Statistics from 1890-1910 are indeed not as complete as in later years, but they're not non-existent either. It was much tougher to analyze those players, though, and we had to use more caution for those years.

Did any of the HoF NeLs play before 1910?

I actually am fine with that, if we think of it in terms of something like Nate's percentile projections for PECOTA, where there's a larger chance that he ends up in the middle of that range than he ends up at either end.

That's exactly how confidence intervals work. Assuming this range is 2 standard deviations, then we'd have an 68% chance he's between 145-165, and about a 5% chance that he's outside this range (either way).

I will confirm that I take you and John absolutely at your word that you are looking at this objectively.

That's all I ask. Thanks.

On Satchel - his peak appears to be less impressive than a few other pitchers, like Bullet Rogan. But his longevity trumps anyone.

That's exactly how confidence intervals work. If 1 standard deviation = 10, then there's a 68% chance he's between 145-165, and a 5% chance he's under 135 or over 175.

Yes. And we would not reject the null hypothesis that Piazza was as good as Gibson.

I'm fairly certain that Satchel was still at least an average big league pitcher when he was in his early 50's, pitching in AAA. I've compared his stats to his teammates who pitched in the big leagues (and pretty much every one of them did at some point).

My guess is a Satchel career where he got to play MLB from the start and was motivated to pitch as long as he could would look like Bert Blyleven or Don Sutton, without a great peak, but he pitched so long he makes Sutton look like a flash in the pan.

On Satchel - his peak appears to be less impressive than a few other pitchers, like Bullet Rogan. But his longevity trumps anyone.

How do you account for the relative lack of pitching talent vs. hitting talent among American blacks?

After integration we got scads of good to all-world American black hitters, and very few good pitchers (only Gibson and Newcombe off the top of my head). It can't be discrimination by coaches, since the NeL obviously had lots of pitchers.

without a great peak

Why do you say this?

Yes. And we would not reject the null hypothesis that Piazza was as good as Gibson.

True. I'd leave it simply as: It's more likely than not Gibson was a bit ahead of Piazza.

> Did any of the HoF NeLs play before 1910?

Yes.

HoF & HoM - Frank Grant, Pete Hill
HoM - Grant "Home Run" Johnson

After integration we got scads of good to all-world American black hitters, and very few good pitchers (only Gibson and Newcombe off the top of my head). It can't be discrimination by coaches, since the NeL obviously had lots of pitchers.

That form of discrimination started after integration. You obviously could not discriminate positionally when in the Negro Leagues. Because somebody had to pitch, and to catch, etc. But these days when you see a black player it's most likely he's an outfielder.

FWIW, we passed over a few HOF players from the NeL because the stats didn't suggest any real greatness from them. Many of us here also don't think Satchel Paige is as great as he's normally thought to have been, either.

I have Joe Williams as the best NgL pitcher and I still think Willard Brown sucks.

Yes.

HoF & HoM - Frank Grant, Pete Hill
HoM - Grant "Home Run" Johnson

OK, never heard of them.

True. I'd leave it simply as: It's more likely than not Gibson was a bit ahead of Piazza.

I'd go stronger that that (it's very likely he was better than Piazza - probably 80%) but it's more likely he was worse than Piazza as it is he was better than Hornsby.

That form of discrimination started after integration. You obviously could not discriminate positionally when in the Negro Leagues. Because somebody had to pitch, and to catch, etc. But these days when you see a black player it's most likely he's an outfielder.

Right.

But that means the NeL pitching couldn't have been very good relative to the hitting.

You didn't see the pitching equivalent of Mays, Campanella, Aaron, Robinson, Irvin, etc. pour into MLB after integration.

Really?

I'd think the ability to pitch to a 125 ERA+ in MLB at ages 41-46 would be an indicator of a pretty premier talent.

Isn't elite RP pretty much the floor to Paige's talent?

Oh, he was unquestionably great, Snapper, but just not the greatest of all-time as some claim. At least based on his NeL stats, that is.

