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

Tuesday, August 17, 2004

Smokey Joe Williams

Who was better during the Deadball Era: Johnson, Alexander or Williams?

That’s how good Smokey Joe was that he doesn’t look foolish being in that question.

John (You Can Call Me Grandma) Murphy Posted: August 17, 2004 at 09:08 PM | 38 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.

   1. The definitely immoral Eric Enders Posted: August 17, 2004 at 10:52 PM (#802570)
Another slam-dunk HOMer. Here's an encyclopedia entry I once wrote on Williams. (It's from the Encyclopedia of Native Americans in Sports, which is the reason for the focus on his matchups with Bender.)


One of the greatest pitchers in baseball’s Negro Leagues, Smokey Joe Williams is one of a handful of Native Americans inducted into the Baseball Hall of Fame. In a career reminiscent of Nolan Ryan’s, Williams used an overpowering fastball to win untold hundreds of games and strike out thousands of opposing batters during a career spanning nearly three decades.

Although official birth records do not exist, Williams is believed to have been born on 6 April 1886 in Seguin, Texas, to an African American father and Native American mother. (His mother was reported to be a Cherokee, although this has not been confirmed.) Williams began his baseball career on the sandlots of Central Texas, where he played for teams in San Antonio and Austin from 1905-09. In 1910 Williams joined the Chicago Giants, a formidable independent team, and a year later he moved on to the New York Lincoln Giants, a premier African American team of the day, for whom he would pitch more than a decade. The Lincoln Giants compiled impressive won-lost records against other black teams and in exhibitions against “white” major leaguers. Williams is known to have faced the other standout Native American pitcher of the day, Charles “Chief” Bender, at least twice. Williams defeated Bender by scores of 2-1 in 1913 and 11-1 in 1917.

In thirty-one documented games against major league competition, Williams compiled a 22-7 record. However, two of Williams’ most impressive feats – a no-hitter against the New York Giants and a 1-0 victory over Walter Johnson, both reportedly in 1917 – have yet to be historically documented and survive only in tales passed down from generation to generation.

A tall, lean man with an extraordinary fastball, Williams claimed to have pitched five no-hit games in his career, including one against ex-teammate Dick Redding on Opening Day 1919. His decade of excellence in New York made him a well-known figure in Harlem, and he developed a reputation as a “stage door johnny” before marrying an ex-showgirl in 1922. In 1924 the Lincoln Giants released Williams because they believed that, at age thirty-eight, his best days were behind him.

In 1925 Williams signed with the Homestead Grays, a powerful Pittsburgh-based team, where he enjoyed an extraordinary late-career renaissance. On 7 August 1930 Williams pitched perhaps the greatest game in Negro Leagues history. In a night game against the renowned Kansas City Monarchs, he allowed only one hit and struck out twenty-seven batters in a 1-0, twelve inning victory. By this time Williams had become famed as the Methuselah of the Negro Leagues, an image he promoted by claiming to be fifty years old although he was really forty-four.

Williams retired from baseball in 1932, at age forty-six, after pitching for twenty-eight seasons. He found work as a bartender in New York, where he traded baseball tales with eager customers. In 1950 Williams, then in ill health, was honored before a Sunday afternoon game at the Polo Grounds. He died in New York on 25 February 1951. The next year, a Pittsburgh Courier poll of black baseball officials and sportswriters named Williams the greatest pitcher in the history of the Negro Leagues. Nearly half a century later, on 25 July 1999, he was finally inducted into the Baseball Hall of Fame.

See also: Bender, Charles Albert “Chief”

Holway, John. 1988. Blackball Stars. New York: Carroll & Graf.
Lester, Larry, and Dick Clark, eds. 1994. The Negro Leagues Book. Cleveland, OH: Society for American Baseball Research.
Research Files. National Baseball Hall of Fame Library, Cooperstown, New York.
Riley, James. 1994. The Biographical Encyclopedia of the Negro Baseball Leagues. New York: Carroll & Graf.
   2. John (You Can Call Me Grandma) Murphy Posted: August 17, 2004 at 11:56 PM (#802701)
Nearly half a century later, on 25 July 1999, he was finally inducted into the Baseball Hall of Fame.

He should have been elected into the HoF as one of the very first Negro Leaguers, IMO.

Nicely written passage, Eric.
   3. Dag Nabbit at Posted: August 18, 2004 at 05:35 PM (#804301)
He died in New York on 25 February 1951. The next year, a Pittsburgh Courier poll of black baseball officials and sportswriters named Williams the greatest pitcher in the history of the Negro Leagues.

Any possibility that the standard post-death misty-eyed treatment pushed him past Paige in the poll? Not saying he wasn't a great pitcher - he was, but this could be a reason to question the Courier's ranking of him.
   4. The definitely immoral Eric Enders Posted: August 18, 2004 at 06:07 PM (#804366)
I've suspected the same thing myself. Alas, there's no way to really know.
   5. The definitely immoral Eric Enders Posted: August 18, 2004 at 06:33 PM (#804398)
Any possibility that the standard post-death misty-eyed treatment pushed him past Paige in the poll?

In a related matter, the sentiment over Williams' death did manifest itself in another way: It started the movement to allow black players into the Hall of Fame.

Sometime in 1951, shortly after Williams died, sportswriter Joe Bostic became the first person (so far as I can tell) ever to publicly advocate for the admission of Negro Leaguers into Cooperstown.

This was followed by a gradual groundswell of support. Satchel Paige got one renegade write-in vote in the 1953 election, and continued to receive a handful per year annually after that. (Although the Hall declined to keep counting them after that first year.) Eventually this all led up to the activism of Stan Isaacs and Ted Williams, and you all know what happened after that.
   6. The definitely immoral Eric Enders Posted: August 18, 2004 at 06:35 PM (#804402)
Here is the aforementioned Joe Bostic column, published in 1951:

Continuing The Plea For Revocation of James Crow's Membership In Baseball's Hall of Fame

News this week of the passing of Eddie Collins, peerless second sacker of the legendary $100,000 infield once put together by Connie Mack seemed, to this pillar at least, to bring into sharp focus the urgent need for the revocation of the membership of James Crow (ol’ Jimcrow to you) in the legion of diamond immortals in the baseball shrine at Cooperstown. This column realizes full well that the late keystone wizard belonged in every sense of the word.

The thing which bothers this observer is that so many superlative baseballers will go on to their reward unsung and unhonored despite the fact they were specialists at their respective positions. We find it hard to accept such discrimination as being either in keeping with the moral code of sports or the noble ideal behind the establishment of the shrine at Cooperstown.

Let's put it another way. If such a violation of the sports code is acceptable to the policy makers at Cooperstown, then the shrine should be blasted to smithereens and torn from its foundations as a blasphemy to the sanctity of the sports ideal.

The inequity and unfairness of such a setup is graphically underscored by the fact that the immortal Cyclone Joe Wllllams passed away some three or four weeks ago and fully nine-tenths of the baseball fans of the country knew nothing about it. Ask any informed baseball authority or any pro who had to bat against the man and you'll be told quickly that the Cyclone could throw with any man who ever tossed the little round pellet plateward. Still, despite this great talent, the present Jimcrow setup at Cooperstown can't accommodate Mr. Williams' name even though he were nominated. This is an injustice that cries to be wiped out.

