Flip.

Has anyone else noticed that there are very significant differences between Seamheads' gWAR for Negro-League players and bbref's bWAR?

I had been assuming that because bbref's data is very similar to Seamheads' data, just less complete because it includes only league games (whereas Seamheads includes independent teams, playoff series, games against white teams, all-starts games, and son on), that the WAR for players on the two sites would be about the same. For some players it is rather similar. Mule Suttles, for example, has 36.8 bWAR in 906 games, while he has 36.3 gWAR in 993 games, about 10% higher when the BWAR is prorated to 993 games. But for Bill Foster, on the other hand, bbref has him at 45.6 pitching bWAR in 1499.7 innings,while Seamheads has him at 39.7 pitching gWAR in 2005.3 innings: he is 50% more valuable in bWAR than in gWAR! In Foster's case, I would guess that some of the difference probably lies in the handling of park factors, but I don't know for sure. In both of these cases, bWAR views the player more favorably, but it goes the other way, too. Newt Allen, for example, has 20.6 bWAR in 945 games but 29.5 gWAR in 1075 games. Some of that difference is fielding value, but gWAR likes Allen's offense better, too.

I'd like to use the fact that the sites have fully calculated Wins Above Replacement for Negro-League players to simplify the process of evaluating them, but the magnitude of the differences between what the systems show adds its own complication. Does anyone have any systematic explanation for the differences between their results?

Chris Cobb,

I don’t know the answers, but…if I had to guess, it starts with fielding. SH uses DRA, so if BBREF shows wide differences from the SH fielding component, they are either computing their own TZ type fielding or they are running some kind of conversion from DRA to whatever units they want to present.

The second piece is the league v non-league games, which likely means not including Cuban League stats in the “league games” bucket.

There could also be small differences in how they compute the positional adjustment, though I am not certain. Also don’t know if they each handle baserunning the same way.

To the best of my understanding they both use wOBA as the foundation for their hitting component, so they should be similar in that regard.

This might be a good question for the always generous and helpful Adam Darowski!

Cross posting from my Twitter account (@EricChalek) b/c this group will find it helpful, I think:

"Early testing results are in...latest update of MLEs are probably conservative by 10-20% on WAA but right on in terms of total WAR. That means they are overshooting playing time but undershooting performance v average. Well, that's regression for you!"

This is an unreleased update, BTW, but I wanted you all to know that I'm finally starting to get some testing going. As I've said before, trust the total WAR more than any season or individual component!

lots of comments on eligibility - which I think means I have to take David Wright off my ballot (I had him 13th last year), right?

Wright remains eligible

Dr. C, I find it interesting that you don't vote for John because of lack of peak, but then vote for Phillips, who has, ostensibly, less of a peak than John (even going by your preferred bWAR which is least favorable to John).

KCgard2, the difference is really about DRA. Once I bring that into the picture, Phillips moves further away from John and gains in peak.

I've counted ballots the past few years and am happy to do it again this year

A note for posterity: CookieDaBookie start with C.

Here's a quick update on the differences and commonalities between the methods of bbref's bWAR and Seamhead's gWAR for Negro-League players.

Differences:

1) Fielding values are different, but predictably so. Bbref starts with the DRA that Seamheads gWAR uses and divides them by 1.75 to bring the variance in fielding values closer to Total Zone's narrower variance.

2) Park factors create difference. Bbref's bWAR incorporates park factors, but it appears that it doesn't have park factors for NeL teams' ballparks: all of them are listed as neutral (100/100). The effect of Schorling part (Chicago American Giants' home park and a notoriously low-scoring park) is clearly evident in the gWAR for CAG players and for the CAG as a whole in the team stats, so I am sure that's part of what is going on. The park factors that gWAR uses do not appear to be available on the site, however, so users of gWAR have to take that component on faith at the moment. (If someone knows where they are, I'd love to know!)

3) I think I remember it being said that gWAR for pitchers was based on RA, with no adjustment for fielding support, but I haven't found documentation of that. If so, that would be another source of difference.

Commonalities

1) Positional adjustments are the same: gWAR uses bbref's.

2) As per Dr. Chaleeko's testimony, batting WAR is WOBA-based in both, and from what I've been able to find, batting value tends to be very similar, once park effects are accounted for.

3) Baserunning value does not seem to be a source of difference between offensive values, but I'm still looking at this one.