I have Joe Williams as the best NgL pitcher

Yeah, Smokey looks like the best - think Pete Alexander.

You didn't see the pitching equivalent of Mays, Campanella, Aaron, Robinson, Irvin, etc. pour into MLB after integration.

Bob Gibson, Juan Marichal, Ferguson Jenkins, Don Newcombe

it's very likely he was better than Piazza - probably 80%

I assume you're just considering hitting. If you add in fielding I don't see any way Piazza could match Gibson.

Yeah, I think I rate Paige higher than you guys do.

Bob Gibson, Juan Marichal, Ferguson Jenkins, Don Newcombe

Sure, but if you look at hitters vs. pitchers, from 1947-70, 8 of the top 20 OPS+ (min 3000 PA) were black, only 3 of the top 20 ERA+ (min 1000 IP), Gibson, Marichal and Jenkins.

It's not controversial to say that American born black talent skewed towards position players from the very beginning of integration.

I assume you're just considering hitting. If you add in fielding I don't see any way Piazza could match Gibson.

Hell we can't even evaluate catcher defense for the guys playing today.

#221 Gibson versus the running game? Unknown but put it this way, I'm far more confident that this is a net plus for Gibson (over Piazza that is) than I am about their relative offensive rankings. Piazza was historically bad against the running game.

Other aspects of catching? Quite literally unknowable. Piazza sure caught an awful lot of very successful staffs despite spotting the league a fair number of runs due to his weakness against the running game.

"Hell we can't even evaluate catcher defense for the guys playing today."

I don't think Yadier Molina would agree with this.

"Hell we can't even evaluate catcher defense for the guys playing today."

I don't think Yadier Molina would agree with this.


But his brother is way better, at 1/15th the cost.

I took AROM's advice (#44) and checked out Seamheads.com. They are everything he says they are. Stuff you can NEVER find anywhere else. I was even able to get lefty/righty splits in ballparks effects, something I've been whining about for years. THANKS, Sean (it is Sean behind AROM, isn't it?). I am now an evangelist and an acolyte. - Brock

Gary Ashwill deserves somewhere between 95% - 99% of the credit for the Negro Leagues research, with a few others of us making minor contributions.

Players who played in MLB and Negro Leagues (1940s-1950s)

Players who played in MLB and the Negro Leagues, 1910s - 1920s:

Dimensions for Josh Gibson's primary parks:

The Competition - Semipro Clubs (Courtesy of Outsider Baseball Bulletin, Scott Simkus, editor)

Competition Part II - MLB exhibition games vs. various classifications, pre-1951. (AAA represents highest minor league class at the time, AA second highest, etc.)


Courtesy of Outsider Baseball Bulletin

Unfortunately Gibson's career begins about where the seamheads compiled data to date ends, but Scott Simkus had compiled 1933, so I can provide a little real data.

Gibson played in 58 of the teams 63 games that have boxscores vs. other Major Negro league teams.
Prorating to 154 games, Gibson's season would look like this:

TOTAL GAMES PLAYED: 142
CATCHER: 96
FIRST BASE: 10
THIRD BASE: 10
LEFT FIELD: 20
RIGHT FIELD: 6

Gibson was 21 years old. Scott noted the similarity to the season of a 22 year old Catcher:

TOTAL GAMES PLAYED: 158
CATCHER: 139
FIRST BASE: 12
THIRD BASE: 1
LEFT FIELD: 15
CENTER FIELD: 2
RIGHT FIELD: 7


This is how the Reds deployed Johnny Bench in 1970.

Here’s Josh Gibson’s splits for 1933:
AS CATCHER: .370 (51-138), 12 doubles, 3 triples, 7 home runs, 36 rbi
OTHER POSITIONS: .493 (33-67), 3 doubles, 4 triples, 8 home runs, 29 rbi

Defense/Stolen Bases:

The only other catcher for Pittsburgh was Bill Perkins, noted defensive specialist.

