-.92 is indicated. I'm sure you've posted it before but what can I say. So much to read, so little time.

OTOH, even with a "double" discount--my point above being that it is NOT really a "double" discount, but rather merely a somewhat deeper competition discount--Moore, Beckwith and Monroe are all on or very near to my ballot. Also Pete Hill is not in my PHoM but otherwise my PHoM matches HoM as far as NeLers are concerned. My method is calibrated to some degree to function in a peak/prime emphasis, not just (or not particularly at all) for a career emphasis.
   78. Chris Cobb Posted: December 30, 2004 at 02:15 AM (#1045958)
Chris, well, I didn't understand that you were already adjusting toward the mean for the short seasons.

I probably didn't state clearly that I was. When I was working more off of i9s, I figured that they had taken care of that. Beckwith and Moore are the first two position players whom I've worked up from the raw data, and so I've done things a bit differently in consequence. If I hadn't been adjusting, your further discount, especially when measuring peak, would have been a completely reasonable step to take.

I look forward to more discussion as to how/why .82-.92 is indicated. I'm sure you've posted it before but what can I say. So much to read, so little time.

I'm not sure that I've ever given a full write-up of this: I've probably just stated the basic methodology. There's so much to study and work on with the Negro Leaguers that I tend not to write up full accounts of this or that until it's needed. Clearly, it is now time for the league adjustment to receive careful scrutiny!
   79. Chris Cobb Posted: January 03, 2005 at 05:33 PM (#1053124)
Bringing this over from the Arlett thread:

Karlmagnus wrote:

Having said that, I've just started on '43 and put Oscar #1 -- there were truly great NL players, just nothing like as many of them as people are claiming. Beckwith in particular looks WAY overvalued.

Oscar Charleston was certainly was ahead of Beckwith in career value, which certainly matters, and may be the only thing that matters to Karlmagnus. But otherwise it's only possible to say that Beckwith is overvalued and Charleston is "truly great" by ignoring the statistics altogether. Beckwith and Charleston were stars at the same time in the same leagues from 1921 to 1933 (both jumped from West to East in 1924).

Here, season by season, are their batting average, home runs, and home runs/550 ab, when the latter stats are available.

Year -- Charleston -- Beckwith (notes)
1921 -- .437/17/39 -- .396/4/16
1922 -- .391/20/50 -- .303 (B. in extreme pitchers' park)
1923 -- .314 -- .304/8/16 (B. in extreme pitchers' park)
1924 -- .342/9/16 -- .382/5/16
1925 -- .416/19/39 -- .406/24/50
1926 -- .281/8/19 -- .311 (both playing for Harrisburg)
1927 -- .342/12/24 -- .362/9/16 (both playing for Harrisburg)
1928 -- .380/10/26 -- .343/2/16
1929 -- .370 -- .443/15/25 (B. in extreme hitter's park 1/2 of season)
1930 -- .337/6/42 -- .493/6/47 (B.'s stats from 19 games in extreme hitter's park)
1931 -- .341/6 -- .350/16/53
1932 -- .313/10/19 -- no data
1933 -- .388/22/35 -- ..391

Judging by these stats, Charleston was clearly the better hitter in 1921, 1922, and 1928. Beckwith was clearly the better hitter in 1923, 1929, and 1931. I rate 1924, 1925, 1926, 1927, 1930, 1932, and 1933 as too close to call given the limited data set for each season. So over this 13 year period Charleston and Beckwith were at least similar as hitters. Charleston was the better player, since he had much more speed and was a great defensive outfielder, while Beckwith was a mediocre-to-poor left-side infielder. But the claim that one is “truly great” and the other is “WAY overvalued” is simply not supported by the statistics.
   80. karlmagnus Posted: January 03, 2005 at 06:08 PM (#1053273)
You see, even with those stats, I would disgree with you about the conclusions. I don't think they show Beckwith as ever clearly a better hitter than Charleston. In 1923 and 1931 there's less than 10 points in it, and in 1929 Beckwith spent half the season in an extreme hitters' park (depends how extreme, I'll bet there were some really funny ones. I also don't have the slightest confidence in park adjustment factors for the NL, except where it shared an ML park -- it's like the park adjustment factor that puts Clarkson ahead of Welch in 1885 -- we simply don't KNOW what those parks were really like, and don't have enough data to assess them statistically.)