Finally, with respect to the data itself, as Dr. Chaleeko noted, bbref is only including regular season games in the NNL1, ECL, EWL, NNL2, and NAL in its data, whereas Seamheads is also including independent teams, selected CWL, FHL, and MXL seasons, as well as championship series, all-star games, and games against white minor-league and major-league opponents. One can view the Seamheads' data by "individual stints," however, and separate out the regular season NeL play, and in all the cases I have seen, Seamheads and bbref are using the same data set.

Also, my re-review of NeL players using the Seamheads and bbref WAR systematically agrees so far with cookiedabookie's and Dr. Chaleeko's findings in elevating George Scales. I've been working backwards from the 1940s because I use that era to develop conversion factors between NeL and NL/AL conditions, and I've only gotten through the 1940s and 1930s so far, but through this period Scales clearly stands out. I don't know if he'll make my ballot, and if he does, I'll likely be placing him closer to cookiedabookie's #15 slot than Dr. Chaleeko's #4, but the question of conversion factors is still open, and I've started with very conservative estimates for conversion factors.

Thanks to kcgard2 and cookiedabookie for counting ballots this year.

Seem pretty light on ballots so far. Hope some folks are planning to vote before new year

Are we fitting the NGL MLE to the wrong curve? I was looking over the numbers today and it tends to pile up the lowest seasons between -1 WAR and 1 WAR. We know that the worst players in the Negro Leagues were typically not good enough to make it in MLB. If you have a season where a Double-A player piles up a lot of plate appearances you would expect to see -3 and -4 WAR seasons pretty regularly in the MLEs. We see this happen in the actual major leagues and they didn't have the same constraints on player pool. If you did an MLE for all Double-A players to MLB in 2022 we should expect that most of the players are below 0 WAR. There are multiple Negro Leaguers in the database who played corner positions and had OPS+ below 60.

Instead of a distribution from 11 to -4 for MLE seasons we might even fit that to a max of 11 and minimum of -6.

314. cookiedabookie Posted: December 22, 2022 at 12:08 AM (#6110490)
Seem pretty light on ballots so far. Hope some folks are planning to vote before new year

I will plan to have one in, was hoping Kiko would share whether he had any updates from his Retrosheet or Negro League projects before finalizing.

I'm planning to vote, my holiday break starts next week so I should be able to get a ballot in before New Year's.

A 50-percentile NgL player is not a 50-percentile MLB player, I'm pretty we sure we all agree (we should, anyway)

From 1920 to about the mid/late 1930s [I would rate the NgL] starting from 70% [of MLB]

We know that the worst players in the Negro Leagues were typically not good enough to make it in MLB

Absolutely no offense intended to anyone involved - you're all people of good will who have given this subject your serious attention - but I disagree profoundly with all of the above.

I see several Negro League players who are left fielders and first basemen with careers of several years and an OPS+ of 45 to 50. They really weren't even good enough to play in the league they were in and yet they had careers that lasted 5 or more years.

The quality of the Negro Leagues varied quite a bit as well depending on the various league expansions and contractions, disputes over jumping leagues and larger economic trends (roaring 20s followed by the depression). For example, the league expansion in 1923-24 surely diluted the quality of the worst players. There were 8 teams in 1921 and 17 in 1924. In 1936 you're down to 6 teams due to team failures and mergers. The next season the NAL begins and you're back up to 15 teams.

DL from MN wrote:

I see several Negro League players who are left fielders and first basemen with careers of several years and an OPS+ of 45 to 50. They really weren't even good enough to play in the league they were in and yet they had careers that lasted 5 or more years.

Yes, the range of ability in the Negro Major Leagues was wider than in the National and American Leagues. The shorter seasons and less regular patterns of competition may amplify the effects of this range in the statistical record, however.

The quality of the Negro Leagues varied quite a bit as well depending on the various league expansions and contractions, disputes over jumping leagues and larger economic trends (roaring 20s followed by the depression). For example, the league expansion in 1923-24 surely diluted the quality of the worst players. There were 8 teams in 1921 and 17 in 1924. In 1936 you're down to 6 teams due to team failures and mergers. The next season the NAL begins and you're back up to 15 teams.