Estimated Stolen Bases per Game:
Gibson - .55 per 9 innings
Perkins - .29 per 9 innings

Over the course of a full 154 game season, Pittsburgh would have stolen
about 275 bases, while holding their opponents to less than 65.

Data again from Outsider Baseball Bulletin.

1933 Batting:


Courtesy Outsider Baseball Bulletin

Players who played in MLB and Negro Leagues (1940s-1950s)

There should be additional data with players who played in AAA/AA and the Negro Leagues.

Jim Pendleton LF

Heh, somehow I thought he was a Jim Lemon-like white lumbering LF.

EDIT: He was on the Reds in '59 when I was 8 and followed the team religiously. Of course my whole experience was via the radio and box scores and I never had one of his baseball cards, so my lack of racial knowledge can hopefully be forgiven.

Thanks KJOK. That's a lot of great data.

Unfortunately Gibson's career begins about where the seamheads compiled data to date ends, but Scott Simkus had compiled 1933, so I can provide a little real data.

Gibson played in 58 of the teams 63 games that have boxscores vs. other Major Negro league teams.
Prorating to 154 games, Gibson's season would look like this:

TOTAL GAMES PLAYED: 142
CATCHER: 96
FIRST BASE: 10
THIRD BASE: 10
LEFT FIELD: 20
RIGHT FIELD: 6

Gibson was 21 years old. Scott noted the similarity to the season of a 22 year old Catcher:

TOTAL GAMES PLAYED: 158
CATCHER: 139
FIRST BASE: 12
THIRD BASE: 1
LEFT FIELD: 15
CENTER FIELD: 2
RIGHT FIELD: 7


This is how the Reds deployed Johnny Bench in 1970.

Thanks for the work.

I think the issue that remains is how was he used in barnstorming.

The pro-rating only represents real wear and tear if he was catching 70+ 9-inning barnstorming games in addition to his league games.

KJOK,

Do you know why the Shades of Glory stats show many fewer games for Gibson? (38 games on his wiki page) Do you think those are a subset or that both sets might contain unique games?

You say you have data for 63 games vs other Negro League teams with boxscores. Any idea of how many games like this there were? Could there be 80 Negro League games and we only have boxscores for some? 100? 120?

After integration we got scads of good to all-world American black hitters, and very few good pitchers

I was thinking about this and it is striking how many Negro League pitchers could really hit and their hitting is a big chunk of their value. Guys like Paige and Cannonball Dick Redding who were strictly pitchers were the exceptions. I think a lot of this is opportunity and part of it is due to limited roster slots. If you're playing baseball and hitting often you're going to get better at it. I think we would see better hitting pitchers today if they practiced it in the minors. With a 16-20 man Negro League roster a guy who could pitch and play LF was pretty valuable.

There should be additional data with players who played in AAA/AA and the Negro Leagues.

Yes, and I have most of those guys identified too, although I haven't yet started on any statistical comparisons. Also didn't want to post such a long list here...

Do you know why the Shades of Glory stats show many fewer games for Gibson? (38 games on his wiki page) Do you think those are a subset or that both sets might contain unique games?

You say you have data for 63 games vs other Negro League teams with boxscores. Any idea of how many games like this there were? Could there be 80 Negro League games and we only have boxscores for some? 100? 120?

It's my understanding the NLRAG (whose data is used in Shades of Glory) used ONLY sanctioned Negro League games. There may also be an issue where they were still missing some boxscores that we now have.

The problem with using only sanctioned games is that 1933 was the re-start of the Negro National League, and some of the top teams, with the top players, had not joined (or rejoined) the league. This includes the Kansas City Monarchs, Pollack's Cuban Stars, Philadelphia Stars, Cuban Stars, New York Black Yankees, Bacharach Giants, Birmingham Black Barons, and a few others. We (seamheads.com researchers, such as Gary Ashwill, Scott Simkus) include those games as being 'major' Negro League games.

We have data for 63 games. There were actually 101 Crawford games vs. other major Negro League teams, and they wwent 54-35-2. The Crawfords also played another 26 games against top white semi-pro teams (going 22-4). They probably played even more barnstorming games against lesser (Division 2 or 3 classification above) teams, but those were likely never reported in any major paper.