The reams of copy that have appeared in the newspapers throughout the country have apprised every sports fan not only of the death of Mr. Collins but also of his brilliant career. And long after the pages of the papers have yellowed with age, the colorful saga of Eddie Collins will be kept alive in the lily white shrine at Cooperstown.

Now what happens to the equally lustrous story that might be written of Cyclone Joe and the others like him, whose skin pigmentation makes them ineligible for baseball's roll call of renown? Why that story fades and passes along with the passing of his contemporary witnesses, who either watched him from the stands or played behind or against him.

Years from now, the only record of the derring do of the Cyclone Joes will be the same one that recounts the legendary deeds – and misdeeds – of the likes of “Charlestown Slim” or "Cornbread," the pool shark. Naturally the word of mouth telling of the stories will become more fascinating with each passing generation, but it will lack the definite authenticity of the written and preserved record.

You think of Hans Wagner, the wonderful shortstop and his rightful place in the all time listing of super stars. He'll never be forgotten, nor his achievements. This too is as it should be.

But what about Pop Lloyd, who had no peer when it came to covering the great expanse of real estate between second and third? Ask anybody who ever saw him go for an estimate of his ability. You'll be told mighty quickly that he was in a class by himself. And when they tell you that, no stipulation is made as to race, color or leagues. He was hell on wheels by any standards. But there is no provision made for keeping for posterity any record of his sparkling scoops and slick maneuvers. In such a case, it seems to me that not only is an immortal being slighted but the Hall of Fame itself is being cheated out of a justifiable membership.

There is but one answer to this obviously wrong and immoral [definitely! – ed.] setup. There must be a complete re-examination of ALL ballplayers in the light of the fact that there have been those with extraordinary talents, even though they performed outside the ranks of organized baseball because the doors were closed to them.
Jimcrow was understandable, if not forgivable, in the frankly commercial baseball setup. But Cooperstown takes its genesis in the nobility and dignity of tradition.

James Crow can't possibly be an acceptable citizen in such a community. Not unless pollution of the membership body is preferable to purity.

I don't think it is.

Or, letting the wish sire a noble thought: I don't want to believe that such is the case.
   7. KJOK Posted: August 19, 2004 at 12:00 AM (#805005)
Here's Williams complete 1928 stat line:

RA/G - 3.21
OppBA - .257
   8. jonesy Posted: August 19, 2004 at 08:29 PM (#807294)
I lean in favor of Williams over Paige. The trouble I have with Paige -- like Dizzy Dean -- is how to wade through all the media hype.

The guy I find real fascinating though, is Will Jackman.

"...enjoyed his prime years in the late 20s and early 30s. During this time he was credited with a 52-2 mark for one season with the Giants and bested Satchel Paige twice in two outings. Jackman's fastball was described as being faster thein either Paige's or Bib (sic, must mean Bob) Feller's..." or so writes Jim Riley.

Argh, how the heck to sort through this.
   9. Patrick W Posted: August 20, 2004 at 04:13 AM (#808760)
Tried to follow Chris C's methodology. Notes:
- i9's ERA reduced 5% to est. MLE (No idea if this is appropriate for this case)
- took the time to avg together the nl & al ERA's for ERA+ calc.
- 8.5 IP per decision
- For WARP, I used 0.035 wins/game for a replacement player (replacement team = .035W/G * 162G * 9 players = 51 wins). So WARP1 = (WS/3) - .035*(IP/9)

Smokey Joe Williams
Yr     IP    ERA+  SN W%  SN W  SN L pWS   pWARP1
1910   294   118   0.584   20   15   22.1    6.2
1911   223   134   0.642   17    9   20.9    6.1
1912   290   108   0.538   18   16   18.8    5.1
1913   334   137   0.654   25   14   30.4    8.8
1914   336   154   0.703   28   12   35.5   10.5
1915   249   161   0.722   21    8   27.4    8.2
1916   351   171   0.745   31   10   41.4   12.4
1917   387   155   0.707   33   13   42.4   12.6
1918   360   182   0.768   32   10   42.9   12.9
1919   327   179   0.762   29    9   39.0   11.7
1920   295   120   0.591   21   14   24.1    6.9
1921   318   141   0.664   25   12   31.4    9.2
1922   291   164   0.728   25    9   32.9    9.8
1923   186   137   0.651   14    8   16.8    4.9
1924   232   168   0.738   20    7   26.5    7.9
1925   206   121   0.593   14   10   15.9    4.5
1926   211   142   0.667   17    8   21.2    6.2
1927   194   147   0.685   16    7   20.3    6.0
1928   155   128   0.620   11    7   13.0    3.7
1929   199   155   0.705   16    7   20.5    6.1
1930   240   111   0.553   15   13   15.9    4.4
1931   236   120   0.590   17   11   19.7    5.6
1932   176   117   0.578   12    9   13.2    3.7
Tot   6090   139   0.660  477  238  592.2  173.7

For WARP2 the difficulty for Walter Johnson (generally the same career timespan) is 90%, while the schedule adj. for Big Train is 103%.
I would estimate Smokey Joe's W2 & W3 at 156.3 & 161.0 respectively.
   10. yest Posted: August 20, 2004 at 06:29 PM (#809372)
some stuu on Williams
His best-documented season was 1930, when, at age 44, he was 7-1 in league play.

In 1924 he struck out 24 batters in a 12-inning game against the Bushwicks, a powerful semi-pro team. In 1930 he struck out 27 while one-hitting the Kansas City Monarchs over 12 innings.
In exhibition games against major leaguers, Williams compiled a 22-7-1 record with 12 shutouts. Two of the losses came when he was 45 years old; two others were in 1-0 games. In 1912 he shut out the National League champion New York Giants 6-0. In 1915 he struck out 10 while hurling a 1-0 three-hit shutout over Hall of Famer Grover Cleveland Alexander and the Phillies. In a 1917 exhibition, he no-hit the Giants and struck out 20, but lost 1-0 on an error. Legend has it that it was after this game that Giants Hall of Famer Ross Youngs tagged Williams with the name "Smokey Joe.
Williams threw approximately 40 no-hitters, some against semi-pro competition, recording his last gem in 1928 at forty-two. (JJM)
Chicago Giants and owner Frank Leland described him "If you have ever witnessed the speed of a pebble in a storm you have not even seen the equal of the speed possessed by this wonderful Texan Giant. "

3 times he faced the NL Champions in post-season play.
The first showdown was in 1912, when he shut out the World Champion New York Giants 6-0.
The second match-up was in 1915, when he struck out 10 batters while throwing a three-hit shutout against Grover Cleveland Alexander and the Philadelphia Phillies, winning by a score of 1-0.
The third time occurred in 1917 when he struck out 20 batters while no-hitting the New York Giants, although he lost the game 1-0 on an error.

In 1930, the Grays won the Eastern Championship by winning a challenge series over the Lincoln Giants. The following season with Josh Gibson, Oscar Charleston, Jud Wilson they repeated as Champions.