Beckwith has a case to have been better in '26 and '27, when they were both in the same park, also maybe in '24. But the sample sizes are really too small to tell.

Try hard enough, and you may convince me not to enshrine Charleston (don't think so, though, on reputational grounds -- James has him #5 all time, and he can't have messed it up THAT badly.) I am not close to being convinced to enshrine Beckwith, or even put him on my ballot.
   81. Chris Cobb Posted: January 03, 2005 at 06:43 PM (#1053376)

All I'm trying to show is that the two were similar hitters, both among the top 4-5 Negro-League hitters during this period (the other two definites being Turkey Stearnes and Mule Suttles). Charleston was probably a little bit better over this stretch of time, but there's not that much difference.

The season-by-season samples are very small in some cases (you can use the ratio of home runs to hr/550 ab to find the size of each season if you want) but the career totals are of respectable size.

Beckwith 2176 ab, 767 hits, .352 avg. 80 hr, 20 hr/550 ab
Charleston 4972 ab, 1689 hits, .340 avg., 169 hr, 22 hr/550 ab

Charleston has a large edge in career (25 seasons to 15), but during 13-year heart of Beckwith's career, he and Charleston were very similar as hitters.

Try hard enough, and you may convince me not to enshrine Charleston (don't think so, though, on reputational grounds -- James has him #5 all time, and he can't have messed it up THAT badly.) I am not close to being convinced to enshrine Beckwith, or even put him on my ballot.

What I'm trying to convince the electorate of (I don't worry too much about convincing you, karlmagnus, in particular) is that Beckwith's value as a hitter is similar, over a thirteen-year period, to Oscar Charleston's, and that both were among the top 4-5 hitters in the Negro Leagues during that period. That doesn't mean Beckwith was nearly as good as Charleston overall. Differences in career length, defensive value, and perhaps in intangibles also bear consideration. But I hope that most of the electorate will agree that similar achievements ought to receive similar value, so that their seeing that for thirteen seasons Beckwith was comparable to Charleston as a hitter will help them to credit Beckwith's achievements appropriately.
   82. Gary A Posted: January 03, 2005 at 09:03 PM (#1053909)
Just to muddy the waters a bit, the Columbus Dispatch (5/1/1921), in its account of a game between the Buckeyes and Chicago Giants, wrote that Beckwith "showed Columbus fans some dazzling fielding around the shortstop's position."

This is purely anecdotal, but I can say, having read a considerable amount of Negro League newspaper coverage in the 1920s and 30s, that I don't recall anybody saying anything bad about Beckwith's fielding (or his character). To be sure, I haven't made a point of reading up on his Baltimore or Harrisburg days, so this is hardly the final word.
   83. Gary A Posted: January 04, 2005 at 02:01 AM (#1054711)
Btw, I'll work up some 1920 and 1922 NNL park factors in the next few days, and will post 1920-23 with additional info (# of home vs. road games, in particular) that'll help make sense of them.
   84. Chris Cobb Posted: January 04, 2005 at 02:55 AM (#1054816)
Here's a summary of the statistical basis for the .87 conversion factor I use, based on the batting averages of players who played in both the Negro Leagues and the majors. I posted it here because that's where the discussion is, but if it should also go to some more general thread, I'd be happy to re-post it there. Just let me know where you'd like it to be!

NeL to ML batting average conversion summary

There are 14 players with full (or almost full) seasons played in both the Negro Leagues to 1950 and in the major leagues post-1946. Of those 14, six either appeared in the Negro Leagues only prior to age 22 or reached the majors well after the age of 30. Because of issues of rise and decline in players’ careers, I exclude these players from the calculation of the batting average conversion factor. Their individual batting average conversion factors are predictably high or low for young players and old players, respectively. They are

Too young
Jim Gilliam (1.01)
Elston Howard (.94)
Willie Mays (1.05 based on his rookie season only)
Al Smith (.93)

Too old
Luke Easter (.81)
Bob Thurman (.70)

The remaining eight players comprise the set of players who played in both leagues during their prime. With adjustment of batting averages for park factors and a few other small adjustments, the ratio of their Negro-League averages to their major-league averages falls into a quite consistent range of values. Since the goal is to find major-league equivalent values for Negro-League performance, my goal in making adjustments has been to eliminate growth, decline, and small sample errors by excluding seasons clearly affected by one of these factors from the data set. Here are the players and their personal ratios of major-league batting average to negro-league batting average.