After integration we got scads of good to all-world American black hitters, and very few good pitchers

Pitching is the most specialized position in baseball. I suspect that one thing that becomes hard when training and practice time is scarce, and rosters are small, is to specialize. I'm guessing a lot of great black athletes who could have been pitchers were instead made position players in the Negro Leagues since position players could play every day, thus being more valuable. And those that were pitchers also had to play RF or 1B, and work on their hitting, so there were only a few of the best (like Satchel Paige) that were allowed to concentrate just on pitching.

It's the jack of all trades ability of Martin Dihigo that makes an MLE argument for him so difficult. What the heck position would he have played in the majors? It's clear he added more value to a Negro League team than he would have to a National League team of the same era. You run into this with other players also. The MLE for their off-day time in the outfield doesn't add anything to their case because it is below MLB replacement level but it clearly helped the team they were on. Likewise Cool Papa Bell's pitching helped his team but he wasn't a big league caliber pitcher.

Likewise Cool Papa Bell's pitching helped his team but he wasn't a big league caliber pitcher.

Bell actually WAS a pretty good pitcher before he converted to the OF. He was his teams #2 starter in his rookie year, and he was the NUMBER 1 starter in 1923.

Bell was a good pitcher before he got hurt but it wouldn't have made him a Hall of Famer.

We have data for 63 games. There were actually 101 Crawford games vs. other major Negro League teams, and they wwent 54-35-2. The Crawfords also played another 26 games against top white semi-pro teams (going 22-4). They probably played even more barnstorming games against lesser (Division 2 or 3 classification above) teams, but those were likely never reported in any major paper.

Great information. So if Gibson caught league games at the same rate overall as for the games we have boxscores, that's probably 65-70 games caught. Plus however many exhibition games he caught (where he may have played more at other positions, but probably caught at least some of them.) Does it seem reasonable to assume his catching workload would be similar to guys like Bill Dickey and Mickey Cochrane, about 120 per year?

Do you have home/road splits for any of his seasons? I know he played mostly in big parks, but my guess is that he personally won't show a huge home/road split. Seems like the great ones are less affected by parks (their homers tend to be no doubters) than mid range power hitters.

is there any progress being made on the real value of SB/CS when considering the leverage of the game when runners do attempt to steal? My understanding might be a little out of date as the last I read of this was Pete Palmer stuff in the 90s and he was still valuing SB at about 0.2 of a run, but I got to think it's more than that owing that runners run more when the game is on the line. But I only know that anecdotally and not everyone agrees pointing to e..g Ricky Henderson. So I dunno. what is the latest in thinking on this?

Back in the 80's Palmer used .3 for steals and .6 for CS. This is too much given their run value, but I think Palmer justified it by claiming leverage. I don't know if Palmer is still using that estimate, but I think research has shown that on average there is little leverage in how the SB is used. I'll have to check the Retrosheet research pages.

Come to think of it, it's probably in The Book, but I can't remember and don't have my copy with me.

#249 Palmer backed off in later versions. While the research he relied on (Dave Smith showed that stolen base attempts are not truly random, they are more frequent in high leverage situations) is correct, it just didn't seem to manifest itself in team wins and losses. Fast teams don't tend to outperform their pythags. Teams with problems with the running game don't tend to under-perform them. (That's not the only way it's been studied. People looked hard -- in every way they could think of)

When you think about it, it might have something to do with the nature of pitchers used in the late innings of close games. There are plenty of closers who are basically hopeless against the running game (almost ignoring it), but they're also murder to attempt to run any form of sequential offense against. (Cue Roberts/Rivera. True enough, it can work. Just doesn't seem to happen often enough to matter)

Hey KJOK,
Did Luque actually play in the Negro Leagues, or just the Cuban winter leagues - which were integrated right?
Luque on Seamheads

Thanks for info on leverage on running game. I think it is an important issue because it is one of the few baseball events that can occur in a non random way. Someone has to decide to steal, whereas hits, walks, strikes etc. all occur on a continuous basis at a more or less predictable rate.