1905 28-4 875 winning%
1906 15-9 625 winning%
1907 20-8 714 winning%
1908 20-2 909 winning%
1909 32-8 800 winning%
1914 41-3 932 winning%
Smoky Joe Williams went 107-57 with a 652 winning% for his career
Smoky Joe Williams won 1 batting title hitting 474 in 1917
Smoky Joe Williams had the most wins 7 times in 8 years (1912,-14 1916-19) going 3- 2 , 12- 5 , 12-2 , 5- 6 , 9- 1 , 7- 2 , and 6- 3.
Ty Cobb, noted for his racial intolerance and general surliness, stated Williams would have been a sure 30-game winner had he been a Major Leaguer.
It was impressive that Williams went 20-7 with 12 shutouts against Major League competition, defeating Hall of Famers Grover Cleveland Alexander, Walter Johnson, Waite Hoyt, Chief Bender and Rube Marquard in the process.
The New York Lincoln Giants, highest-paid player, earning $105 per month.
His signature season was 1914, when he rung up 41 victories against a mere three defeats, a record compiled while pitching against all levels of competition (in league play he was 12-2 with 100 Ks in 17 games).
on Aug. 7, 1930 against Chet Brewer and the Kansas City Monarchs Smokey Joe one-hit the Monarchs and fanned 27 batters in a 12-inning game winning 1-0
"On certain days, our Negro National or Negro American League clubs could have been Major Leaguers," said former teammate Bill Yancey in Robert Peterson's 1970 book "Only the Ball Was White". "With Smokey Joe Williams or Cannonball Dick Redding or Phil Cockrell or Nip Winters pitching, we could beat anybody."
He threw so hard and so fast, he earned the nicknames "Cyclone" and "Strikeout."
legend has it, after Williams struck out 20 white New York Giants in a 1914 exhibition game, one of those Giants players patted Williams on the rear end and said: "Nice job, Smokey."
Records show Williams outdueled the great Walter Johnson twice, some say more, as well as Grover Cleveland Alexander, and Williams reeled off a 20-7 record in exhibitions against white major-league competition.
On Aug. 2, 1930, as a 44-year-old pitcher for the Grays, he struck out 27 Kansas City Monarchs and threw a one-hitter in a 12-inning game.
Satchel Paige, considered him to be the greatest pitcher he had ever seen.

His Hall Of Fame Plaque reads
NEGRO LEAGUES, 1910-1932
   11. OCF Posted: August 20, 2004 at 06:53 PM (#809415)
One thing about that 1930 27-strikeout game that is so central to Williams's legend (note how many times the reference gets repeated in yest's multi-source complilation): That was a night game, right? With the lights that were available in 1930 (several years before being adopted in the majors)? It doesn't take much to assume that the batters were having severe visibility problems.

I'm not knocking Williams. Taking Andrew Siegel's numbers in #9: maybe he's 100% of that, maybe he's 70% of that, maybe he's somewhere in between - but in any of those cases, he's obviously qualified for the HoM. I have no clear idea how to rank him compared to Speaker or Collins, or, for that matter, Alexander, whom he may share a ballot with. But they'll all get their turn.
   12. Chris Cobb Posted: August 20, 2004 at 07:13 PM (#809437)
Here's Williams complete 1928 stat line

KJOK, I've been meaning to ask -- Do you know what percentage of the recorded Grays games Williams pitched that year?

Are we looking at 3 starts and one relief appearance among 20 recorded team games, or 40?
   13. andrew siegel Posted: August 20, 2004 at 07:50 PM (#809506)

Those aren't my numbers in #9; Patrick W should get the credit.

I'm having a hard time believing Williams really would have won 477 games in the major leagues between 1910 and 1932. That's substantially more than Johnson won in roughly the same period and he's probably the best pitcher ever. It's almost as many as Young won in a much easier environment and he's considered a one-of-a-kind freak for his achievment. It's basically 100 wins more than anyone other than Johnson and Young have ever won. Let me put it this way: if those numbers are right, then Williams (not the usual suspects Johnson, Young, Grove, Paige, or Clemens) is the greatest pitcher of all-time.

My subjective assessment is that Williams was a great pitcher and an inner-circle HoMer, but not on par with Johnson or Grove. I think the best case scenario is that he was even with Alexander; the worst case scenario is, say, Robin Roberts. I'll split the difference and peg him to Christy Matthewson. That's fifth on his ballot but in the top 50 All-Time.
   14. John (You Can Call Me Grandma) Murphy Posted: August 20, 2004 at 08:05 PM (#809526)
I'm having a hard time believing Williams really would have won 477 games in the major leagues between 1910 and 1932.

Andrew, since his WARP numbers are smaller than Johnson's, plus his WS numbers are only slightly greater, Chris' projections are not suggesting the superiority that you are implying. It does suggest the problem of translating Wins though.
   15. OCF Posted: August 20, 2004 at 08:23 PM (#809557)
Andrew - that was careless of me. Sorry for the misidentification.

I think the best case scenario is that he was even with Alexander; the worst case scenario is, say, Robin Roberts. I'll split the difference and peg him to Christy Matthewson.

That's just about what I'm thinking. I know for sure that Alexander was better than Mathewson, that Pete was a lot closer to being Johnson than to being "merely" Mathewson. It's within the range of reasonable possibilities that Williams was better than Alexander, but it's a pretty wide range, and contains some lower values as well.
   16. Chris Cobb Posted: August 20, 2004 at 09:15 PM (#809656)
I'm having a hard time believing Williams really would have won 477 games in the major leagues between 1910 and 1932.

I'm curious as to whether Patrick (to whom I'm immensely grateful for producing these numbers!!) made any downward adjustments for the W-L record produced by my method in Williams' top ERA+ seasons. My experience with Foster and Mendez indicated that at the high end of ERA+, the pythagorean method produces results that are too high. You can see what's going on in comparing Johnson's best seasons to Williams', as projected by i9s.

Here are Johnson's two best seasons
1912 -- 369 IP, 240 ERA+, 33-12 (RSI 100, slightly above avg. fielding support)
1913 -- 346 IP, 259 ERA+, 36-7 (RSI 96, slightly above avg. fielding support)

Here are Williams' two best
1916 -- 351 IP, 171 ERA+, 31-10 (assumed support-neutral)
1917 -- 360 IP, 182 EERA+, 32-10 (assumed support-neutral)

The system is cleearly projecting more wins for Williams than actual play indicates he would get with his ERA+.

Johnson's 1910 seasons is pretty comparable to Williams' best
1910 -- 340 IP, 183 ERA+, 25-17 (RSI 112, avg. defense)

Johnson may have been a bit unluckly that year, but it seems unlikely that Williams would have achieved 30+ support-neutral wins in these seasons.

I suspect that using Pythaganport to determine the exponent for turning ERA+ into a winning percentage might correct part of this problem. Patrick could tell us whether he used the straight pythagorean method (as I do, since I'm computationally challenged) or the more complex pythaganport one (which I'm not equipped to explain, but has been explained in the discussions several times).

Another consideration: WARP shows that the great pitchers tended to underperform their expected wins, even with proper exponents used for projections. Johnson misses his expected wins by 41! So it's unlikely, even if this projection is as accurate as possible, that Williams would have won close to 477 games.

