|
|
Hall of Merit— A Look at Baseball's All-Time Best
Tuesday, January 11, 2005
Willie Foster
I’m looking forward to see what the numbers say about him.
|
Support BBTF
Thanks to Dock Ellis for his generous support.
Bookmarks
You must be logged in to view your Bookmarks.
Hot Topics
Reranking Pitchers 1893-1923: Ballot (4 - 4:30pm, Oct 02)Last: cookiedabookieReranking Shortstops: Results (7 - 8:15am, Sep 30)Last: kcgard2Reranking First Basemen: Results (8 - 4:22pm, Sep 21)Last: Chris CobbReranking Pitchers 1893-1923: Discussion (38 - 7:19pm, Sep 20)Last: DL from MNReranking First Basemen: Ballot (18 - 10:13am, Sep 11)Last: DL from MNReranking First Basemen: Discussion Thread (111 - 5:08pm, Sep 01)Last:  Chris Cobb2024 Hall of Merit Ballot Discussion (151 - 6:33pm, Aug 31)Last:  kcgard2Hall of Merit Book Club (15 - 6:04pm, Aug 10)Last: progrockfanBattle of the Uber-Stat Systems (Win Shares vs. WARP)! (381 - 1:13pm, Jul 14)Last:  Chris CobbReranking Shortstops Ballot (21 - 5:02pm, Jun 07)Last: DL from MNReranking Shortstops: Discussion Thread (69 - 11:52pm, Jun 06)Last: GuapoCal Ripken, Jr. (15 - 12:42am, May 18)Last: The Honorable ArdoNew Eligibles Year by Year (996 - 12:23pm, May 12)Last:  cookiedabookieReranking Centerfielders: Results (20 - 10:31am, Apr 28)Last: cookiedabookieReranking Center Fielders Ballot (20 - 9:30am, Apr 06)Last: DL from MN
|
|
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. John (You Can Call Me Grandma) Murphy Posted: January 11, 2005 at 01:55 AM (#1071126)1923 Bill Foster (his rookie season)
NNL Memphis Red Sox / Chicago American Giants
W-4
L-2
SV-0
ERA-2.42
G-10
GS-8
CG-5
SHO-1
IP-52
H-32
HR-1
W-14
K-38
This is mostly for Memphis; for Chicago he pitched 8 innings with no decisions.
NNL Chicago American Giants
*-led league
W-15 (2nd)
L-10
SV-3 (tied for 2nd)
G-31*
GS-24*
CG-23*
SHO-3 (tied for 2nd)
IP-230*
H-217*
HR-11 (2nd)
R-92
W-73*
K-140*
HB-9
SH-31
SB-6
DP-15
Just for fun, I made up a rudimentary metric to measure stolen bases allowed per opportunity, taking taking stolen bases allowed divided by (hits+walks+hit batsmen minus home runs and triples). Foster's rate is .022; Chicago's team rate was .062, and the NNL rate was .071. These are the best pitchers among ERA (or TRA) qualifiers, with team average in parentheses:
Army Cooper, KC, .018 (.042)
Bill Foster, Chicago, .022 (.062)
Bullet Rogan, KC, .024 (.042)
Luther McDonald, StL .032 (.068)
J. Burdine, Bir .034 (.093)
The worst:
Reuben Currie, Det .114 (.062)
Ted Trent, StL .120 (.068)
William Nash, Bir .157 (.093)
Robert Poindexter, Bir .167 (.093)
Heliodoro Diaz, CSW .176 (.121)
Cooper and Foster are both lefties; Rogan famously had no windup. McDonald's righthanded, Burdine's unknown. Four of the bottom five are righthanders (Nash is unknown, but probably RH), and two of them (Trent and Diaz) are notable power pitchers.
Willie Powell, Chi 2.50 (144 ip)
Satchel Paige, Bir 2.93 (126 ip)
Ted Trent, StL 3.16 (205 ip)
William Bell, KC 3.56 (146.7 ip)
Bill Foster, Chi 3.60 (230 ip)
1924 6-0 for Chi Am Giants; 1-0 for Memphis Red Sox; Chi 63-28, Mem 26-36
1925 6-0 for Chi Am Giants; team 39-37; all-star
1926 11-5 for Chi Am Giants; TRA 2.03 1st, 77 K first; team 49-25; all-star
2-2 in playoff vs. KC (concluded with a double-header shutout, beating Rogan both games) 0.84 TRA
2-0, 2.57 TRA in World Series vs. AC Bacharachs
1927 21-3 for Chi Am Giants; wins 1st, TRA 2.67 3rd, .875 wp 1st, 106 K 1st; all-star, GSA (Cy Young); team 54-28
2-0, 2.00 TRA in playoff vs. Birmingham
2-2, 3.38 TRA in World Series vs. AC Bacharachs
6-8 in Cuban Play
1928 14-10 for Chi Am Giants; wins 3rd, 118 K 1st; team 46-37
2-2 5.00 TRA in Playoff vs. St. Louis
1929 15-10 for Ch Am Giants; wins 4th; team 48-25
1-0 in post-season tournament against Homestead; 0.00 TRA
2-1, 1.42 TRA vs. major-league competition
1930 16-10 fro Chi Am Giants; wins 2nd, TRA 3.71 5th, 134 K 1st; team 48-52
1-0 vs. major-league competition; TRA 1.00
1931 12-2 for Homestead Grays, 1-1 for Chi Am Giants; 1st in wins, wp .857 1st, 3.05 TRA 3rd, 111 K 1st (in 109 ip); Grays 46-19; all-star, GSA
3-0, 3.24 TRA vs. Monarchs in “Championship Series”
1-0 vs. major-league competition, 3.00 TRA
1932 19-8 for Chi Am Giants; 1st in wins, .704 wp 1st, 2.67 TRA 3rd; 73 K 1st; all-star, GSA; team 62-31
2-0, 3.50 TRA in playoff vs. Nashville
1933 12-7 for Chi Am Giants; 2nd in wins, 3.08 TRA 3rd, 52 K 3rd; team 36-18
1934 6-5 for Chi Am Giants; 2.92 TRA 4th, 49 K 3rd; team 30-20
also pitched for KC Monarchs in the Denver Post tournament
1-2, 2.00 TRA in playoff vs. Philadelphia
1935 4-2 for Chi Am Giants; team 18-24
1936 3-2 for Pgh Crawfords; team 39-27
1937 5-4 for Chi Am Giants; team 25-14
0-0, .260 TRA in playoff vs. KC
1-1 in World Series vs. Eastern all-star team (only win for West team in 9 games)
career totals
152-69, .688
146-66, according to Holway (why his career numbers don’t match his seasonal numbers in Holway’s book I have no idea)
team record 603-385, .610 wp (not counting stint with Memphis in 1924 & Chi in 1931)
teams without Foster 453-317, .588
Foster 22 wins above team
Black Ink 49, Gray Ink 97
18-9, .667 in post-season competition
4-1 vs. major-league competition (Holway has 3-0, career)
6-8 in one season of Cuban play
---
because it's Holway....