Just right
Bob Boyd, .86 (1949 & 1950 NeL seasons excluded due to questions about level of competition, part-seasons excluded from ML average)
Roy Campanella, .85 (seasons before age 20 and two decline seasons excluded)
Larry Doby, .84 (two decline years excluded)
Monte Irvin, .86
Sam Jethro, .81 (Jethroe is a low outlier because he reached the majors at 32 and was a speed player)
Minnie Minoso 1.00 (Minoso is a high outlier, but he played in the NeL ages 24-26. A late bloomer?)
Jackie Robinson .88
Hank Thompson .86

These players’ batting average ratios average to .87, which is the figure I use.

Given that Minoso’s inclusion pulls that figure upward, exclusion of Minoso and Jethroe to get the most narrowly focused group might make sense. That would lower the ratio to .86.

This is enough data to give us a pretty fair starting point, I believe.

There are, however, several issues that might affect the conversions.

1) League offensive levels. Was the level of offense in the NeL during this period about equal to the level of offense in the majors? What effect would variations in offensive level in the majors (esp. during WWII) have on these conversion factors?

2) Quality of competition variations due to World War II and to the travel of many players in the Negro Leagues to Mexico during the early 1940s and to the entry of top black players into the majors and minors from 1946-1950. If these factors dropped the quality of competition in the Negro Leagues during the period of 1941-48 when all but a couple of the Negro-League seasons in this sample were played, that might mean that the conversion factor for 1920s and 1930s Negro-Leaguers should be higher.

3) Power conversion. With incomplete statistics, slugging average is very difficult to calculate consistently for Negro-League players. Limited data available suggest that power numbers may need to be prorated differently, but what the rate of conversion should be I don’t know. Regressing to the mean tends to draw high slugging averages down faster than batting averages, so that may be sufficient, but I’d be easier in my mind with conversion factors more firmly based in actual data.

I hope this data shows clearly why I use .87 as my conversion factor and the absence of evidence to suggest any factor far removed from .85 to .87.

If one believes that the conversion for power numbers should be lower, a factor of .81 or .82 might be used for converting slugging. If one believes that levels of competition were higher 1920 to 1940 in the Negro Leagues than after 1940, a conversion factor of .90 or .91 might be used for earlier stars.

Evidence of significant differences in offensive levels between the Negro Leagues and the major leagues would also affect the conversion ratio. More data is needed for this element to be accounted for systematically.

Any other issues I've missed?

Comments welcome!

More detailed breakdowns of player conversions available upon request.
   85. sunnyday2 Posted: January 04, 2005 at 03:13 AM (#1054864)
Chris, thanks, this is really good stuff. First, to clarify, the BA ratio means a .300 hitter in the NeL (X .87) comes out at .261 in the bigs, .350 = .308, etc?

Secondly, in moving from hits (BA) to XBH and/or HR, it seems to me you would need to figure out a ratio empirically rather than assuming .87 works or adjusting it arbitrarily. The .87 is empirical but for hits only. I think we would have to know, empirically, what the distribution of those hits was.

Third, therefore there is no way to assume that .87 would work for OBA or SA or OPS or OPS+. I've seen some massive OPS+ numbers for some NeLers--KJOK sent out some MLEs, which are great, by the way, more info rather than less and all of that, but if I recall we had Beckwith at a career 160. It would be really good to validate those kinds of numbers a little bit more than they are so far.

Of course, I'm inducting Dobie Moore into my PHoM this year.... Am I the first?
   86. Brent Posted: January 04, 2005 at 03:40 AM (#1054952)
Thanks, Chris. Very helpful.

If one believes that the conversion for power numbers should be lower, a factor of .81 or .82 might be used for converting slugging.

In the AAA minors->majors conversion I did on the Buzz Arlett thread, which is based on the Bill James 1985 Abstract, I notice the conversion factor he uses for hits is larger than for slugging. (The factor for hits is .89 = .98 * sqrt(.82), while the factor for HR is .82); see my post on that thread for all the factors.) I know other analysts have done more sophisticated work in recent years, but I haven't kept up on the factors that they use.
   87. Chris Cobb Posted: January 04, 2005 at 04:40 AM (#1055117)
Chris, thanks, this is really good stuff. First, to clarify, the BA ratio means a .300 hitter in the NeL (X .87) comes out at .261 in the bigs, .350 = .308, etc?