So, all that considered, these projected numbers, which apparently show Williams to be better than Johnson, probably shouldn't be interpreted as doing so. The ERA+ itself (a notch lower than Johnson's) is evidence that there's a problem is in the projection of ERA+ into support-neutral wins. Patrick has corrected for the problem pretty well in estimated WARP I think, but the win numbers themselves are too high.

I think the projection does claim that Williams is somewhat better than Alexander: his era+ is slightly better than Pete's, and he threw nearly 1000 more innings. If this ERA+ and IP is reasonable, I think Williams would probably have been in a position to win about 400 games.

Two questions that we could try to answer to place Williams more precisely in the pantheon of all-time great pitchers:

1) is an MLE of 6000 innings pitched reasonable for Williams' career?

2) is an ERA+ of 139 reasonable? (Pete Alexander's is 135.)

I've just begun compiling my data on Williams, so I have no answers to offer myself, and I don't have a system for estimating playing time in any case (though I'm going to try to think about that), so I'd really like to see what others come up with on these.
   17. OCF Posted: August 20, 2004 at 09:30 PM (#809677)
Just for some backup data to go with Chris C's post above (and fixing his typo on 1910 IP):

Using the sliding exponent that adjusts for run environment, and using 1 decision per 9 IP (which adjusts Johnson's decisions downward), I get the follow equivalent records:

1913: IP: 346; RA+ 275; equiv. record: 32-06 (actual 36-07)
1912: IP: 369; RA+ 204; equiv. record: 32-09 (actual 33-12)
1910: IP: 370; RA+ 160; equiv. record: 28-13 (actual 25-17)
   18. Patrick W Posted: August 20, 2004 at 10:28 PM (#809762)
More info:
- I'm not smart enough to know how to calc py-port; my sn w% is classic pythag.
- I have not adjusted Williams' peak seasons down.
- I'm throwing numbers up on the board, but I'm just like everyone else here, asking: what do these numbers mean?

For comparison purposes, here is Johnson by the same method

Walter Johnson
Year   IP   *ERA+  SN W%  SN W SN L  pWS   pWARP
1907   110   129   0.622    8    5    9.4   2.7
1908   256   139   0.658   20   10   24.9   7.3
1909   296   109   0.544   19   16   20.2   5.6
1910   370   183   0.770   34   10   45.5  13.7
1911   322   172   0.748   28   10   36.7  11.0
1912   369   240   0.852   37    6   52.4  16.0
1913   346   259   0.870   36    5   51.1  15.7
1914   372   164   0.729   32   12   41.6  12.4
1915   337   191   0.785   31    9   41.5  12.5
1916   370   147   0.685   29   14   36.4  10.7
1917   326   119   0.587   22   16   24.9   7.0
1918   326   214   0.820   31    7   42.9  13.0
1919   290   214   0.821   28    6   38.8  11.8
1920   144   118   0.584   10    7   11.3   3.2
1921   264   117   0.578   18   13   20.3   5.7
1922   280   129   0.626   21   12   25.2   7.3
1923   261   109   0.541   17   14   18.2   5.1
1924   278   148   0.685   23   10   29.1   8.6
1925   229   137   0.653   18    9   22.3   6.5
1926   261   106   0.531   16   15   16.1   4.4
1927   108    80   0.387    5    8    3.2   0.6
Tot   5915   146   0.693  483  214  612.0 181.0

Compared with actual,
W-L: 417-279 (the Senators in the Johnson years were still the Senators - a bunch of 6-7-8th place finishes in there. I think the SN W-L is reasonable)
pWS: 532 (95% of 560 - don't have the elec. version. My estimate above is 15% high)
WARP1,2,3: 211.7, 189.7, 195.6 (based on the 3 RAR components, WJ's value is 98.8% PRAR; I'm saying WARP ~= pWARP. My estimate above is 14.5% low)
   19. OCF Posted: August 20, 2004 at 11:54 PM (#809948)
Patrick W:

That differs from my numbers in three ways:
1. You used ERA+; I used RA+
2. You used a fixed exponent; I used a variable (Pythaganport) exponent.
3. You pegged equivalent decisions to actual decisions; I pegged them to IP/9.

Johnson's actual record was 417-279; your version of an equivalent record is 483-214; my version of an equivalent record is 427-230.

Points #1 and #2 have gotten a lot of publicity; I want to concentrate here on point #3.

I can see now that 9 innings per decision isn't quite right, but I still like it. It works pretty well in modern baseball. The one modern pitcher I've worked up completely is Roger Clemens, and he has 9.06 IP per decision.

I have about 60 pitchers from 1893 through about 1940 in my database. Of course, they're all very good pitchers - at worst, being investigated for possible HoM qualification. For them, the median number of IP per decision is 8.75. Of them 8 have more than 9.00 IP per decision. The most extreme is Walsh at 9.23 and the others are the AL part of Young (I don't have the rest of Young in the system), Reulbach, Leever, Mays, Joss, and Cicotte. At the other end, I have 6 pitchers with under 8.50 IP per decision: Shawkey, Quinn, Dauss, Wood, Griffith, and Ferrell. Griffith, at 8.28 IP/decision, is a fairly extreme case. I was once a strong supporter of Griffith, but that was based in part on an error in my records. As my support for him faded, he became quite popular with the rest of the electorate. Why do other voters think more of him than I do? Possibly because I'm giving him credit for fewer equivalent decisions than they are.

Walter Johnson checks in at 8.50 IP/decision.

There are some others in that neigborhood: Mathewson at 8.52, and Brown at 8.65 (8.53 if you leave out the FL).

But Johnson does have a rather low IP/decision ratio, or another way to put it, a high number of decisions for as much as he pitched.

Of course, I am NOT talking about durability or about IP/GS. The high number of decisions reflects a high number of decisions in relief.

For the record, Alexander is closer to the other end of the scale: 8.93 IP/decision.
   20. Patrick W Posted: August 21, 2004 at 12:25 AM (#810048)

My intent wasn't to match any numbers, yours or otherwise - it was to use the exact same method that i had used for Smokey in #9, and hopefully the same method that Chris used for Mendez on that thread. My hope is that you might compare #9 to #18 and compare #18 to whatever numbers you have for Walter, then adjust the #9 stats into your system accordingly.

I used ERA+ because that was what was used for Mendez; i9s don't provide an RA and i think that assuming any percent of unearned runs on Joe only adds to the uncertainty of the analysis. If you disagree, I hope I have provided enough information that you can modify the numbers accordingly.

I didn't peg actual W+L = SN W+L - I used IP/8.5 because Chris used IP/8.5 in his analysis previously. Johnson's decisions just happened to match that value rather snuggly.
   21. Chris Cobb Posted: August 21, 2004 at 02:03 AM (#810362)
OCF and Patrick,

Thank you both for most informative posts that clarify greatly how the projected statistics produced by my far-from-perfect projection method should be interpreted.

I picked 1 decision per 8.5 innings pitched because it was a pretty typical ratio for the era, and I wanted to create a record that would be readily comparable to the actual records of contemporaries.

I really wish we could get a set of pythaganported projected support-neutral won-lost records for Williams: that would be a lot better than having a projection that we know is off by some factor and then adjusting backwards.