I was just reading that Jackie Robinson stealing home thread on Hardball times site. I think it was 32 lifetime attemps, with about half them in +/- 1 run games. Not that this has much to do with stealing second for the average MLB player, but it got me to thinking about it. He actually succeeded I think 74% when you take out certain miscues.

Did Luque actually play in the Negro Leagues, or just the Cuban winter leagues - which were integrated right

He definitely played in the Cuban Winter Leagues, which where integrated, and some seasons I think had more U.S. black players than Cuban players.

He played for what would later be considered a Negro League team - the Long Branch Cubans, in 1913, but they were playing in the New York/New Jersey league that year, and not in the "negro leagues". He susposedly played for the Cuban Stars in the U.S. in some season, but we haven't found any games yet where he played for them against other major Negro League teams.

Do you have home/road splits for any of his seasons? I know he played mostly in big parks, but my guess is that he personally won't show a huge home/road split. Seems like the great ones are less affected by parks (their homers tend to be no doubters) than mid range power hitters.

Road here includes a few 'neutral site' games in a couple of parks where balls hit over the fence were ground rule doubles and not home runs:

1933:
Home: .390 ba 22/g 77/ab 30/h 5/2b 3/3b 6/hr 4/bb

Away: .422 ba 36/g 128/ab 54/h 10/2b 4/3b 9/hr 9/bb

On a related note, the Journal 21: The Illustrated Journal of Outsider Baseball will contain an in-depth look at the 1933 Pittsburgh Crawfords and will feature never-before seen statistics and 17 original illustrations.

The journal is $25 (shipping and handling included) and can be purchased at http://infinitecardset.blogspot.com/p/how-to-buy-21-illustrated-journal-of.html (Books will be shipped starting on Saturday April 9th)

You've seen some samples of the statistics above from Scott Simkus, and if you've never seen the baseball illustrations of Gary Joseph Cieradkowski you should definitely check them out http://infinitecardset.blogspot.com/

And in addition to the collaboration with Cieradkowski, Simkus has a weekly baseball magazine, the Outsider Baseball Bulletin, which contains all manner of historical baseball analysis, and is dirt cheap for the fantastic content you'll get:

Outsider Baseball Bulletin

End of advertisements.

FWIW, we passed over a few HOF players from the NeL because the stats didn't suggest any real greatness from them. Many of us here also don't think Satchel Paige is as great as he's normally thought to have been, either.

To the extent that this represents the consensus view of the members of the HoM project, I think it's one of our biggest mistakes.

The MLEs for Paige (and for most other NeLg players) were done before the results of the Hall of Fame (NLRAG) study were published for selected players. They were based mostly on Holway's Complete Book, supplemented in some cases by data from Cuban, Mexican, or Puerto Rican leagues and from a couple of seasons that Gary Ashwill had tabulated. For pitchers, that means the MLEs were mostly based on W-L records relative to team, supplemented for a few seasons by R/9 data.

For batters, the results of the early MLEs have held up pretty well when we've compared them to the more complete NLRAG data. For pitchers, and especially for Satchel Paige, that isn't true. The MLEs show Paige as a very good pitcher during the 1930s, then dropping to an average or below average pitcher during 1940-47. That result appears to be an artifact of how he was used during that period and how it affected his W-L record. During that period, he frequently started and pitched for about three innings to take advantage of his drawing power as a gate attraction. The effect was that he seldom received credit for the wins in those games, but was charged with many of the losses.