Yes, exactly. It's clear from the examples we have that the ratio is pretty consistent for major-league career averages from about .260 to about .310. Since the highest NeL career averages based on large numbers of at bats are in the .350s, I think we are fairly safe in assuming that this ratio is about right for all players we're considering, if competition level is consistent.

Third, therefore there is no way to assume that .87 would work for OBA or SA or OPS or OPS+.

Basically true, though my experience with doing conversions for Moore and Beckwith makes it clear that the ratio should be lower for SA.

I don't have the data to attempt an OBA conversion; when I match players to find win shares, I try to match the Negro Leaguers to major leaguers with similar levels plate discipline, as far as it can be determined.

For slugging average, some more work ought to be able to generate at least some empirical data.

Brent's comment above offers one model we might apply in the absence of data. James's old formulas would suggest a HR factor of .79, which is a little steeper than what I would have suggested. This steeper discount would be counterbalanced to some extent in the 1920s by the lower power levels in the Negro Leagues in comparison to the majors.

When I started work on conversions, I decided slugging wasn't all that important, as the correlation between value and batting average for hitters was very strong in the deadball era. If you matched up high batting averages and batting win shares, you would miss a few guys like Jimmy Sheckard who took a lot of walks, but you wouldn't include anybody who really wasn't a top-value hitter. After 1920, that correlation loosens a lot, and I haven't yet done a conversion study on slugging.

Having revisited and updated the basic data set for the batting average conversion study today, it should be a fairly quick process to put together some slugging conversion estimates. They'll be based on a smaller data set with more estimates built in (I seldom have totals for 2b, 3b, and hr for a player in a single year), but hopefully they'll show some consistency.
   88. jimd Posted: January 04, 2005 at 05:16 AM (#1055184)
Big Assumption: Assume that .87 also works for OBA and SLG.

Mappings from NeL OPS+ to MLB OPS+.
210 => 170
170 => 135
130 => 100
100 => 74
   89. jimd Posted: January 04, 2005 at 05:38 AM (#1055238)
An average major league line from 1930 was: .295/.353/.434 (.787 OPS). The equivalent NeL line would look like .339/.406/.499 (.905 OPS). Using .87, this would translate to an average major league hitter (assuming neutral parks, etc.).
   90. Chris Cobb Posted: January 04, 2005 at 03:37 PM (#1055675)
Report on slugging percentage conversions

Having reviewed my data, I find that there are only 10 player-seasons in which I have enough data to calculate an estimated slugging percentage for Negro-League players with major-league careers: three for Monte Irvin, three for Larry Doby, two for Roy Campanella, one for Jackie Robinson, and one for Sam Jethroe.

That’s not a lot to go on/ It's about 1/3 the NeL seasons contributing to the batting average conversion without the old and young players as outliers to show that the middle range is truly a representation of playes in their prime, and with a degree of uncertainty in every calculation: I do not have the full set of ab, hits, 2b, 3b, and hr for any season, so at least one component of the data is estimated in every case.

Here’s the player-by-player summary of results

Robinson .762
Irvin .816
Doby .833
Jethroe .874
Campanella .961

To reach a reliable general conversion factor, I’m inclined to leave Campanella out. His two seasons were at age 19 and 24, when we would expect his power still to be developing, so it’s not surprising that his NeL slugging is closer to his major-league slugging.

Weighting the four remaining player-conversions equally, they average to

Weighting them by the # of NeL seasons contributing to them, they average to

I’m thus inclined to use .82 as the conversion factor for slugging.

Below are the season-by-season breakdowns of the players. All numbers followed by a ? are my estimates. Given the small data set, a change of a couple of extra-base hits to singles can swing slugging percentages by 20-30 points, so we are not on terribly firm ground here.

Comments welcome!

1941 138 ab 53 hits, 9 hr, 7 2b, 2? 3b --> .659 slg, .646 park-adj.
1946 169 ab, 59 hits, 9 hr, 11 2b, 2? 3b --> .597 slg, .586 park-adj.
1947 296 ab, 94 hits, 14 hr, 18 2b, 2? 3b --> .534 slg., .523 park-adj.