OCF, what data would we need to put together for Williams in order to be able to apply correctly a sliding exponent to his won-lost record?
   22. OCF Posted: August 21, 2004 at 08:34 AM (#810821)
What do I need to know? Either a RA+ or an ERA+ for each year, a number of IP for each year, and a local run environment. Everything you know about league runs, park factors, defensive quality, league strength - that all goes into "local run environment." It also goes into RA+ or ERA+. Of course a great pitcher will by his own presence lower the run environment. We will use that to calculate the Pythaganport exponent.

I decided to interpret Chris's challenge as narrowly as I could, by taking the data in Patrick W's table at face value. Any doubts you have about that data becomes doubts about what I'll do with it. Patrick said he was assuming a run environment that was an average of the American and National leauges, so I used that raw run data.

There are a number of very large uncertainties behind everything in Patrick's table, notably including the IP and the ERA+.

I then took the following steps:

1. Based on my subjective impression that ERA+ tends to exaggerate extremes compared to RA+, I applied a function that pulled ERA+ back toward 100. Johnson's 5 best ERA+ were 259, 240, 214, 214, 191. This function pulled those back to 214, 201, 184, 184, 168. For the sake of calibration, note that his RA+ for those five years was 275, 204, 181, 160, 182.

2. Calculated the PythPort exponent as P = [(R/G)*(1+100/ERA+)]^.286. The ERA+ used there was the modified version from #1.

3. Computed a winning percentage: [(ERA+/100)^P]/[(ERA+/100)^P+1]. Again, this uses the reduced ERA+.

4. Took IP/9 to get a number of decisions. As I noted in a previous post, pitchers seem to range from a little under 8.5 IP/dec to a little over 9.0 IP/dec, with a pretty big cluster around 8.75. I'll stick with 9.0 because it's what I've been using.

5. Took decisions time winning percentage to get a year-by-year record. Career sums are sums of this record.

Apply that to Walter Johnson and I get a career record of 421-236. That's 6 games worse than my RA+ record of 427-230, which is really pretty close. Because I lost the great RA+ of 1913, this procedure made 1912 his best season at 32-9, although 1913 was close behind at 30-8.

The next post will contain the table I got for Joe Williams.
   23. OCF Posted: August 21, 2004 at 08:38 AM (#810822)
Joe Williams, table after Patrick W, but with the adjustments indicated in the previous post:
Year     W   L    
1910    18  14 
1911    15  10 
1912    17  15  
1913    23  15  
1914    24  13  
1915    18  09  
1916    26  13  
1917    28  15 
1918    28  12  
1919    25  11  
1920    19  14 
1921    22  13  
1922    22  11  
1923    13  08  
1924    18  08  
1925    13  10  
1926    15  09  
1927    14  08  
1928    10  07  
1929    15  08  
1930    14  12  
1931    15  11 
1932    11  09  
       420  256  

Williams lasted for a very long time, and for the last decade was a less than full time pitcher. If we compensate for the fact that he was pitching in some mixture of majors and minors by taking away, say, his last four years, the sum would be 366-217. This is right on Alexander's record.

I'm still taking this as the result of taking several very uncertain things (IP, run environment, and "league" strength) and interpreting them mostly in Williams's favor. I accept that it is possible that Williams was better than Alexander and nearly as good as Johnson, by I'm not sure that it's likely that that's the right estimate. I will probably continue to estimate him as behind Alexander - but far, far, ahead of the "mere mortal" candidates like Waddell, Ruffing, or Vance.

I said something earlier about the Johnsons having the best all-name team. Probably yes - but the Williamses do have a top-end pitcher to go with their superior outfield.
   24. Chris Cobb Posted: August 21, 2004 at 01:48 PM (#810869)
Thank you, OCF, for this analysis!

On the Williams-Alexander comparison: since Alexander missed 1 1/3 prime seasons due to WWI, my view is that the best-case analysis of Williams, which you've given us here in accurate form, definitely leaves Alexander slightly ahead. (That is, I think 1 1/3 more seasons of Alexander at his peak would put him ahead of Williams even if one keeps his four final seasons.)

I've started to work on Williams' raw data: I hope to have a coherent post sometime over the weekend.

For now, let me throw out a few questions about analyzing it that are on my mind:

1) It's my sense that for pitchers generally, there is a tradeoff between innings-eating and per-inning quality. Negro-League pitchers worked on a different pattern than their major-league counterparts, which might lead to a different balance between the two kinds of pitching contributions. Negro-Leaguers were pitching quite a few more games that major-league pitchers were, so they may have been pitching at less than their optimal effectiveness. How ought their records to be adjusted to translate them into major-league values? If one accepts the evidence of durability at face value, they look _very_ durable in relation to their major-league counterparts, but not particularly impressive in terms of effectiveness. That's what i9s is showing us, I think. But is that right? Should we cut the innings but assume that the ERA+ would be a bit better than a straight translation would produce?

2) What evidence do we have about run-environments in the Negro Leagues in comparison to the major leagues? My method (and OCF's experiment at a more refined treatment of it above) assumes equivalent run environments. I'm near certain that i9s, because they are preparing records suitable for simulated games, are creating their projections to fit contemporary major-league run environments, so when we use their projections to calculate ERA+, we are trusting that they have made the correct translation from Negro-League environments to major-league ones. Is there data that we could use to check their projections in this respect, or to try to create ERA+ values for pitchers with respect to Negro-League competition?

3) How would it be best to estimate the ratio between runs and earned runs in working with Negro-League data, and what bearing, if any, should that ratio have on assessment of the pitching performance of Negro-Leaguers? In OCF's study above, he applies a standard ratio between RA+ and ERA+. Since the earned runs allowed he's using are major-league projections, it's correct for him to use a standard major-league ratio.

In working with the Negro-League data, however, we usually have (from Holway, at least), Runs Allowed Average. In working with this data, should we assume that the ratio between earned runs allowed and runs allowed is the same in Negro-League play as in contemporary major-league play? That seems to me a shaky assumption, since the overall quality of play was lower, as was quality of equipment and quality of playing conditions. One might decide, then, to project the Negro-League pitchers as having somewhat lower ERAs in relation to their RAs than major-league pitchers (prior to making a translation for level of competition, of course). On the other hand, one might infer that the disadvantages of lesser fielding were counterbalanced by advantagegs derived from looser rules about treatment of the baseball, pitching motion, etc., that gave advantages to Negro-Leaugue pitchers that they would not enjoy under major-league conditions. Given these differences, the safest thing to do might be to assume that pitchers in the Negro Leagues had exactly the same degree of responsibility for runs scored against them as their major-league counterparts, and leave it at that. Obviously, having a run-environment for the Negro Leagues would help us to deal with this problem, also.

Well, that's more than I thought I had to ask. But since there are folks joining in the discussion now with quite a bit of knowledge about the history of the Negro Leagues, I'm hoping that we'll be able to come up with some answers.
   25. Chris Cobb Posted: August 21, 2004 at 06:34 PM (#811222)
Here are the published numbers on Smokey Joe Williams from Riley and Holway.