As far as I'm aware, the NLRAG project has published 1920–1949 pitching statitics for 16 pitchers who were either Hall of Famers or candidates (I didn't include players like Oscar Charleston who pitched occasionally, but weren't regular pitchers): Cool Papa Bell, William Bell, Chet Brewer, Ray Brown, Bill Byrd, Andy Cooper, Leon Day, Martín Dihigo, John Donaldson, Willie Foster, José Méndez, Satchel Paige, Cannonball Dick Redding, Bullet Rogan, Hilton Smith, and Cyclone Joe Williams. Against this group of elite pitchers, Paige compares very well. Here are the top 5 in several categories:

R/9
1. Paige 3.31
2. Foster 3.36
3. Smith 3.37
4. Rogan 3.66
5. W. Bell 3.86

ERA
1. Smith 1.68
2. Paige 2.02
3. Foster 2.40
4. Rogan 2.59
5. Brewer 2.89

K/9
1. Paige 7.4
2. Rogan 5.3
3. Smith 5.2
4. Day 5.1
5. Foster 4.9

BB/9
1. Smith 1.1
2. Williams 1.1
3. Cooper 1.4
4. Byrd 1.4
5. Paige 1.5

K/BB
1. Smith 4.9
2. Paige 4.9
3. Williams 3.8
4. Byrd 2.5
5. Méndez 2.5

H/9
1. Paige 7.0
2. Foster 7.4
3. Smith 7.5
4. Day 7.7
5. Rogan 8.0

A caveat is that we don't have league averages or park factors from the NLRAG data, so comparisons remain somewhat risky. But I'm comfortable rejecting the HoM MLEs as flawed, especially for Paige's 1940-47 period. Here are the MLEs:

1927-37: 2678 IP, 3.56 DERA (ERA adjusted for defense), 126 DERA+, 193-122 neutral W-L
1940-47: 1512 IP, 4.93 DERA, 91 DERA+, 81-97 neutral W-L

In comparison, here are his rate statistics from the NLRAG data:
1927-37: 3.38 R/9, 2.28 ERA, 7.1 H/9, 7.2 K/9, 1.6 BB/9, 4.5 K/BB, .667 W-L pct
1940-47: 3.15 R/9, 1.45 ERA, 6.7 H/9, 7.7 K/9, 1.3 BB/9, 5.8 K/BB, .547 W-L pct

Every rate statistic except his W-L percentage is better during the 1940s than in the 1930s. While it's possible that the 1940s were a better pitching environment, I think the main anomaly here is the W-L record. While I can't quite be sure exactly how good Paige was at his peak, I'm confident that he was an excellent pitcher for a very, very long time.

I don't think there was ever any doubt he was an excellent pitcher, just that his peak wasn't the best among his peers. The fact that he pitched for a long time is beyond dispute. He makes Nolan Ryan look like Koufax in that regard. The data above certainly changes my mind though, especially the strikeout rate. By those numbers he does appear to be the top Negro League pitcher under any criteria.

One of the reasons he dropped is an arm injury. He was as good as it gets for a period of time, got hurt, rehabbed and re-invented himself as essentially a short reliever, though he was pitching the beginning of the ballgame.

While Paige's usage did change after his 1937 arm injury, I don't think it would be accurate to describe him as a "short reliever... pitching at the beginning of the ballgame."

His usage statistics for 1927-37 are similar to the average of the 16 pitchers from the NLRAG study. Keep in mind that Negro league teams had small rosters and they generally didn't carry dedicated relief pitchers. That meant that their top starters were being used in relief more often than top major league starters of the same era. During 1927-37, 69% of Paige's appearances were starts; he completed 76% of his starts; and he averaged 6.2 IP/G. In comparison, the averages of the 16 pitchers for which we have NLRAG data are: GS/G = 69%; CG/GS = 72%; and 6.3 IP/G.

In contrast, during 1940-47 Paige's statistics were GS/G = 85%; CG/GS = 28%; and 5.0 IP/G. So it's true that he was pitching fewer innings per game and was less likely to complete a game, but he was also making many fewer actual relief appearances. So, some of the change was that they were substituting short appearances at the beginning of a game for short appearances at the end, which meant that the overall drop in innings pitched per game (from 6.2 to 5.0) wasn't huge. My guess is that about half of his starts (probabaly against weaker opponents) were of the 3-inning variety, while he was expected to pitch deep into the game in the other half. He continued to pitch quite a few complete games. (In fact, he was still pitching many complete games in Class AAA for Miami at ages 49 to 51.) I don't think that the relative strength of his strikeout and walk statistics during the 1940s can be explained by the change in usage; he was not being used like a modern relief pitcher.