Keep in mind, as you read the following, that these are estimates. They are based on a number of interpretations of Foster's career that I think are reasonable. Unlike my MLEs for hitters, they are not based on a fairly strict conversion of available statistics according to pre-established conversion rates. The margin of error is thus considerably greater, in both directions.
Bill Foster, Career Estimate
3029 innings pitched, 125 support-neutral ERA+ (DERA 3.60)
Support-neutral w-l record, estimating 1 decision per 8.7 ip
212-136
Career ERA+ was derived by estimating the quality of the offenses and defenses that Foster pitched for, and then using the pythagorean method to calculate Foster’s contribution to his career wins. I started using this method when I was working on MLEs for Joe Rogan (which I’ve finished but never written up cleanly – it contains no surprises), so I started off comparing the offense of the Chi Am Giants to the KC Monarchs from 1923 to 1928, with suitable park adjustments. Then I continued, making my best guesses. Adding these up and prorating them to Foster’s # of decisions in each season, I calculated his team’s offense as having a RS+ of 108.7 for his career. The Chi Am Giants were always known for pitching and defense, so having nothing else to go on, I set their defensive RA+ at 120 for Foster’s career. I then selected a 4.5 r/g environment as fairly close to Foster’s career average conditions, and used the pythagorean method to find the RA+ needed for Foster to reach his career winning percentage of .688 (giving fielders 1/3 of credit for runs saved and the pitcher 2/3 credit). That shows Foster with an NeL support-neutral ERA+ of 147, which, using .85 as the conversion factor to major-league estimates for pitchers (between the batting and slugging conversion factors for hitters), leads to an estimated support-neutra major-league ERA+ of 125.
The innings pitched estimate is derived from two things: an estimate, based on Foster’s decisions year by year, of whether he would have been used as a #1, #2, #3, #4, or #5 starter in the majors (or have pitched but not as a regular starter), then calculating the average innings pitched by the pitchers in that rotation spot for each season and taking that as the estimate of Foster’s ip. For years when he was not in the rotation, I prorated his decisions to 154-game seasons and multiplied by 8.7.
Here are my estimates of Foster’s seasonal usage:
#1 Starter 29-33
#2 Starter 26-28
#3 Starter 34
#4 Starter 37
#5 Starter 35
Part-time pitcher 23-25, 36
In Foster’s 4 seasons as a part-time pitcher, his decisions pro-rate to 46, giving him 400 innings pitched for those years. (I give him full major-league credit for 23-25 because it is clear that he was a highly effective, and highly desired pitcher [given his half-brother’s efforts to pry him away from the Memphis Red Sox], but as he was going to college, he did not pitch full seasons.)
In Foster’s 2 final years in rotations, 35 and 37, #5 and #4 starters averaged 139 and 187 innings pitched, respectively, giving Foster 326 innings pitched for those years.
In the heart of Foster’s career as a top starter, 1926-1934, the averages for #1, #2, and #3 starters in the relevant seasons adds up to 2203 innings pitched for those 9 seasons.
This is a conservative estimate for Foster’s pitching in those years. Here are the major-league innings pitched leaders for 1926-34, as far as I have been able to determine.
2313.3 Lefty Grove
2177.3 Freddie Fitzsimmons
2172 Earl Whitehill
2153.3 Ted Lyons
2137 Charlie Root
Here are top 9-year stretches a few years later by other top pitchers
2530.7 Carl Hubbell, 29-37
2389.7 Wes Ferrell, 29-37
2235.7 Red Ruffing, 28-36
Foster had no injuries during this period and was at least as effective pitcher as everyone in these groups except for Grove and Hubbell, so I think an estimate of 2203 innings pitched for him is entirely reasonable and probably underrates him by as much as 200 innings pitched. Hubbell’s total shows us what a really great pitcher throwing 9 consecutive seasons without injury could do. Grove’s innings include an injury year in 1934 and Ferrell’s a lesser year in 1934 as well. So I think adding 200 innings to an estimate of Foster’s career would be justified. I have added only 100, bringing his career total to 3029, as listed.
In sum, I see Foster’s career as quite similar to Stan Coveleski’s and Dazzy Vance’s, mapping somewhere between the two in shape. He didn’t quite have Coveleski’s steadiness, but he was steadier than Vance. He doesn’t match Vance’s peak (he has only one mind-boggling peak season in 1927), but his best couple of years are more outstanding than Coveleski’s. He had a great prime, 9 years from 1926-1934 over which he was consistently one of the top pitchers in baseball, preceded by a strong part-time run of three seasons and followed by a decent three-season decline.