Expansion and contraction are definitely factors in the level of competition in Black Baseball. Describing that expansion and contraction solely at the level of the league presents a more dramatic view of its impacts than is warranted, I think. Because the Negro Leagues were always only one among multiple components of the business operations of Black Baseball teams, my sense is that while leagues came and went, the top teams where talent was concentrated often continued to exist independent of the leagues. In the pre-league period and for some of these teams during the 1930s, their play is well documented. Because much of league creation involved establishing agreements among existing teams, the presence or absence of the league is not the decisive determinant of competition levels. Also, because the talent distribution (and the economic incentives) in Black Baseball were less stratified by league than in the White Baseball structure (and prior to the advent of the farm system was less stratified there than is often recognized), expansion and contraction when teams move in and out of leagues also has less impact, I think, because the teams that are joining a league are not necessarily weaker than the teams already in the league, although they probably were in the majority of cases. It's not like major-league expansion where a new team gets created from cast-offs and AAAA players from other teams; when a pre-existing team that was independent or in a lower-prestige league of some kind switches leagues, it brings its existing mix of talent into the league it joins. (Analogies to conference membership in college football come to mind.) Because the Seamheads database includes records from both league and non-league teams, wherever the data is available, it gives us some view into top-level Black baseball beyond the league structure.

The minimal effect of expansion on the quality of play is particularly the case with respect to the creation of the ECL in 1923. For top-level Black Baseball in 1922, Seamheads includes data from 11 independent teams. In 1923 six of those teams (one with a change of name), form the Eastern Colored League. So the ECL isn't in any way an expansion of teams: it formalizes the relationship between the teams to raise their profile. Many (if not all-given the incompleteness of records it's impossible to tell without research beyond Seamheads) of the 1922 non-league teams continue on in 1923, with some of them employing highly talented players. As part of its profile enhancement, the ECL teams do begin "raiding" NNL teams for their star players, so the creation of the ECL does shift competition levels between the now-two top leagues somewhat, but the degree of change is much less than is implied if one simply says, "In 1922, there were 8 major-league teams, and in 1923, there were 14. That must have brought in a lot of less talented players." The creation of the ECL is the extreme case, of course, but it still provides a telling example of why, if one really wants to track what is going on with competition levels in Black Baseball, one has to track what's happening with teams and who is playing on them in addition to what's happening with leagues.

All of this movement is intricate and chaotic, and I don't pretend to have a clear mental map of all that was going on: I would just caution against applying analytical frameworks based on the economic organization of the National and American Leagues and their farm systems to Black Baseball without adding a lot of qualifications.

I'll conclude by saying that in the aggregate the case that competition levels in the top leagues of Black Baseball were lower than in the National and American Leagues is strong, not least because there was less stratification within Black Baseball: in 1922-23, for example, the most talented players were spread among 25 teams tracked in the Seamheads database and beyond. There is just no corresponding situation in the National and American leagues where a top player like Joe Williams up and goes to a non-league Homestead Grays or where another top player like Ben Taylor goes to the non-league Washington Potomacs to be their manager, or when Roy Parnell stays in the southern leagues to manage at the peak of his playing career rather than staying in the Negro National League, not to mention the situation when a bunch of top players go to Santa Domingo in 1937 or to the Mexican Leagues in the early 1940s. Even if one believes that the overall talent level, player for player, in the Negro Leagues was as high as in the National and American Leagues, factors like these necessarily prevent the kind of talent concentration that maintained competition levels in the National and American leagues. The different and looser organization of Black Baseball led to a more diffuse distribution of the most talented players.

@318: The second point is certainly up for debate. There probably won't ever be a clear answer. However, the first and third are not really matters of opinion. To disagree profoundly is...your prerogative, I suppose.

@Dr. Chaleeko

In your current version of the Hall Report Card spreadsheet, it looks like you've assigned all pre-1947 Black and Cuban players to the National League.

Is this correct? If so, then this inflates Report Card Scores for segregated AL players relative to their peers because they have less competition in the multi-season achievement categories.

was hoping Kiko would share whether he had any updates from his Retrosheet or Negro League projects before finalizing

My apologies for not contributing here as much this year. For those who haven't heard, I've been pretty busy with my new role at Retrosheet. In case anyone missed it, we did our semi-annual release within the past couple of weeks. We now have partial play-by-play back to 1914, full play-by-play (including deduced games) back to 1921 (102 total years - a Retro-century!). And game accounts for Negro League games back to 1942.