This thread and the mention of Scott Simkus reminds me that I should do a project and a Hartford semi-pro team called the Savitt Gems and other CT teams like the Meriden Insilcos when I have time.

That is all

I absolutely agree with Brent in 260. They were picking spots for Paige - CG when they need the win, 3 innings when they need the gate. Hilton Smith finished a big chunk of those games and there wasn't a sizeable dropoff for the Monarchs.

LEVEL WINS LOSSES TIES PCT.
Negro League 91 110 6 .452

How good were these Negro Leagues and MLB teams relative to the average team in each league? Did MLB teams play primarily the best Negro League teams? Did they tend to rest some of their players more often? I'm wondering if we could use this data to estimate an MLE factor for the pre-1940s Negro Leagues (as it stands, the conversion rate would be greater than 1, which seems implausible).

How good were these Negro Leagues and MLB teams relative to the average team in each league? Did MLB teams play primarily the best Negro League teams? Did they tend to rest some of their players more often? I'm wondering if we could use this data to estimate an MLE factor for the pre-1940s Negro Leagues (as it stands, the conversion rate would be greater than 1, which seems implausible).

It would be hard to do using the MLB vs. Negro League games. Certainly, the conversion factor shouldn't be greater than 1, but it does provide evidence that the Negro League teams were better than top minor league teams. The problem is you'd have the "Lefty Groves All-stars" (later the Dizzy Dean All-stars, etc.) playing the Baltimore Black Sox. Lefty would have some A's teammates, but it wouldn't be the entire A's team. You might have a guy from the Senators along. Then you'd have a couple of International League players too. On the other side, you'd have most of the Black Sox, but then they'd get Judy Johnson and Frank Warfield from Hilldale to supplement their lineup.

There is also data of Negro League teams vs. minor league teams. In these games, you were much more likely to have the 'real' teams playing each other. In those games, the Negro League teams won the majority vs. the highest minors (like IL, American Assoc), then even higher percentage vs. Eastern League (next level down), etc. just like you would expect as the competition lessened. I would put a lot more faith in those numbers personally.

#264 Do they have number versus PCL teams? There was more major league ready talent in the PCL than any place else in the minors.

Now that I think about it, it seems that if there are enough known games versus PCL opposition it ought to be possible to calculate some kind of conversion factor. Devil is as always in the details, since you'd have to know who was actually playing in any given game (for both sides).

Seems to me you could use Arpad Elo's methods. Problem being that you'd need ~30 games to get any kind of meaningful results and I don't think they were playing 30 times within any kind of useful time period.

1932 Crawfords - ALL games (Charleston second best hitter on team)


from the Pittsburgh Courier, January of 1933.

Negro League teams vs opponents 1901-1950 by classification:

Top Negro League teams vs. Semi-pro opponents:



Data courtesy Scott Simkus and Outsider Baseball Bulletin

1932 Crawfords - ALL games (Charleston second best hitter on team)

Pos 1932 Crawfords G AB R H 2B 3B HR SO BB BA OBP SLG
C Josh Gibson 123 490 114 186 45 16 34 41 31 .380 .417 .745
1B Oscar Charleston 135 430 104 156 22 19 13 37 75 .363 .457 .593
1932 Crawfords 135 4513 898 1449 263 109 101 500 407 .321 .377 .495


from the Pittsburgh Courier, January of 1933.

That's a pretty high offensive context.

Was the whole league like that?

Semi-Pro 1682 911 49 .649 5.78 4.09

I'm surprised the W% isn't much higher.

I'm wondering if we could use this data to estimate an MLE factor for the pre-1940s Negro Leagues (as it stands, the conversion rate would be greater than 1, which seems implausible).