I think a more precise estimate based on more season-by-season study is possible, but for now, I’m very comfortable on the basis of this analysis slotting him into Dazzy Vance’s slot in my ballot.
You compare Foster to Vance and Coveleski. I guess all 3 were power pitchers, but Vance had much higher strikeout rates than Coveleski (or anyone else in the majors at the time). Do you see Foster as closer to Vance or to Coveleski in getting strikeouts?
Support-neutral w-l record, estimating 1 decision per 8.7 ip
212-136
This invites comparison to my RA+ PythPat records for the white pitchers. I've been using 9.0 IP/decision. Even though 8.7 may be closer to the truth, I'd rather use 9.0 just for comparability; hence, I'll render his record as 205-131 or 205-132. A good 3000-inning pitcher; who do I have in that neighborhood?
209-134 Stan Coveleski
201-129 Dazzy Vance
211-143 Mordecai Brown (defense-adjusted)
200-129 Rube Waddell
201-132 Babe Adams (with no adjustments)
With some others a little further away but still comparable:
210-119 Ed Walsh
---------
216-160 Clark Griffith
190-124 Tommy Bridges
209-149 Ed Cicotte
179-113 Wes Ferrell (offense-adjusted)
181-117 Urban Shocker
203-154 Dolf Luque (majors only)
So who is Foster? Three-Finger Brown? Stan Coveleski? Babe Adams?
"Will you come HOM, Bill Foster?
Will you come HOM?
Chris makes MLEs all night long.
Gary will give us some stats
OCF will make some comps
I hope none of it is wrong
Remember that Cooperstown noon time
When you got your plaque
Will lightning strike twice and Murph add you to his tome?
karlmagnus won't have your name,
Well, is that a shame?
Bill Foster, will you be in our HOM?
(Rube is there already)
Bill Foster, will you be in our HOM?"
:-D
I probably want fewer NLers than we seem likely to elect, but I too would like to get the right ones!
That's my view of the matter, though in my system they are very close -- currently they are 6 & 7 on my evolving ballot -- and in my view they are both Hom-worthy. Lundy is currently at 31. I hope that before we catch up with the present we'll have fuller data for him that will make his position vis-a-vis the in-out line clearer.
Foster vs. Beckwith (again, taking Chris's translations at face value for now) depend on your view of the Vance/Coveleski style pitcher. karl doesn't like them, so he won't like Foster. Others, like me, had them in a top 5 ballot position, so I do like him.
IP - 3,029
Supt Neutral Fibonacci Wins - 205
ERA+ - 125
Juan Marichal
IP - 3,506
Supt Neutral Fibonacci Wins - 206
ERA+ - 122
Supt Neutral Fibonacci Wins - 205
ERA+ - 125
I've seen lots of people use "looks like Coveleski & Vance" as a case for induction, that's dangerous because although I agreed that those two were the best two pitchers we could have inducted at the time, there were many close behind.
Newhouser without the war discount is a great candidate.
Anyhow, Foster probably has another year to wait, but he's well ahead of the backlog and will most likely go in next year. He's still on my ballot, but I'd just like to echo John's call for MLE's & WS estimates. I'd like to see why he has a #3 ranking on the all-time NeL pitching depth chart.
We'll probably never be satisfied that we know the answer here.
However, pending such analysis a few specifications of his value can be made, I think.
Coveleski and Vance are very likely to be his best major-league comps because of the shape of his career and the period in which he was pitching.
He is almost certainly better than Babe Adams, having a peak that is more like Coveleski's and Vance's, because Adams accumulated his career innings while seldom throwing all that many innings in a given season, and he pitched when offensive levels were much lower, so that his ip are less meaningful than they are in the 1925-1940 period when Foster starred.
He is almost certainly better than Carl Mays. My ERA+ estimate and support-neutral records estimates are adjusted, as much as is possible given the limitations of the available data, for both run support and fielding support. The level of support that I estimated for Foster fits a pythgorean projection of his w/l record, and I tried to give as much credit to batters and fielders as I reasonably could to make sure that I wasn't overestimating Foster's contributions as a pitcher. The only way that Mays looks as good as Coveleski or Vance is if these two factors are _not_ taken into account sufficiently, so I'm pretty sure that Foster was better than Mays. Even with those accounted for, Mays is not a bad candidate, of course, in part because his own abilities as hitter and fielder contributed to his own support.
In conclusion, if you think that Vance and Coveleski were mistakes, season-by-season estimates based on my career estimates are unlikely to lead you to see Foster as a much better than they were.
If you think that V & C were _not_ mistakes, I think it very likely that seasonal estimates will show Foster to be similar to them, just as the career estimate did.
--
karlmagnus to David Foss:
with the exception of Coveleski and Vance (a mistake, IMHO) we haven't elected any of the pitchers on your list, nor are we likely to (I suppose Newhouser's possible, though I don't think I shall vote for him). I suggest Foster loses on a 6-2 vote.
Back in the preceding professional wrestling epoch, there was a wrestler stage-named Killer Kowalski. I wonder whether he was a relative of Stan Coveleski.
Bob Lemon. Wasn't he the sabr-tooth tiger who took on George Steinbrenner and won? ;-) My "system" says that the HOM will approve him but that's only my system.
My problem is I had Coveleski near the top of my ballot, while I had Vance closer to the bottom. :-)
There was a thread a long time ago on this site, probably about 3-4 years ago, that had the tombstones of various players. I seem to recall that there was a discussion about Turkey Stearnes grave, which is unmarked. People here were mentioning at the time that they might try to get a collection together to put a proper tombstone on his grave. Not sure whatever happened with that.
Turkey Stearnes grave
By the way, overall I think we do a nice job of avoiding either knee-jerk sympathy or hard-ass 'don't see it in the Baseball Encyclopedia so can't elect him' thinking. These guys get a fair shake, but no free ride.