I was going to write some more about the Negro Leagues stuff but may save it for the 2024 HOM Discussion Thread.

I have not updated my Player won-lost records in a few years now. I really need to do that. Hopefully, I'll be able to make some time for that this spring.

I was planning to vote, but haven't really had a lot of time to do any significant analysis. So, my plan was to start from last year's ballot, remove the guys we elected and add any interesting newcomers - which is easy enough because I think there's only one interesting newcomer.

Here's my preliminary ballot. I'll probably move it over the the ballot thread early next week.

Happy Holidays to all of you!

1. Carlos Beltran
2. David Ortiz
3. Tommy John
4. Tim Hudson
5. Vern Stephens
6. Lance Berkman
7. Jason Giambi
8. Wally Schang
9. Darryl Strawberry
10. Urban Shocker
11. Tommy Henrich
12. Toby Harrah
13. Bert Campaneris
14. Dave Concepcion
15. Dizzy Dean

Required disclosures:

Berkman, Bando, Ortiz, and John are on my ballot
Bonds and Johnson are in my top 40, probably in my top 30
Munson is probably to 50 or so; I might be under-rating him
Bell is probably outside my top 200. I've discussed him extensively; my system just isn't a fan.

Willis and Taylor are inadequately handled by my system which isn't an excuse for excluding them (my system is also missing significant portions of the careers of Schang and Shocker and I've voted for Johnny Evers, among others, in the past).

My reading of Dr. C's latest MLEs on Ben Taylor leave him outside of HOM-worthy. But I'm certainly open to being wrong about that.

Vic Willis is the type of pitcher that my system tends to like and I've had him on-ballot in the past. Given the perpetual eligibility and the knowledge that eventually Retrosheet will get back far enough that I will have a better handle on Willis's full career, I feel more confident in the HOM-worthiness of the 15 players that I've listed. But I will certainly re-evaluate Willis's case as I get more data.

I don't think there are any first-year eligibles outside of Beltran who seem worthy of comment (no offense to any of them).

The second point is certainly up for debate. There probably won't ever be a clear answer.

My book spends hundreds of pages addressing this very question.

However, the first and third are not really matters of opinion.

Um... clearly they are, because my opinion differs from yours.

Your opinion differs from that expresed by numerous MLB managers and players at the time. It's also flatly contradicted by the results in actual play when MLB integrated.

I'm not going to re-litigate issues I've spent years analyzing and writing up. Once my book comes out (this should happen in April), you are of course welcome to dispute specific arguments I make therein.

For now, I merely note that my evaluation of NgL players diverges radically from yours - and so, therefore, will my ballot.

Happy Hollandaise everyone!

Two cents:
“A 50-percentile NgL player is not a 50-percentile MLB player, I'm pretty we sure we all agree (we should, anyway)”
This requires a legalistic sort of answer. The reasons are twofold per my research. One is that the NeLs concentrated more PAs in their best hitters than MLB did. But they also gave a higher percentage of PAs to very bad hitters (wOBA < .290). So the mean performance might be similar to MLB but it doesn’t mean the league had the same top-to-bottom quality.

Also, the STDEV of both wOBA and RA9 is wider than MLB. Perhaps most importantly my latest QoP research, using a really strong third-party method, shows that QoP was always below MLB overall, though it got wicked close a couple of years. More on this in a moment.

“From 1920 to about the mid/late 1930s [I would rate the NgL] starting from 70% [of MLB]”
My research, noted a moment ago, says that this is too low a starting point. If my work is accurate, and I don’t see any glaring input errors so I feel good about it, AAA is a more accurate starting point since the very early 1900s.