.580 NeL career slg., park adj.
.473 ML park-adj.

Ratio .816

1945 102 ab, 35 hits, 10 2b, 5 hr, 2? 3b --> 64 tb, .627 slg
major-league slugging, less 2 decline years, park-adj. .478

Ratio .762

1941 96? ab, 33 hits, 5? 2b, 1? 3b, 4? Hr --> 52 tb, .542 slg
1945 116 ab, 37 hits, 7 2b, 1? 3b, 4 hr --> 58 tb, .500 slg

.533 NeL career slg
ml slugging, less 2 decline years, park adj. = .512

Ratio .961

1943 115 ab, 37 hits, 7? 2b, 3? 3b, 5 hr --> 65 tb, .565 slg, .554 park-adj.
1946 182? Ab, 65? Hits, 6 2b, 6 3b, 5 hr --> 98 tb, .538 slg, .528 park-adj.
1947 164 ab 51 hits 10? 2b, 4? 3b, 14 hr --> 111 tb, .677 slg, .663 park adj.
.594 Nel career slg, park-adj.
.495 ML park-adj.

Ratio .833

1946 220 ab, 68 hits, 10? 2b, 6? 3b, 6 hr --> 105 tb, .491 slg
major-league slugging, park-adjusted .429

Ratio .874
   91. Gary A Posted: January 05, 2005 at 06:02 AM (#1057978)
I guess this is the right place to post this.

1920 NNL park effects, for 6 NNL home parks plus associate member Bacharach Giants

team--park factor, adjusted park factor (home/road games)

Indianapolis--102, 101 (48, 43)
Bacharach Gts--119, 107 (8, 15)
Chicago Am Gts--85, 90 (34,20)
Detroit--121, 116 (51, 15)
Dayton--101, 100 (3, 18)
Kansas City--110, 106 (48, 29)
St. Louis--68, 86 (16, 20)

These are still simple park factors (total runs per game at home divided by total runs per gam on the road). Adjusted park factor is the average park effect faced by a team, assuming its road parks cancel out to 100.

I put in the home and road games (neutral-site games count as road games) to show how fantastically unbalanced schedules could. Though 1920 was an especially bad year, and the scores I have are incomplete (I'm missing several series played in St. Louis, in particular). There were an unusually low number of games this year, as the NNL was just getting organized.

Obviously, these data are of limited usefulness by themselves (samples are mostly too small--Dayton, for example). But 3-4 years of data start yielding meaningful patterns.

I don't have stats for 1920 beyond game scores, but teams averaged 4.49 runs/game, compared to 5.03 in 1921.

The Cuban Stars and Chicago Giants were both road teams in 1920.
   92. Gary A Posted: January 05, 2005 at 06:31 AM (#1058038)
By the way, the St. Louis park in 1920-21, known as Giants' Park, was *not* the same as the later Stars' Park. The newer park was built in the spring of 1922 when a new group of investors took over the St. Louis franchise in the NNL and founded the Stars.
   93. Gary A Posted: January 05, 2005 at 06:42 AM (#1058056)
Forgot to include the names of the 1920 parks:

Indianapolis--Washington Park
Bacharach Gts--Inlet Park (aka Bacharach Park)
Chicago Am Gts--Schorling Park
Detroit--Mack Park
Dayton--Westwood Field
Kansas City--Association Park
St. Louis--Giants' Park
   94. KJOK Posted: January 06, 2005 at 01:14 AM (#1059622)
By the way, the St. Louis park in 1920-21, known as Giants' Park, was *not* the same as the later Stars' Park. The newer park was built in the spring of 1922 when a new group of investors took over the St. Louis franchise in the NNL and founded the Stars.

And so far, the data I've seen all indicate that that later Star's Park was hugely hitter friendly, perhaps similar to Baker Bowl...
   95. Gary A Posted: January 06, 2005 at 02:59 AM (#1059797)
Yeah, the data are very different. In 1920-21, Giants' Park registered PFs of 68 (incomplete data) and 102; in 1923 Stars' Park got a 155, in 1928 a 125 (iirc).