Hell, no doubt that's all any of them - all the way back to Frank Grant (who came close but never did get an elect-me spot from me) - ever would have wanted.
From The Negro League Courier, Oct 2002.
>>
A replica of Turkey Stearnes' Hall of Fame plaque now serves as a headstone for the previously unmarked grave of the Detroit Stars legend. It was dedicated August 12 and marks the first time that permission has been granted by the Hall of Fame to replicate a plaque. [. . .] the unmarked grave was reported to the world in Black Baseball in Detroit by [Dick] Clark, Larry Lester, and Sammy Miller in 2000.<<
--Dick Clark, co-Chair, Negro Leagues Cmte, SABR
Willie Bill Foster
From The Negro Leagues Courier, Oct 1998.
>>
Ernest Nagy reports that Alcorn State in Lorman, MS is beginning the process of gathering funds to erect a befitting headstone for Bill Foster's gravesite.<<
--from a Google citation! of the reprint on Jim Riley's website.
http://www.blackbaseball.com/resource/sabr1098.htm
(WARNING. I started to read that number and I was redirected to the fancy homepage before I learned more. From the homepage, I couldn't find the newsletters! If you use this url, try to copy the page ASAP, before you are redirected. Format is simple html.)
I’m going to get season-by-season projections for Bill Foster completed over the weekend. I’ll be posting these in a few separate posts as I complete each step of the projections. The first piece is establishing innings pitched projections. That’s the subject of this post. Season-by-season support-neural W-L records and DERA+ will be next; win-share projections will be last.
Innings Pitched
#1 Innings listed in MacMillan Encyclopedia, 10th edition
#2 Innings modified according to w-l data in Holway, where he lists more or fewer decisions than MacMillan
#3 Innings projected from Foster’s team’s games to 154-game seasons
#4 Total from #3 multiplied by .85. I use this multiple, set to match the BA/SA conversion factor, on the assumption that the number of innings a mature pitcher can throw will be reduced similarly to his reduction in effectiveness as a result of stiffer competition. It is probably not coincidental that this multiple, when applied to Foster’s projected games started (not shown here), tends to reduce projections of 39-42 starts in a season—unlikely by ml usage patterns—to 33-36 starts, which is a typical total of starts for #1 starters in the majors in this period.
#5 Projection adjusted for seasons outside the range of probable values. Even with the .85 factor, two of Foster’s seasons, 1929 and 1933, end with totals so far above the major-league leader in innings pitched as to be implausible. I have reduced each of these seasons by 30 innings. I have also reduced Foster’s final season by 50%, as a final season of 200+ innings pitched seems inconsistent with his decline in 1934-36
Notes
This projection raises my initial IP estimate for Foster by 142 innings, from 3029. The change is a result of season-by-season inspection justifying a stronger prime for Foster. My first estimate had been based simply on the average usage of #1 and #2 starters during this period. It’s clear that Foster’s usage exceeded the average of a #1 starter in several of his seasons, hence the higher total.
1659 ip is the highest total of recorded ip in MacMillan. Our view of Foster’s career thus rests on a larger statistical base than that of any other NeL pitcher. He was the best pitcher in the NeL when its level of competition was at its highest.
Foster’s fully adjusted IP would have him leading the majors in IP in 1927 and 1929. Had he been in a major league, he could have led his league, but not the majors, in 1928, 1932, and 1933.
During his 9-year prime, 1926-1934, Foster projects to 2554 innings pitched. This is slightly higher than the top 9-year total among ML pitchers of this era, which belongs to Carl Hubbell at 2530.7. Hubbell and Foster are the only two top pitchers, 1924-1938 to throw 9 consecutive seasons in their prime without an injury year, so I think this projection for Foster is reasonable. Had Ferrell or Grove not been injured during their 9-year primes, they would have reached a very similar total. Since we know Foster carried this load without injury, he ought to receive credit for it.
It's this sort of thing that is throwing these Nel projections odff, and making these guys look just a bit better than they actually were. We are electing too many borderline cases (Beckwith and BOTH Fosters?) and this is making us do so. Conversely, we are currently being too harsh on ML players such as Sisler, Rixey, and, yes, Beckley, who actually put up the numbers in the real world.
Your objections don't match either what I actually did or the historical record in this case.
Four points you have overlooked:
1) It's not that Foster was _resting_ when he wasn't pitching against other teams in the Negro Leagues; he was pitching games against white semipro teams, probably throwing many more innings over the course of the season than big-league pitchers were. Spalding's situation was completely different. We know the NeL teams were playing these games: we just don't typically have full statistical documentation for them. So I'm only projecting Foster back into innings he has already pitched. I think that the 3922 inning projection is a conservative estimate of the number of innings Foster _actually threw_. That number then needs to be reduced to take account of the greater strain of facing ML competition in every start, so
2) I didn't just "normalize to 154". I _reduced_ his innings by .85 in each season, as I explained in the methodology section above. This is to factor in the "additional strain on his system."
3) Pitchers tended to break down _faster_ in the Negro Leagues than they did in the majors. When I say that Foster and Hubbell are the only pitchers during this period with a nine-year run of greatness uninterrupted by injury, I am _including_ Negro-League pitchers as well as major-league pitchers in that judgment. This, coupled with point 1 above, strongly implies that, in career terms, _more demands_ in terms of durability were placed on NeL pitchers than on ML pitchers. _Fewer demands_ were made in terms of quality.
4) How would it be _possible_ for Foster to have any reputation as a great pitcher if he as a player who managed to squeeze out fewer innings than Dizzy Dean against NeL competition?
I'll stop there in order to stay polite, but the "actually put up the numbers in the real world" line is, frankly, offensive.
This statement seems to fit my historical impression of Bill Foster, as one of the top three Negro League pitchers of all time (only Williams and Paige are generally thought to have been better). I have always linked him to Carl Hubbell rather than Vance or Coveleski, and the new innings projection may lend support to this hypothesis.