“We know that the worst players in the Negro Leagues were typically not good enough to make it in MLB.”
I’m my opinion, this is absolutely true. And we know that for lots of reasons. And it stands to reason. As a thought experiment, we know that a replacement player in MLB is roughly an average AAA player. I have not calculated what a AAA replacement player truly is, but theoretically and empirically, it’s a AA player. Let’s say the average NeL QoP was .9 of MLB. That would mean that the replacement player in the NeL was worse than AA by some baseball equivalent of the transitive property. Is that thought experiment result reasonable? Well, I think so. How might we assess this? Here’s one thought. The NeLs tried again and again to create seven or eight team leagues. These attempts repeatedly failed. Why? A couple reasons. One, talent was not evenly distributed around the league. For example, Cubans rarely played on teams that weren’t touring Cuban teams (Cuban Stars, Stars if Havana, Pollock’s Cuban Stars, All Cubans, whatever). And there was usually an eastern and a western Cuban team. So a lot of that talent pool didn’t diffuse around the other 12 teams in the league. Why is that important? Because it meant the talent pool of other six teams was really stretched. What this led to was repeated collapse of franchises in Ohio (esp Cleveland) or wherever the weak link was (Indy, Atlanta, lesser Chicago team sometimes). There weren’t enough good players around and replacement-bad teams don’t make money. The most stable period in NeL history was the NAL/NNL period during which the leagues quickly pared down to six teams each.

In the 1970s, Black participation in MLB topped out around 30% of all players (IIRC). In a league with 24 teams that’s about 7 teams worth of players. If I’m remembering incorrectly and it’s 40%, that’s 10 teams. But let’s say it’s 30% for the moment. That’s around 1976. In 1937, there was not much competition from other sports at all. So it’s not difficult to imagine that that there was 8-10 teams of MLB caliber talent (from replacement to remarkable). But the NeL then we’re running 12-14 total teams. There’s a gap there that could represent 25-50 players who were not MLB talents. If my logic is even close to accurate, it would clearly support the idea that replacement level in the NeL would have to have been lower than in MLB.

In general, the ideal situation is one in which the four things align: The statistical record, analysis of the record, our theoretical understanding of the game, and the narrative history/contemporary opinions. When they don’t all align, it is, IMO, more likely to
Provide a satisfactory result if we place far more weight on synthesis of the first three things than the fourth because the “error bars” on narrative/opinion is far, far, far wider and more slippery. But that’s just me.

I was going to wait for the new Discussion Thread to talk Negro Leagues but maybe I'll go ahead and chime in here.

We know that the worst players in the Negro Leagues were typically not good enough to make it in MLB

I think this depends on exactly what you mean by "MLB". If you mean "leagues recognized as 'major-league' by MLB", this statement is factually false. If you mean, "the segregated AL/NL if they were suddenly integrated but not expanded, this statement is factually true, but it is also factually true that there are players who played in a segregated white AL and/or NL who would not have been good enough to play in an integrated 16-team "MLB", which makes it a true, but ultimately meaningless, statement.

The interesting question, then, is - what if MLB had expanded to accommodate the larger player pool. How would the mix of White/Black players in such a league have compared to the mix of White/Black players in the segregated leagues of the time? And here, I think the most likely answer is that the racial mix wouldn't have looked very different at all.

The earliest Negro League season for which Retrosheet has done a deep dive is 1942. So, I'll use that season as my example.

In 1942, it appears that there were 454 players who played at least 5 games in the AL/NL (I get 512 total players). For 16 teams, that works out to 28.4 players per roster (32 if you include 'cup-of-coffee' guys). Retrosheet has found 315 players who played in at least one game for an official Negro League team in 1942. But only 233 of these players played at least 5 games. There were 12 Negro League teams, so that works out to 19.4 players per Negro League roster, excluding cup-of-coffee guys. [The Negro Leagues did have more cup-of-coffee guys. It seems clear to me that such players are of relatively little interest in the context of discussing the Hall of Merit. YMMMV]

So, excluding your cup-of-coffee guys, we have a total of 687 "major-league" players (454 AL/NL, 233 NAL/NNL), 66.1% of whom were White. That percentage that is White is broadly similar to the racial mix of players since the 1970s or so (there are far fewer African-Americans today but they have largely been replaced by Afro-Latinos). The top Google hit on the current racial mix of MLB says "players of color constituted about 38% of players in the league on Opening Day 2022."

Suppose we used the 1942 AL/NL roster size as our guide - 28.4 players, excluding cup-of-coffee guys. How many teams could you fill with 687 players (all of the non-cup-of-coffee guys from any of the four "major" leagues in 1942)? The answer is 24.

So, if we imagined an integrated "MLB" in 1942 with 24 teams, there would theoretically be room for everyone who actually played for an AL/NL/NAL/NNL team to fit on a roster of said integrated league. Could a few White guys who weren't good enough to make the AL/NL have nudged a few Black guys out of roster spots? Maybe. But could a few Black guys who weren't good enough to make the NAL/NNL have nudged a few White guys out of roster spots? Also maybe.