If that data represented average strength teams in each leagues and was a reliable sample of games, that .55/.45 split in wins would be equivalent to a very slight adjustment at the player level. Something like a guy hitting .300 with 30 homers in the weaker league would hit .285 with 28 in the tougher league.

I think we can get a reasonable estimate from the data of Negro league players who went on to play MLB, by age adjusting the data points. Some of my initial attempts at this show that a player would lose a lot in BA but gain a bit in power and more in walks. MLB in those years saw a change in style of play to lower BA and more power/walks thanks to a generation of power pitchers. So a league adjustment is needed as well.

One thing to be careful of though - this will tell you what a player who played in the 30's/40's in NLB might have hit in the 1950's MLB.

It does not tell you what they would have hit in 1930/40's MLB. Both MLB and NLB leagues before 1947 were weaker than the integrated league that followed.

Something like a guy hitting .300 with 30 homers in the tougher league would hit .285 with 28 in the weaker league.

Got that part mixed up and edit won't work. A guy hitting .300 with 30 homers in the weaker league would hit .285 with 28 in the tougher league

It does not tell you what they would have hit in 1930/40's MLB. Both MLB and NLB leagues before 1947 were weaker than the integrated league that followed.

Couldn't we use the opponents' data to try to figure this out? Since the barnstorming data is from 1901-1950, we could get a pretty good idea of the Negro Leagues' quality through the whole time period. To adjust the data, we could calibrate it against direct estimates based on players' stats--we'd multiply the 1940-1950 data by some constant to force it to equal the players conversion rates, and then multiply the 1901-1940 data by the same constant.

Semi-Pro 1682 911 49 .649 5.78 4.09

I'm surprised the W% isn't much higher.

There is a severely disproportionate amount of Brooklyn Bushwicks and other NY teams included in the Negro Lg vs. Semi pro data. From memory, there may be 600 or 700 such games, which skew the data. The winning percentage goes way up against other teams. This is all a work in progress, and has actually been updated quite a bit since originally published in the Outsider Baseball Bulletin.

That's a pretty high offensive context.

Was the whole league like that?

Remember, the 1932 data is not of the same quality as 1933 - it is ALL games, not just 'league' games, and based on contemporary stats tracking. Unfortunately, pitching for the Crawfords wasn't reported, but I would guess the opponent offensive totals would be MUCH less than the Pittsburgh offense totals, which would bring down the offensive context for Crawfords games quite a bit.

What was their W-L record? That combined with runs scored would give a good indication of runs allowed.

I found one source on the web saying they had a 99-36 record. With 898 runs scored in 135 games, that makes 510 runs allowed most likely, plus or minus 100 runs (they could have beat or underperformed pythag to that degree). So the likely offensive context for the Crawfords was between 9.7 and 11.4 runs per game.

MLB average run context was 9.8 that year, down from 11.1 in the crazy 1930 season. So there's nothing unusual at all about that team's run environment, that was just baseball in the 1930's.

20. AROM Posted: February 28, 2012 at 09:19 AM (#4070161)
Using .2 runs for a SB and .5 for a CS, Piazza cost his teams about 70 runs over his career. That's giving him 100% of the blame and not adjusting for pitchers.

I have his game calling being worth a bit more than that. Max Marchi is working on a similar metric and I can't wait to see what he comes up with.

My apologies for laying low. Between working two jobs and commuting, I burn 80 hours a week.

On to the good stuff:

As AROM mentioned, Max Marchi formerly of The Hardball Times, has moved on to Baseball Prospectus, and has done studies on historical catching value.

http://www.baseballprospectus.com/article.php?articleid=16199

Catchers improving pitching staffs - runs prevented:
248 - Tony Pena
210 - Mike Scioscia
205 - Javy Lopez
205 - Mike Piazza
191 - Carlton Fisk
178 - A.J. Pierzynski
161 - Russell Martin
150 - Jim Hegan
150 - Jose Molina
146 - Andy Etchebarren

When I get an opportunity, I hope to run through the revised Baseball-Reference WAR rankings and share my findings.