I for one am looking forward to Chris' final numbers.
The New Bedford Standard, August 10, 1928:
"Wills Jackman, pitcher extraordinary of the Colored Giants of Philadelphia, and iron man of the team in constitution, ingratiated himself still further into the hearts of the fans of this city last night when he appeared on the mound at Sargent field and blanked the Frates Dairy team 4-0. As Jackman had pitched the previous day's game against Osterville, he was expecting to take a well earned rest in last night's game. On arriving in New Bedford, the visiting management was notified that Jackman was expected to pitch, announcement having been made to that effect. Rather than disappoint the good sized gathering at Sargent field, Manager Burlin White, in his stentorian voice - he doesn't need a megaphone - announced 'that Jackman has kindly consented to twirl for two innings, probably not more. There is a limit to human endurance. Everywhere the Giants play, it is announced that Jackman will pitch - sometimes without permission or knowledge of the Giants - and I am sure you will excuse the big boy after pitching two innings. Wylie will succeed him on the mound.' As a matter of fact, the hardest work that Wylie was called upon to do was a little coaching at third base, as Jackman remained on the mound throughout the full nine innings. It was the second time in successive weeks that this pitcher has appeared in in the box on consecutive days, and the ease in which he subdued the dairyman suggested that he could take his place on the hill one day after another and still more that hold his own with the opposition."
Jackman fanned ten in this game.
Osterville was th best team in the Cape Cod League that year.
The Frates Dairy team was one of the top semi-pro teams in southeastern New England. The week before they lost a 6-5 game to the Boston Braves. Rogers Hornsby's four-for-four day saved the game for the Braves.
Same here, Ron.
I am becoming increasingly convinced that we will end up having elected far too many NEL pitchers, because the conversions took figures that were very uncertain, made assumptions that were very uncertain, and treated the result as solid.
I've just looked up Campanella -- no way he was the player Wally Schang was. It's not just us, theerfore; the HOF only didn't elect 40 NELers because the committee concerned hadn't heard of 40 NElers.
It's not that Foster was _resting_ when he wasn't pitching against other teams in the Negro Leagues; he was pitching games against white semipro teams, probably throwing many more innings over the course of the season than big-league pitchers were. Spalding's situation was completely different.
karl's point about Spalding "125-20" in prorated 1875 is true, of course. But Spalding's situation was not completely different, so Chris would be conceding too much if this were a debate about Spalding and others from the 1870s. (Some Spalding data is below the line.)
I suspect that mlb teams commonly fielded most to many of their best players for in-season exhibition games, from their tiny to medium-size rosters, as long as in-season exhibitions were common. And I suppose that, in turn, was one reason in-seasons exhibitions were gradually phased out.
Regarding the 1899 Milwaukee Brewers:
Some time after May 21, "Connie Mack changed course, steering his team away from off-day exhibitions and focusing on batting practice instead. He decided his players benefited more from the respite." --more than his club benefited from the money, I suppose.
--Brian Podoll, The Minor League Milwaukee Brewers, p89, relying on the Milwaukee Sentinel
--
The Vaccaro logs (Frank Vaccaro, All Games Baseball) include those in-season exhibition games reported in the newspapers. For Boston NA1875, Vaccaro lists five games (8/2, 9/29, 10/4, 10/5, 10/23) of which Spalding pitched the first three, the fourth is unknown, and Joe Borden pitched the fifth. For Boston NA1871, Vaccaro lists 11 games of which Spalding pitched ten and one is unknown.
1875-08-02 underscores that old-time exhibition games should not casually be interpreted as 21st century games between a major league team and its AA affiliate.
- 7/24 - the last known game before 7/31: championship game pitched by Spalding v St Louis (strong team)
- 7/31 - champ game pitched by Manning v New Haven (weak)
- 8/2 - exhib game pitched by Spalding v Philadelphia (strong)
- 8/4 - champ game pitched by Spalding v Philadelphia (strong)
It looks like Spalding was saved for Philadelphia.
Karlmagnus wrote:
I am becoming increasingly convinced that we will end up having elected far too many NEL pitchers, because the conversions took figures that were very uncertain, made assumptions that were very uncertain, and treated the result as solid.
I wrote in my first data presentation on Foster in post 8 above:
Keep in mind, as you read the following, that these are estimates. They are based on a number of interpretations of Foster's career that I think are reasonable. Unlike my MLEs for hitters, they are not based on a fairly strict conversion of available statistics according to pre-established conversion rates. The margin of error is thus considerably greater, in both directions.
It would be tiresome to repeat this every time I post new data, and I assume people remember that this is the case. But clearly not.
I will add that, as the electorate well knows, we have so far elected three NeL pitchers, Joe Williams, Rube Foster, and Joe Rogan, the latter of whom was elected as much for his hitting as for his pitching. We are certain to elect Satchel Paige, and unless current trends change, we are likely to elect Willie Foster. To say that electing 5 NeL pitchers is "far too many" is typical hyperbole.
Let me say again that Foster's numbers cut both ways. If I had done no analysis at all, and we had approached Bill Foster on reputation alone, we probably would have elected him easily, and his reputation may still (possibly rightly) override any questions about his quality that my numbers produce.
If the numbers are as uncertain as karlmagnus sees them as being, they are as likely to be underestimating the merit of Negro-Leaguers as overestimating them. It is clearly prejudicial to assume that an effort to make a more precise estimate of merit must necessarily lead to an overestimation of that merit.
You're right, of course, Paul, and I made arguments on this point to justify projecting the NA players to longer seasons, back when we were debating whether or not Cal McVey had played enough games to be electable. I was concerned with the immediate issue about Foster's innings and was tempted by the hyperbole into overstating the case.
Does anybody know anything about Willie Foster's reputation/ability as a hitter?
The bio in Riley says not a word on the subject.