In terms of "does it make sense to think about 24 'major-league' teams in 1942", let's compare the 1910 and 1940 Census.

In 1910, there were 16 "major-league" teams. The U.S. Census reported a population of 92.2 million, 88.9% of whom were white. So that'd be 16 teams drawing from a population of about 82.0 million (88.9% of 92.2 million). That's 5.125 million people per MLB team.

In 1940, the U.S. Census reported a population of 132.2 million (89.8% of whom were white). One team per each 5.125 million people in the U.S. in 1940 would be 26 teams. And that's not even taking into account the increased number of Latin American players who were playing in 1942 (overwhelmingly in the Negro Leagues). [Obviously, one team per 5 million people would imply a vastly larger MLB in 2022 than currently exists.]

Bottom line: I don't think you can get to "the AL/NL were clearly stronger than the NAL/NNL" or any sub-variant of that (the average player was better, replacement was lower, etc.) from a purely theoretical, logical path. Rather, this is an "empirical question". I use that term a lot in my work (I'm an econometrician); it means that you have to actually do the math to figure out the right answer. And I will say that I have a great deal of respect for the math that Dr. C., in particular, has done on this topic.

Lots of highly interesting information in Dr. Chaleeko's and Kiko Sakata's recent posts! I have three "empirical questions" to ask in response to them.

First, Dr. Chaleeko, is your assertion about the connection between poor player quality and team failure based on empirical evidence from the history of major teams in Black Baseball?

You wrote: The NeLs tried again and again to create seven or eight team leagues. These attempts repeatedly failed. Why? A couple reasons. One, talent was not evenly distributed around the league. For example, Cubans rarely played on teams that weren’t touring Cuban teams (Cuban Stars, Stars if Havana, Pollock’s Cuban Stars, All Cubans, whatever). And there was usually an eastern and a western Cuban team. So a lot of that talent pool didn’t diffuse around the other 12 teams in the league. Why is that important? Because it meant the talent pool of other six teams was really stretched. What this led to was repeated collapse of franchises in Ohio (esp Cleveland) or wherever the weak link was (Indy, Atlanta, lesser Chicago team sometimes). There weren’t enough good players around and replacement-bad teams don’t make money.

From my own limited knowledge of the historical situation, it seems to me that the correlation between team talent and team viability may not have been as direct as this statement implies. I'm also unsure that an insufficiency of talent to support sixteen competitive teams was the primary reason that a stable suite of such teams never emerged, as there were a variety of other factors involved in player movement.

Second: Dr. Chaleeko, when you say that the NeLs concentrated more PAs in their best hitters than MLB did, to what extent is this fact (which I am sure you have established empirically) due to a different style of player usage, to what extent is it an artifact of league economics, and to what extent is it an artifact of data preservation? With respect to league economics, the top teams with the best players tended to play more league games. With respect to data preservation, the top teams tended also have more detailed records printed of their games. Are those the factors that concentrate NeL PA in the best hitters, or are there others? If one looks at the records of a top team over the course of a season, are the PAs of that team more concentrated among top players than they would be for NL/AL teams? Given that the competition level we are interested in is the competition level of play in the games of which we have detailed statistical records, it may not matter for HoM purposes if the concentration of plate appearances is an artifact of data preservation and/or league economics, but if it is a result of these factors, we should think through the implications of that fact. To the extent that it is due to a different style of player usage, the implications of that usage pattern for player performance seem likely to be important. So, question 2A, did the NeLs concentrate more IP in their best pitchers than MLB did (I am pretty sure that the answer is yes, at least for top-league-level play), and what are the implications of that for how we interpret pitcher quality in the context of NeL play?

My third question follows from that one, and is a general informational question rather than one that follows directly from these two recent posts: to what extent are there sources available that document exhaustively what a season or a sequence of seasons of play was like for Negro-League teams? I was reading the SABR bio of Willie Foster this morning, and it quotes him describing how his team would play nine games in two days:

"I played with the Homestead Grays in ’31, and we had nine starters. I pitched on a Friday, pitched my ball game and finished it. Do you know that Sunday evening it was my turn again? Everybody had pitched and everybody had stayed in the ball game? Yes sir, we played nine ball games in two days!"