Smokey Joe Williams Data

<u>From Riley</u>

1905-1909 (in Texas play) 28-4, 15-9, 20-8, 32-8

1911 – 10-7 in Cuban play; 22-15 lifetime in Cuba
1912 – 9-1 for Chi American Giants against PCL teams; 8-2 for NY Lincoln Giants
1913 – 18-3 for NY Lincoln Giants
1914 – 41-3 vs. all levels of competition; 12-2 vs. black teams, 100 K in 17 games, all for NY Lincoln Giants
1921 18-2 “through mid-August” for NY Lincoln Giants
1929 12-7 for Homestead Grays

20-7 vs. major-league competition, lifetime

1912-1914 Hit .320, .299, .175
1917 hits .281

<u>From Holway</u>

1910 3-4 for Chicago Giants, TRA 3.12 (4th in West – 4 teams, top TRA 2.00)
7 decisions out of 13 recorded for team.
1911 1-0 for Chicago Leland Giants (of 5 team decisions)
10-7 in Cuban play (Frank Wickware 10-4, Jose Mendez 9-5, Dick Redding 4-8)
1912 3-2 for NY Lincoln Giants, TRA 2.20 (9 team decisions)
9-7 in Cuban play (Dick Redding, 7-2, Jose Mendez 7-5)
2-0 vs. major-league competition (beat a Giants/Yankees mixed squad, 6-0, 6-0). May have won a third game against the Giants, 2-0
1913 12-5 for NY Lincoln Giants, TRA 3.87 (23 team decisions); Holway all-star, Holway George Stovey award
1-0 vs. major-league competition (beat Phillies & Pete Alexander, 9-2)
1914 6-4 for NY Lincoln Giants, (29 team dec.)
1-0-1 vs. major-league competition (beat Phillies, 10-4, tied Rube Marquard & a part-Giants team 1-1)
1915 2-0 for NY Lincoln Giants (5 team dec.)
Missed time due to broken arm and broken wrist
2-2 vs. major-league competition (beat Buffalo of Federal League, 3-0, lost to NY Giants, 2-4, beat the Phillies (less Gavvy Cravath), 1-0, lost to the Phillies, 2-4)
3-3 in Cuban play (Dolf Luque 12-5, Frank Wickware 2-4)
1916 5-6 for NY Lincoln Giants (19 team dec.), TRA 4.60
1-0 vs. “Major-League” competition – beat Rube Marquard, backed by a minor-league squad, 5-4.
2-1 in resort ball in Florida
1917 9-1 for NY Lincoln Giants, 3.22 TRA (of 14 team decisions); Holway All Star, George Stovey award, Fleet Walker award
1-1 vs. major league competition (beat Joe Bush & a weak, part big-league squad 6-2, lost to Bush and a stronger squad 4-10) .
1918 7-2 for NY Lincoln Giants, TRA 2.23 (19 team decisions); Holway all star, George Stovey award, Fleet Walker award
1-0 vs. major-league competition (won by forfeit while trailing 3-4 in the ninth), 4-0 vs. big-league pitchers backed by minor-league squads (8-0, 2-0, 1-0, 14-1)
1919 9-2 for NY Lincoln Giants, 2.32 TRA (of 20 team decisions); Holway all-star, George Stovey award
1920 0-3 for NY Lincoln Giants (two losses to Dick Redding; Redding pitched shutouts in both) (4 team decisions)
1921 7-1 for NY Lincoln Giants (17 team decisions); Holway all-star, George Stovey Award
1922 4-1 for NY Lincoln Giants (12 team decisions), 5.40 TRA
1923 5-4 for NY Lincoln Giants (38 team decisons)
1924 3-4 for Bkn Royal Giants (35 team decisions)
1925 2-2 for Homestead Grays (? Team decisions)
1926 1-0 for Homestead Grays (6 team decisions)
1-0 vs. major-league competition (beat AL all-star team (good team!), 6-5)
1927 2-0 for Homestead Grays (7 team decisions)
1-0 vs. major league competition (beat good AL all-star team, 5-0. Joe’s pitching was obviously good; opposing pitcher Rube Walberg was pitching his second game of the day)
1928 2-1 for Homestead Grays (13 team decisions)
1-0 vs. major-league competition (beat good AL all-star team, no box score, but Holway says he and his opponent, Rube Walberg, were “hit hard.”)
1929 8-2 for Homestead Grays (58 team decisions), 4.76 TRA
1930 7-2 for Homestead Grays (18 team decisions), 3.00 TRA
1-2 in playoff vs. Lincoln Giants (2-6, 11-3, 2-6)
1931 10-6 for Homestead Grays (65 team decisions), 2.54 TRA
0-1, 6.57 TRA in “World Series” against KC Monarchs
1932 6-2, 3.29 TRA for Homestead Grays (82 team dec)

Holway’s listed career totals
107-57, .652 wp
8-4 vs. major-league competition.

From Holway’s listed seasonal records, I count
114-52, .687 wp

10-4-1, .700 wp vs. major-league competition (counts forfeit win as a loss, doesn’t count victory against Rube Marquard & a minor-league squad, does count victory vs. Federal League team, calculates the tie as 1/2 win)

Total score in 13 games with box-scores: 64-36, so Williams’ record is pretty much in line with expectation, given the scoring.

In the 11 seasons for which Holway lists a total run average for Williams, he pitched 107 games. Treating each game as complete, Williams' career TRA comes out at 3.47. (4 of the 11 seasons are the last 4 of Williams' career.)

Don’t know why there’s a discrepancy between Holway’s seasonal records and his career records, but there it is.
   26. andrew siegel Posted: August 22, 2004 at 05:42 PM (#812357)
I will move this to the 1934 Discussion Board when there is one, but I had it done and wanted to put it out there now:

The onslaught of top new candidates in 1934 has made me take a look back at the full slate of candidates we have thus far vetted. In my books, we've done a wonderful job in separating the wheat from the chaff. I've only got three discrepancies in my retroactive personal HoM with Cupid Childs at number 59 the highest rated player we have thus far missed. One of the three guys you've elected that I would have skipped will eventually make my HoM (Sam Thompson) and the other two are folks on whom the limited evidence makes a wide variety of rankings plausible (Foster and Pearce).

I am going to add a new feature to my ballot every year, giving the ranking among all the players we have seen so far of every player who makes the top 100. To do that, I am going to slot the new players in every year and review the overall list every 10 years to make any modifications I find necessary.

For those who care, this is my view of the Top 100 candidates through 1934. (Pretend that I'm publishing the Andrew Siegel Historical Abstract in 1934 and not considering anyone not yet eligible for the HoM).

1) Wagner
2) Johnson
3) Cobb
4) Young
5) Collins
6) Lloyd
7) Speaker
8) Lajoie
9) Williams
10) Matthewson

Note, that's numbers 3,5,6,7, and 9 eligible for the first time in 1934-wow!