NNL Chicago American Giants
G-33
AB-90
H-20
D-3
T-0
HR-0
R-9
W-3
HP-1
SH-6
SB-0
AVE-.222
OBA-.255
SLG-.256
Not bad for a pitcher, I guess, considering the park, but nothing special.
The HOF didn't want to elect any to begin with; then they wanted to put them in a separate wing. Only after pressure from people like Ted Williams did they agree to induct NeLers on an equal basis with everybody else. When they did, the Negro League Committee picked a semi-symbolic team of 9 players through 1977, then closed up shop.
Btw, Roy Campanella was certainly not elected to the HOF for his Negro League career. Most voters at the time were probably barely aware he had one. The same goes for Jackie Robinson, who was cited as one of the deserving NeLers. His Negro League career, such as it was (he played a single season for the Monarchs), had nothing to do with his HOF induction.
Starts playing in NeL at age 15. At 19, he is a starter and an all-star for the first time. The next two years, he is touring and playing in the Mexican League.
At ages 22 and 23, he is an all-star catcher again.
At age 24 he is the league MVP for Nashua while playing in the Dodgers' system.
At age 25 he is the league MVP for Montreal while playing in the Dodgers' system.
Reached the majors at 26.
All-star at ages 27-34. Probably 2 were on reputation in the off years (54 and 56). Oh, and 3 MVP awards.
I have exercised judgment and made guesses
1) when estimating innings pitched, as discussed above
2) in estimating Foster's run support and fielding support, estimates necessary to turn Foster's w-l record into an ERA+ to which a conversion factor can be applied
3) in estimating the conversion factor. I have used .85.
Let it further be noted that I have modified the system from the simpler system I applied in creating estimates for Dick Redding in two ways.
First, I have used a .85 conversion factor here. With Redding, I used .87. Redding's ERA+ and WS numbers in comparison to Foster's should be slightly reduced as a result of this change.
Second, before making a win-share estimate, I have turned ERA+ into DERA. This eliminates the overestimation of pitching win shares that was observed in my estimates for Nip Winters a while back. As you'll see, these win shares are scaled to match Bill James win shares almost exactly. For this reason, Redding's listed win shares should be REDUCED, I estimate, by 7.5%. This has nothing to do with conversions -- it has to do with finding a way to model the win shares system properly. I'll be lowering him in my rankings somewhat in consequence of this finding. I'm not sure whether Jose Mendez needs similar adjustment: his projections were based more on pitching versus major-league competition.
As I gradually work out a better system, and as we approach a more accurate estimate of conversion factors, I hope to get my data into spread sheets and re-run them for eligible and elected players, but that's a big job and time is short.
Without further preamble, I'll put the Foster data up for consideration.
From NeL w-l record to MLE DERA+
Explanations of data
NeL IP: column #2 in post above
NeL W-L: Won-Lost record as shown in Holway. A notable exception is 1923, where Holway does not have data for Foster; for this season I have used MacMillan. For 1925 and 1931, the posted record here differs from the one posted near the top of the list slightly, as I found that Foster had a few decisions for another team that I missed in my first pass through Holway.
RSI: my estimate of the percentage above league average of Foster’s team run support – it is the equivalent of Chris J’s measure, except that mine is an educated guess based on examination of team batting averages (eyeball-adjusted for park-effects) rather than on Retrosheet game scores. I have made some attempt to keep these in keeping with the range of values observed in major-league baseball. I’m guessing that for selected seasons Gary A. has the data we would need for true projections at the team level or even for Foster personally. If that data appears, I’d be happy to use it.
fRa+: my estimate of the percentage above league average of Foster’s team fielding. This is based more on team reputations than anything else. More study of both team RSI patterns and team fRA+ patterns would be advantageous to the modeling process, but since these numbers are at best educated guesses without more data, much guesswork would still remain.
NeL ERA+: Calculated by reverse-Pythagorean method. Assuming a 4.5 run/game environment, RSI and frA+ values are used to calculate Foster’s runs-allowed relative to his league necessary to achieve his w-l record. The more sophisticated pythaganpat formula might help here, but I don’t have the computational sophistication to use it. What I’m calling ERA+ here probably does not scale exactly against major-league ERA+ values, which I increasingly see as of dubious reliability because of the interaction of fielding value with pitching value in the data used to calculate them. In other words, use these ERA+ numbers in comparison to other ERA+ numbers at your own risk.
Projected ERA+: NeL ERA+ multiplied by .85 . Although the info from Gary A. on Herrera is not enough to firmly establish a conversion ratio for the 1920s, and I have not yet incorporated gadfly’s additional data into my conversion ratio for batters, that information is sufficient to show that the conversion factors that I am using are on the conservative side. Discussion of whether using the same ratio for pitcher ERA+ that is used for batter’s BA/SA would be useful. Recent discussions of conversion of pitcher OPS+ into ERA+ lead me to infer that the ratio should be the same, but I’d welcome confirmation of that from those who are fully able to follow the conversation between jimd and Paul Wendt.
Regressed ERA+: The projected ERA+ regressed to the mean. Baseline is the ERA+ for a rolling 5-year period centered on each season, where the 5 years are sufficient to yield a base of c. 700 NeL ip, When that is insufficient, innings in shorter seasons are weighted more heavily as necessary. For the beginning and end of Foster’s career, I have improvised as seemed appropriate. The regression changed Foster’s career average very slightly.
Proj. DERA: This adds in average defensive support to runs allowed to get a figure that is scaled exactly as WARP DERA is scaled. I provide this number as the best comparison of Foster’s quality to WARP1’s data on contemporary pitchers, to which it is directly comparable. It is also a ncessary stepping stone to DERA+, from which an accurate support-neutral winning percentage can be derived.
Proj. DERA+: 4.50 / DERA. This puts us on the threshold of an accurate support-neutral winning percentage.