I'm compiling 1922 scores next; it'll be interesting to see if the pattern holds up.
   96. Gary A Posted: January 06, 2005 at 03:15 AM (#1059824)
Here, btw, are updated 1921 park factors (I've been able to add a few scores, and made some corrections).

team--park factor, adjusted park factor (home, road games)

Indianapolis--107, 103 (48, 51)
Bacharach Gts--99, 100 (20, 49)
Chi. American Gts--54, 77 (41, 40)
Columbus--129, 113 (37, 46)
Cuban Stars (W)--111, 104 (29, 44)
Cleveland--110, 104 (10, 12)
Detroit--107, 104 (43, 33)
Kansas City--115, 108 (50, 50)
Pittsburgh--75, 79 (10, 2)
Hilldale--97, 98 (30, 14)
St. Louis--102, 101 (45, 30)

The parks:
Indianapolis: Washington Park
Bacharach Gts: Inlet Park
Chi American Gts: Schorling Park
Columbus: Neil Park
Cuban Stars (W): Redland Field, Cincinnati
Cleveland: Tate Field
Detroit: Mack Park
Kansas City: Association Park
Pittsburgh: unknown
Hilldale: Hilldale Park
St. Louis: Giants' Park

Redland Field was the Cincinnati Reds' home park at the time. Its 1921 park factor in the NL was 99; its HR factor was 14 (one of the all-time lowest).
   97. Chris Cobb Posted: January 07, 2005 at 03:53 PM (#1063391)
Some notes on Beckwith:

Kelly from SD wrote:

Before the park adjustments, he was 30th. After the park adjustments 9th. Without having park adjustments for the other NeL’ers I did not feel right about boosting Beckwith that much. For me, he epitomizes the talent/numbers dichotomy of the Negro Leagues. Gadfly and other say he had 40-50 homerun ability. His real numbers in league play show a little over 20 per 550. I prefer to err on the side of caution.

1) Most of the change in Beckwith’s totals did not come from park adjustments. In my second estimate of Beckwith’s win shares, I recalculated five seasons. Here are the reasons for the changes

1921 – Changed to in accordance with new, more complete statistics from 23.1 to 27.8
1922 – Changed because of park factors from 14.3 to 19.1
1923 – Adjusted for new, lower statistics and park factor, which offset, to leave total at 28.
1924 – Adjusted for new evidence about previously undocumented playing time for non-league (but strong) Homestead Grays from 22.3 to 29.9
1928 – Adjusted for new evidence about playing from 17.5 to 20.7.

2) In my use of park factors for _two_ seasons out of Beckwith’s career, I have not used the raw park factors Gary A has provided at full value. Rather, I have used them as an indicator of general park tendencies. Chicago was obviously an extreme pitchers’ park, but with small samples the degree of effect changed quite a bit from year-to-year. I have treated it as a having a park factor of 90, which I think is conservative.

3) Beckwith is not the first player for whom park adjustments have been made. I have used park adjustments for all Chicago American Giants players, because all the evidence shows that their park was _always_ a strong pitchers’ park. I made these adjustments for Pete Hill, Bruce Petway, and Christobal Torriente among serious candidates. I have also made park adjustments, albeit smaller ones, for Dobie Moore, Joe Rogan, and Heavy Johnson. All of Beckwith’s contemporaries from the 1920s will have park-adjustments made as appropriate when they become eligible. For players prior to the 1920s, we lack sufficient evidence (except in the case of Chicago), to make adjustments. But we can and should do it for players in the 1920s, when records are better.

4) The MLE win share calculations I have made are based on his real numbers, not on gadfly’s 40-home-runs-per-year view. I think it possible that Beckwith might have hit 40 home runs a couple of times in the majors if he had played in a good home run park, but he would have averaged more like 15-20. There are not many players who were doing that in the 1920s.

Brent wrote:

I greatly appreciate and admire the work that Gadfly, Chris Cobb, KJOK, Gary A, and others have contributed in gathering data, analyzing, and interpreting it. But I also greatly respect experts like Holway and Riley who have devoted years to study of the Negro Leagues, and I am troubled by the disparate interpretations of Beckwith’s career that we receive from these sources.

Holway, it should be noted, named Beckwith as a candidate he would elect to the Hall of Fame in the _Cool Papas_ survey, listing him among the shortstops deserving of election. He also listed him as his #2 utility player for the all-time Negro League team, (after Josh Gibson). The “disparate interpretation” we have is Riley’s, and the disagreement is over character, not talent. Riley begins his bio of Beckwith as follows: “The right-handed pull hitter was a powerful and consistent hitter whose slugging prowess was extraordinary. Despite facing a severely overshifted defensive alignment, he maintained a high batting average and proved that nothing could stop his prodigious blasts out of the ballpark.”