11) Nichols
12) Brouthers
13) Connor
14) Delahnty
15) Anson
16) Davis
17) Dahlen
18) Ewing
19) Clarkson
20) Walsh

21) Crawford
22) Torriente
23) Hamilton
24) G. Wright (the first non-1st ballot guy on my list)
25) Rusie
26) Baker
27) Home Run Johnson
28) Jackson
29) Hines
30) White

31) O'Rourke
32) King Kelly
33) Barnes
34) Plank
35) Santop
36) McVey (probably my biggest discrepancy with many of you)
37) Burkett
38) Hill
39) Gore
40) Brown

41) Clarke
42) Glasscock
43) Bennett
44) Wheat
45) Collins
46) Flick
47) Ward
48) Spalding
49) McPhee
50) Sheckard

51) Richardson
52) Radbourn
53) Wallace
54) Sutton (I jumped the gun on him a bit)
55) Joe Kelley
56) Galvin
57) Stovey
58) Keefe
59) Childs (the top reurning player from my ballot, personal retroactive HoM)
60) Grant

61) Start
62) Van Haltren (personal retroactive HoM)
63) Keeler
64) McGinnity
65) Jennings (returning, personal retroactive HoM)
66) Caruthers
67) Coveleski
68) Magee (last player in thus far)
69) Chance
70) Duffy

71) Pike
72) Groh
73) Thompson (not in personal retroactive HoM)
74) Beckley
75) Ryan
76) Bresnahan
77) Doyle
78) Charley Jones
79) Willis
80) Veach

81) Welch (n.b., only 23 places behind Keefe but still off my ballot while Keefe is still in my personal HoM)
82) Dunlap (very underrated)
83) Williamson
84) Ben Taylor (tentative)
85) McGraw
86) Cooper
87) Griffin
88) Mendez
89) Monroe
90) Browning

91) Griffith
92) Rube Foster (not in personal HoM)
93) Poles
94) Tiernan
95) Leach
96) Cravath
97) Shocker
98) Cicotte
99) Vaughn
100) Konetchy

Dickey Pearce at number 106 is the only HoMer out of my top 100.
   27. TomH Posted: August 23, 2004 at 02:18 PM (#813641)
Andrew, your list is so close to mine that it's very scary. We match top 5 precisely, and thru top 17 we have no one out of place by more than 3 positions. I'll have to let you proxy vote for me in the future if I miss an election :)
   28. KJOK Posted: August 25, 2004 at 06:16 PM (#818325)
KJOK, I've been meaning to ask -- Do you know what percentage of the recorded Grays games Williams pitched that year?

Are we looking at 3 starts and one relief appearance among 20 recorded team games, or 40?

Sorry, I missed this question somehow. It's 3 starts and 1 relief appearance in the 19 games we have boxscores.

The Grays played about 130 other games that year, and not sure how many Williams pitched, but seems as if Oscar Owens and Sam Streeter did the majority of the pitching.
   29. Chris Cobb Posted: September 04, 2004 at 06:49 PM (#837165)
Smokey Joe Williams Analysis

There are three approaches available to developing an MLE for Williams, given what we know at present about the relation of quality of competition Negro Leaguers faced to the quality of major-league competition. They are

1) study of his won-lost record versus major-league competition

2) study of his won-lost record versus black teams

3) study of his runs allowed averages.

I’m going to look at the first two of these; I’d like to get to the third, eventually, but I won’t manage it before the balloting begins for 1934.

1) Williams versus major-league competition

Reported 22-7 for career

10-4-1, .700 wp vs. major-league competition in games with box scores or other newspaper records (counts forfeit win as a loss, doesn’t count victory against Rube Marquard & a minor-league squad, does count victory vs. Federal League team, calculates the tie as 1/2 win)

Total score in 13 games with box-scores: 64-36, so Williams’ record is pretty much in line with expectation, given the scoring.

Williams runs allowed average in those games is 2.77

Williams’ winning percentage against actual major league teams is impressive, but the teams he beat generally were not. Unlike Mendez, who pitched mostly against championship-calibre teams touring Cuba together, Williams tended to pitch against cobbled-together squads, often made up largely of marginal major leaguers. He did pitch against some good teams, and the all-star teams he faced late in his career with the Homestead Grays had considerable talent, but I estimate that the combined winning percentage of the teams Williams faced against major-league competition would have been about .450.

In contrast, the black teams assembled to play against major-leaguers tended to be the strongest of the black teams, and they were often supplemented by other black stars. No easy estimate is available, but I think Williams’ teams ought to be looked at, collectively, as .550 teams at against major-league average competition. Normalizing Williams’ record to major-league average opposition and major-league average support, his winning percentage drops to .600 (if one counts the tie as a half win).

That winning percentage pythags out to an ERA+ of about 122.5; using Pythaganport to adjust for the run environment might lead to a slightly different result.

I believe this is a conservative estimate of Williams’ effectiveness in these games. I feel like my assessment of the quality of the opposition is pretty good, but it’s possible that my estimate of the quality of Williams’ support is too high. It’s certainly significantly higher than the quality of Williams’ average support during his career. It definitely _was_ higher, but it’s hard to say by how much.

Next part to follow shortly.
   30. Chris Cobb Posted: September 04, 2004 at 07:14 PM (#837209)
Smokey Joe Williams analysis, continued

2) Wiliams versus black teams

115-54, .680 wp, in recorded games, career.

If one assumes that, for his recorded games, Williams’ support was league average, one can calculate an ERA+ from Williams’ Negro League play and apply the standard .87 conversion multiple. That yields an ERA+ of 146 for Williams against Negro-League competition and an ERA+ 127 MLE. An ERA+ of 127 pythags to a wp of .617. This is still close to Williams’ support neutral winning percentage vs. major-league competition, but is a bit higher.

Reliance on Williams’ recorded career record in this way has two major drawbacks.

1) The assumption that Williams’ support was average could be wrong.

2) The uneven documentation of Williams’ career could lead to his better years or his worse years receiving undue weight in the analysis.

I’m going to examine each of these issues in turn.

1) Team support, offensive and defensive

On the whole, Williams’ teams were probably slightly better than average vs. Negro-League competition, but their calibre varied over the many years of his career. I studied the records of Williams’ teams in games not pitched by Williams, using Holway’s data, and the shape of this career looks like this.

1910-11 – With Chicago Leland Giants. Average teams
1912-16 – With New York Lincoln Giants during a championship run. Good teams.
1917-22 – With New York Lincoln Giants. Team thrives, but is below .500 without Williams, who pitches a majority of teams’ games against other top black teams.
1923-24 – With Lincolns and Brooklyn Royal Giants. Bad teams, sub .400 wp without Williams pitching. He manages .500 record.
1925-28 – With Homestead Grays as a great team is being built. Because the Grays are outside league, they play few games vs. black teams, but they win most of them, and they have success vs. major-league opponents also. Good to very good teams, but pitching was probably their greatest strength.
1929-32 – With Grays as they join regular league play. Very good teams,would probably have been .600 or better vs. Negro League competition with average pitching. (They were considerably better than that, because their pitching was great, too.)

Considering team seasons without accounting for Williams’ usage patterns, I estimate Williams’ career support was about .506 . If Williams’ usage was evenly distributed across his career, that would indicate a career ERA+ of about 118.

2) Adjusting usage. To weight both Williams’ recorded wins and his won-lost record properly, they need to be adjusted for his usage.

I don’t have any method for estimating major-league equivalent usage. Therefore, I am, as in the past, relying on i9s for usage estimates. If anyone has suggestions for ways to improve usage projections, I’d love to improve this part of the analysis!

I prorated Williams’ winning percentage from each of the above stages of his career according to his innings pitched in each part. This calculation drops his career wp to .645. (The reason why his documented wp is higher is this: In his last four years with the Grays, his record is very well documented, and he had great support in those years. When the rest of his career is its due weight, his winning percentage goes down.)