( DERA+ squared)/ 1 +( DERA+ squared) = support-neutral wp
*For 1924 through 1930, because Foster is so very nearly unbeaten in 24, 25, and 27 and so much below his career winning percentage in 28-30, I have modified his winning percentage for these six seasons, treating him as 1.5 win above Pythagorean expectations in 24, 25, and 27, & 1.5 wins below in 28 and 29, and 30. This smoothes the extreme values, which would happen anyway in the regression, but it makes it more possible to calculate DERA+ in the first place. The Pyth method just doesn’t work well for seasons with winning percentages above .900 .
From Projected DERA and Projected Innings Pitched to Projected Pitching Win Shares
Explanation of data:
Projected DERA and Projected DERA+ are from the table above.
Projected IP: column #5 from innings pitched post. This is my fully-adjusted estimate of Foster’s MLE IP.
Proj. snW%: Projected support-neutral winning percentage. See DERA+ explanation in previous post on DERA for derivation.
Proj. W-L: support-neutral winning percentage derived by the pythagorean method from the regressed DERA+ multiplied by projected IP / 8.7. From here it’s an easy step to calculate pitching WS.
Pitching Win Shares. There are two elements in this calculation. First, pitcher receives credit for win shares per inning as an average pitcher (I use .0546 ws/ip for 1921-30, .058 ws/ip after 1930). To that is added 3 win shares for each win the pitcher is above average, calculated by finding pitcher’s wins above an average pitcher (support-neutral wins - .5 * decisions). Saves are not included in this estimate; perhaps they should be. Foster was credited with 12.
The appropriateness of the scale of these win shares can be checked by comparing the pitching win shares of pitcher-seasons with similar DERA and IP. See below for a spot-check of Foster against Wes Ferrell.
It has been observed in the past that my projected win shares for NeL pitchers seemed 5 to 10% higher than what pitchers with similar ERA+ and ip would earn. Turning ERA+ into DERA+ corrects this problem. As I noted in the explanatory post, Dick Redding’s win shares need to be reduced for this reason, as may those of Jose Mendez.
As a cross-check on the accuracy of this win-share estimate based on the projected DERA and IP: Wes Ferrell has a DERA in WARP1 of 3.90 in 2623 IP, so Foster’s career rate of quality as a pitcher is just slightly higher than Ferrell’s. Ferrell has 208 pitching win shares. Pro-rated to Foster’s innings, his win-share total rises to 251. Multiply that by 3.9/3.87 (the ratio of their DERAs) it rises again to 253. So I think it is safe to conclude that this method produces win shares that are scaled pretty much the same as Bill James win shares. Individual seasons also match up very well.
Whether the IP and quality conversions are accurate is, of course, another question. But I think that, if the conversions are correct, the win-share totals can used for comparison purposes without further adjustment, at least for pitchers who are used mainly as starting pitchers.
Taking these numbers, and then taking Redding's, reduced by 7.5%, Foster and Redding look nearly inseparable. Foster will be one spot ahead of Redding next ballot. Don't know where that will be.
As one of the few with Mendez in a really high position, how long do you think it'll take to look back at him, and see what, if anything, needs to be done with his estimates?
I don't know. Here are the next projects I am planning in the order I plan to do them:
1) going over the conversion ratios again in light of the data gadfly provided to see if they should be changed
2) re-doing projections for Beckwith, Moore, and Lundy with systematic regression to the mean
3) preparing projections for Stearnes, Suttles, Wilson, and Allen for 1946
4) doing a comprehensive review of NeL pitchers, hopefully after discussion on the MLE thread has settled how their conversion ratio should be pegged to that for hitters. It's in this project that I plan to revisit Mendez.
Realistically, it'll probably be about a month before I get to #4.
The passage I quoted in #30 and in the below letter to Dick Clark is the whole item published in The Negro Leagues Courier, Oct 1998.
"The headstone has been dedicated and looks fine."
--Dick Clark
Chair, Negro Leagues Cmte, SABR
---------- Forwarded message ----------
To: Dick Clark
Subject: Bill Foster gravesite
Hi, Dick.
I read this note in the Negro League Courier, October 1998, on Jim Riley's website.
>>Ernest Nagy reports that Alcorn State in Lorman, MS is beginning the process of gathering funds to
erect a befitting headstone for Bill Foster's gravesite.<<
Do you know the outcome of that effort?
Paul Wendt
1. It may change a few details, but basically I had him about right on my 1944 ballot.
2. Now that I see that. I think I underestimated him in 1944 and will bump him up in 1945.
3. That's all? I think I overestimated him in 1944 and will lower his ranking next year.
If the most common response is #1 - in fact, if it's anything other than a very large number of #3's - then Foster will be elected in 1945.
0. Have you seen Chris Cobb's posts #46-48?
Seriously, here as elsewhere I wonder what is the read rate?
Another indication that Chris is trying to find the true value of the Negro League players and not rely on quotas or anecdotes instead.
minimum 10 G pitched in that year by any player, but player only counts against the "pitchers total" if he has at least 1000 career IP. Ruth gives 1916 "the lead" with 10 HOM pitchers.