I’ve always found Chris’s ML projections to be very helpful, but they need to be checked and prodded, just as my own Arlett projections need to be. In Beckwith’s case there are some technical features, such as the influence of an extreme park factor in 1922, that may merit further inspection. (For example, I noticed in the data that Gary A posted last night that the park effects for Schorling Park jump around quite a bit from year to year).

See above on my use of park adjustments.

I’m concerned about the dissonance between the rosy biography presented by Gadfly and the decidedly negative biography of Riley.

I think voters have perhaps been too quick to dismiss character issues, especially to the extent that they may have affected team performance. Ty Cobb’s behavior could sometimes be abhorrent, but the Tigers never gave him an unconditional release in mid-season as the Homestead Grays are reported to have done with Beckwith (according to Riley).

If we can resolve some of these issues, I’m willing to give my support to Beckwith, but his is the most complicated case before us, and I would be more comfortable if we took a few more “years” to consider it.

On the unconditional release due to unreliability reported in Riley, I am doubtful. If Beckwith was so unreliable with the Grays, why did Homestead bring him back for a while in 1926 and for a full season in 1928? Why (as gadfly justly pointed out) did teams make him the manager? I am dubious about Riley’s treatment of Beckwith in general because, although _many_ black players jumped teams, often during the season, only Beckwith is judged to be a bad apple because he was doing it.
   98. Gary A Posted: January 07, 2005 at 06:23 PM (#1063752)
I've got the Pittsburgh Courier's Negro League coverage for 1924 photocopied--I will go through it over the weekend to see if there's any commentary on Beckwith and the Grays. I can also go through the Chicago Defender for that year, and 1922-23, when he was with the American Giants.

The Baltimore Afro-American might have stuff on Beckwith with the Black Sox, but I don't have those years in my files.
   99. jimd Posted: January 07, 2005 at 07:25 PM (#1063912)
1917   96     97     97     94    101    101
1918   98     97     98     95    102    102
1919   97     95     99     96    101     99
1920   99     97     99     96    104    101
1921   95     95    100     97    102     99     98     97    106    105
1922   98     97    101     98    102     99     95     94    104    103
1923   98     97                                 98     98    106    105
1924   99     97                                 98     98    106    107
1925   98     96                                102    101    105    106

Park Factors (Lahman DB, 2002 edition) for Redland Field/Crosley Field and for the shared fields between the NL and AL: New York's Polo Grounds pre-Yankee Stadium, St. Louis' Sportsmans Park post Robison Field.

Redland Field plays as a mild pitcher's park in the NL, not that different from the Polo Grounds or Sportsmans Park in its effects on runs (its effects on run elements are quite different though; it was very hard on Home Runs while the other two were both excellent Home Run parks). As one can see, the latter two parks play as hitter's parks in the AL.

This implies that there was a very distinct difference between the relative park factors in the NL and the AL. The shared parks were pitcher's parks in the NL and hitter's parks in the AL. This implies that the remainder of the NL was more hitter friendly than the remainder of the AL.

The Redland Field discrepancy implies the same thing, that the NL was more hitter friendly than the Negro League parks. Trying to compare the AL with the NeL is getting very shakey (trying to get too much inference out of one data point), but that first data point could imply that they might be similar.
   100. Gary A Posted: January 08, 2005 at 07:36 AM (#1065378)
From the Chicago Defender, 12-29-1923:

The announcement of Beckwith’s signing with the Homestead Grays “comes not as a surprise as the first baseman’s playing of last season was not up to what it had been the previous years. He was unable to hit in pinches and it is a known fact that he was dissatisfied with the salary he was receiving although he signed a contract for that amount. It is believed that if he has signed a contract to play in Pittsburgh, he believed that Foster was about to trade him.”
Page 1 of 4 pages  1 2 3 >  Last ›

You must be Registered and Logged In to post comments.



<< Back to main

BBTF Partner

Dynasty League Baseball

Support BBTF


Thanks to
for his generous support.


You must be logged in to view your Bookmarks.


Page rendered in 1.1120 seconds
39 querie(s) executed