However, when I prorated his team support in the same manner, it dropped from .506 to .499, because Williams pitched more innings in the earlier seasons with poorer support. Thus, one can take his usage-adjusted .645 winning percentage as very close to support neutral. It pythags out to a Negro-League ERA+ of 135.

This ERA+ can then be translated into an MLE by the usual method, yielding an ERA+ of 118, which pythags out to a .582 support-neutral winning percentage.

Study of Williams’ records against major-league competition and against black teams thus offers two different projections of Williams’ career ERA+: 118 (from Negro-League competion) and 122 (from direct competition vs. major-leaguers).

The 118 value draws on a much larger sample, but the 122 was achieved against actual major-league competition, so there is less uncertainty in the conversion factor. The ERA+ against actual major league competition leaves out the beginning and ending of Williams’ career, however. I think splitting the difference and setting Williams’ career ERA+ at 120 seems reasonable. This implies, by the straight Pythgorean method, a career support-neutral winning percentage of .590. (If one wants to cut off the latter seasons from Williams’ career, figuring his major-league equivalent career would end in 1928, then one should raise his career ERA+ in consequence.)

Applying that winning percentage to Williams’ career ip as projected by i9s, and assuming 1 decision per 9 ip, Williams’ record comes out to 399-278. I’m guessing that, given the low run-environment from the first half of Williams’ career, pythaganport projections of his record from his ERA+ would be lower, something like 377-300.

If you want to knock off his last four seasons, taking 851 innings of league average pitching off his record, it would come out to about 330-253.

In an ordinary year, this value would easily make Williams a first-ballot HoMer. Looking through baseball history for a good comp, I’d say Bob Feller without time missed for World War II and with more durability in the latter part of his career is the best I can find. James, fwiw, has Williams at #52 all time, Feller at #56.

Based on this analysis, Williams will be at #5 on my ballot in 1934. If he’s not elected in 1934-35, he’ll be #2 on my ballot behind Pete Alexander in 1936.

I haven’t done a season-by-season win share estimate for Williams, but I think he can be placed reasonably well on the basis of innings pitched and estimated ERA+. Eventually I may do WS estimates, if I can find the time.
   31. Dr. Chaleeko Posted: May 22, 2005 at 10:33 PM (#1355006)

I guess this represents a little less than half of Williams's career, and, like Redding, it's not a particularly well document phase either.

WINS t-40th with 57

LOSSES t-91st with 29

DECISIONS t-55th with 86

(50 Decisions Minimum) 17th
(25 Decisions Minimum) 28th
(10 Decisions Minimum) 50th

(50 Decisions Minimum) 30th
(25 Decisions Minimum) 42nd
(10 Decisions Minimum) 66th

WAT 31st at 7.3

(50 Decisions Minimum) 25th
(25 Decisions Minimum) 41st
(10 Decisions Minimum) 70th


1921 3rd in East with 7 wins.

1922 4th in East 4 wins.

1923 6th in ECL with 5 wins.

1929 t-7th in ANL with 8 wins.

1930 4th in East with 7 wins.

1931 3rd in East with 10 wins, 5th in NgLs.
   32. OCF Posted: April 03, 2009 at 05:19 PM (#3123316)
Now that Williams is in the top group of our current ranking exercise, I felt like bumping his thread. I'm surprised at how active a participant I was in this thread - I don't remember it all that well.

Although official birth records do not exist, Williams is believed to have been born on 6 April 1886 in Seguin, Texas, to an African American father and Native American mother. (His mother was reported to be a Cherokee, although this has not been confirmed.)

I know something of the area, because my wife was born and raised in New Braunfels. New Braunfels and Seguin are only about 15 miles apart and are county seats of adjacent counties. And, going all the way back to their pre-Civil War foundings, they couldn't be more different, and the differences persist to the present.

New Braunfels is located along the Balcones Escarpment - the edge of the Texas Hill Country, which is itself the easternmost part of the massive block of limestone called the Edwards Plateau. Seguin is located in the flatlands out away from the escarpment. If you look to each location with a 19th century eye, the biggest natural advantage New Braunfels had was water - plentiful spring water, water power. The biggest natural advantage Seguin had was fertile farmland.

New Braunfels was settled in the 1840's by a society that promoted emigration from Germany to the new world, and grew rapidly into a town of overwhelmingly German character. (It was well into the 20th century before German language newspapers faded out and the schools stopped seeing kids who spoke only German at home.) Seguin is slightly older, and although named for a Tejano who spoke only Spanish, it became a center for migrants from older parts of the South who brought their slaves with them. In the 19th century, Seguin was the center of an agricultural area; New Braunfels was (at least in 19th century terms) more industrial. The Germans farmed, too, but were less focused on commodity crops like cotton. Seguin grew cotton, of course, but claims the pecan as its signature agricultural item.

Because of its slave-holding history, Seguin has always had a sizable African-American population, although over the years, much of that population has migrated away to big cities. The German settlers of New Braunfels generally didn't own slaves, and the African-American population of that town has always been miniscule.

As for Williams's ancestry: Cherokees are not native to Texas, but there was some significant migration into the area - at least into East Texas - prior to the forced relocation of most of the tribe to Oklahoma. I think there's quite a bit of history of Sam Houston having negotiations and dealings with Cherokees. Seguin isn't East Texas, but people do move around.
   33. KJOK Posted: September 17, 2011 at 07:43 AM (#3927882)
   34. KJOK Posted: September 17, 2011 at 04:56 PM (#3928076)
   35. KJOK Posted: September 17, 2011 at 08:15 PM (#3928161)
OK, now that this morning's player indexing hiccup is fixed, let's try again:

Joe Williams' Real Stats
   36. Alex King Posted: January 17, 2012 at 06:14 PM (#4038936)
I was poking around the Negro Leagues database and I found that shockingly, Williams had a 147 OPS+ over 482 PA in the Negro Leagues. This would probably translate to a better hitting record than that of Wes Ferrell, generally regarded as among the best-hitting 20th century pitchers, if not the best (excluding Ruth obviously). Was Williams considered a good hitter reputationally? Did he bat high up in the lineup? If Williams was really as good as his stats suggest, his batting would have added significant value throughout his career, more so even than someone like Walter Johnson (76 OPS+).
   37. Gary A Posted: January 19, 2012 at 04:20 PM (#4040631)
No, I don't think Williams had a reputation as a great hitter at the time or subsequently, despite his evident performance. Like almost every pitcher he did some duty in the outfield or at first base, but he wasn't a regular when he wasn't pitching, unlike, say, Rogan, Dihigo, Luther Farrell, or other two-way players. And he usually batted at the bottom of the order--maybe as high as sixth when he wasn't pitching.
   38. Dr. Chaleeko Posted: October 21, 2017 at 11:37 AM (#5559523)
Hey, everyone,

Please find my latest MLE for Joe Williams here. The MLE is articulated in an article linked therein.

You must be Registered and Logged In to post comments.



<< Back to main

BBTF Partner

Dynasty League Baseball

Support BBTF


Thanks to
cHiEf iMpaCt oFfiCEr JE
for his generous support.


You must be logged in to view your Bookmarks.


Page rendered in 1.0327 seconds
62 querie(s) executed