1868-75 (1) - Spalding
1876 - (1) Spalding (McVey)
1877 (0) - (McVey)
1878 (1) - Ward
1879 (2) - Ward Galvin
1880 (3) - Ward Galvin Keefe
1881-83 (4) - Ward Galvin Keefe Radbourn
1884-88 (5) - Galvin Keefe Radbourn Clarkson Caruthers
1889 (6) - Galvin Keefe Radbourn Clarkson Caruthers Rusie
1890 (8) - Galvin Keefe Radbourn Clarkson Caruthers Rusie Young Nichols (Burkett)
1891 (8) - Galvin Keefe Radbourn Clarkson Caruthers Rusie Young Nichols
1892 (7) - Galvin Keefe Clarkson Caruthers Rusie Young Nichols
1893 (5) - Keefe Clarkson Rusie Young Nichols
1894 (4) - Clarkson Rusie Young Nichols
1895 (3) - Rusie Young Nichols (Wallace)
1896 (2) - Young Nichols (Wallace)
1897-98 (3) - Rusie Young Nichols
1899-00 (3) - Young Nichols McGinnity
1901 (5) - Young Nichols McGinnity Plank Mathewson
1902 (5) - Young McGinnity Plank Mathewson Foster
1903 (6) - Young McGinnity Plank Mathewson Foster Brown
1904 (7) - Young Nichols McGinnity Plank Mathewson Foster Brown
1905 (8) - Young Nichols McGinnity Plank Mathewson Foster Brown Walsh
1906 (7) - Young McGinnity Plank Mathewson Foster Brown Walsh
1907-08 (8) - Young McGinnity Plank Mathewson Foster Brown Walsh Johnson
1909 (7) - Young Plank Mathewson Foster Brown Walsh Johnson
1910 (8) - Young Plank Mathewson Foster Brown Walsh Johnson Williams
1911 (9) - Young Plank Mathewson Foster Brown Walsh Johnson Williams Alexander
1912-13 (8) - Plank Mathewson Foster Brown Walsh Johnson Williams Alexander
1914 (8) - Plank Mathewson Foster Brown Johnson Williams Alexander Faber
1915 (9) - Plank Mathewson Foster Brown Johnson Williams Alexander Faber Ruth
1916 (10) - Plank Mathewson Foster Brown Johnson Williams Alexander Faber Covaleski Ruth
1917 (7) - Plank Johnson Williams Alexander Faber Covaleski Ruth
1918 (5) - Johnson Williams Faber Covaleski Ruth
1919 (6) - Johnson Williams Alexander Faber Covaleski Ruth
1920-21 (6) - Johnson Williams Alexander Faber Covaleski Rogan
1922-26 (7) - Johnson Williams Alexander Faber Covaleski Rogan Vance
1927 (6) - Johnson Williams Alexander Faber Rogan Vance
1928 (6) - Williams Alexander Faber Covaleski Rogan Vance
1929 (5) - Williams Alexander Faber Rogan Vance
1930-32 (3) - Williams Faber Vance
1933 (2) - Faber Vance
1934 (1) - Vance
1935 (1) - Vance
Welch would be 1880-91
Griffith would be 1891 and 1894-1906
Waddell would be 1899-1910
Mendez would be 1908-09 and 1912-25 (roughly)
Redding would be 1911-17 and 1919-30 (roughly)
Rixey would be 1912-17 and 1919-33
Grimes would be 1917-34
Foster would be 1924-37 (roughly)
Starts: 24 (15 at home, 11 road)
Team W-L: 15-9 (Foster 14-9)
Run Support: 145 runs, 6.04 per game (Am Gts averaged 4.75 runs/game for 83 games; NNL ave 5.03)
Runs Allowed (by team, not just Foster): 86, 3.58/game (Am Gts allowed 3.80 runs/game overall)
Teams Foster started against:
7 St. Louis (includes 4 playoff games)
5 Birmingham
4 Kansas City
4 Detroit
2 Cuban Stars
1 Cleveland
1 Memphis
(NOTE: This does NOT include games for which I have only line scores.)
Can anyone make this Foster teams list more accurate?
BILL FOSTER
1924 CHI AM GIA (Mem)
1925-30 CHI AM GIA
1931 HOME GRE (Chi Am Gia)
1932-35 CHI AM GIA
1936 PIT CRAW (part-time)
1937 CHI AM GIA
Should 1935 with Chicago American Giants also be a "part-time" season, for instance?
Thanks!
We've talked about good pitchers facing better teams in the major leagues: it looks like it was very much the reality in the NeL.
Leaving out the playoff games, here are Foster's # starts per team and the winning percentages of the teams:
5 Birmingham .449
4 Kansas City .613
4 Detroit .604
3 St. Louis .731
2 Cuban Stars .241
1 Cleveland .284
1 Memphis .366
The typical winning pecentage of Foster's opponents was .617. The Chi Am Giants, Foster's team, had a winning percentage of .549.
Chris J. might be able to tell us how this sort of a season compares to the workloads against good teams that ML pitchers of the 20s and 30s might have faced?
Foster did get good run support against these teams (he underperformed his pythagorean record by quite a bit), but his team's runs allowed against this roster of opponents is also quite impressive.
BILL FOSTER'S PLACEMENTS ON CAREER NGL LEADERBOARDS
WINS t-4th at 144 (with Rogan)
LOSSES 14th with 70
DECISIONS t-4th at 214
WINNING PCT .673
(50+ decisions) 14th
(25+ decisions) 24th
(10+ decisions) 40th
ADJ PCT OF TEAM DECISIONS 22.1%
(50+ decisions) 36th
(25+ decisions) 49th
(10+ decisions) 75th
WAT 5th 19.0
WAT/DEC .089
(50+ decisions) 21st
(25+ decisions) 37th
(10+ decisions) 64th
APPEARANCES ON YEARLY WINS LEADERBOARDS
1926 t-5th most wins in NNL with 11, t-8th most in NgLs.
1927 Led NNL and NgLs with 21 wins.
1928 t-3rd in NNL and NgLs with 14 wins.
1929 t-3rd in NNL with 15 wins, 4th in NgLs.
1930 2nd in NNL and NgLs with 16 wins.
1931 Led NgLs with 15 wins, led eastern pitchers with 12 wins.
1932 Led NSL with 19 wins, 3rd in NgLs.
1937 4th in NAL with 5 wins, 6th in NgLs.
Please find my my latest MLEs for Bill/Willie Foster here. The MLE method is articulated in an article linked therein.
You must be Registered and Logged In to post comments.
<< Back to main