I don't know if this statement should be taken purely at face value, but it nevertheless highlights the fact that the pattern of play in the Negro Leagues was different from the pattern of play in the NL and AL at the same time. I'd like to have a better macro-scale sense of what that pattern was so as to consider how to account for it in interpreting the statistical record of Negro-League players. Are there sources that provide this kind of detail at the seasonal level of just how many games NeL teams played against all levels of competition and how those games tended to be paced across the baseball season?

As usual, very thoughtful comments, Eric.

Happy Hollandaise everyone!

Ooooh, that’s a cheesy joke.

“A 50-percentile NgL player is not a 50-percentile MLB player, I'm pretty we sure we all agree (we should, anyway).

…The reasons are twofold per my research.

This implies, Eric, that you agree with the contention that MLB > NgL – yes? Don’t allow me to misquote you, please correct me if I’m wrong. If so, however, this explains two tautologies that I perceive in your arguments.

First tautology:

Let’s say the average NeL QoP was .9 of MLB. That would mean that the replacement player in the NeL was worse than AA by some baseball equivalent of the transitive property.

Your initial “let’s say” biases the result. You assume that NgL QoP is less than MLB, which leads to the conclusion that, um, NgL QoP is less than MLB. It’s a feedback loop, A = A.

Second tautology:

We know that a replacement player in MLB is roughly an average AAA player. I have not calculated what a AAA replacement player truly is, but theoretically and empirically, it’s a AA player.

Agreed. However:

It’s not difficult to imagine that that there was 8-10 teams of MLB caliber talent (from replacement to remarkable). But the NeL then we’re running 12-14 total teams. There’s a gap there that could represent 25-50 players who were not MLB talents.

Again, the foundational assumption of NgL inferiority biases the thought experiment. The statement “it’s not difficult to imagine” tells the story.

There’s also a mathematical bias, inherent in the nature of NgL play, that must be compensated for to obtain valid results:

The NeLs concentrated more PAs in their best hitters than MLB did. But they also gave a higher percentage of PAs to very bad hitters (wOBA < .290)…Also, the STDEV of both wOBA and RA9 is wider than MLB.

Sample size, sir, sample size. Short seasons lead to high StDevs.

Most NgL seasons were about the length of a modern MLB month. Check out the incredible variations in MLB play month-by-month, rather than season-by-season, and you’ll see precisely the same kinds of StDevs as in NgL play.

My latest QoP research, using a really strong third-party method, shows that [NgL] QoP was always below MLB overall, though it got wicked close a couple of years.

I can’t comment on your research, because I haven’t seen it. Your research is generally outstanding; it's one of the basic reasons I hang out here.

Having said this, I asked a question earlier in this thread, and its answer may illuminate (at least for me) your thinking on these issues: How do you compare, say, QoP in 1900 MLB to 1920 MLB?

One other point:

The NeLs tried again and again to create seven or eight team leagues. These attempts repeatedly failed.

The reasons you cite for this hold validity, certainly. But the absolute #1 reason isn’t mentioned: lack of capital. Consider Black owners’ access to bank capital in any era of NgL play as compared to their white counterparts in MLB, and the true economic circumstance of the Black leagues becomes readily apparent.

As to the general nature of your belief (assuming you hold it – again, please correct me if I’m wrong) that NgL play was inferior to MLB play:

In general, the ideal situation is one in which the four things align: The statistical record, analysis of the record, our theoretical understanding of the game, and the narrative history/contemporary opinions. When they don’t all align, it is, IMO, more likely to provide a satisfactory result if we place far more weight on synthesis of the first three things than the fourth…

* With NgL play, the statistical record is now largely stable.

* Analysis of that record, though, is in its infancy; hence, this debate.

* Our theoretical understanding of the game is, I think, tied up in the outcomes of the analysis.

* Narrative history and contemporary opinions are split exactly down the middle. Here’s a simplification that generally holds true: The commissioners and owners of the 1920s-30s-40s believed that Black ballplayers were inferior to white ballplayers of the same era. The players and managers of the same era held exactly the opposite opinion.

Now, as you yourself observe:

The “error bars” on narrative/opinion [are] far, far, far wider and more slippery.

True, of course - but now ask yourself: Why did almost 100% of MLB executives and owners both oppose integration, and insist that Black players were inferior to white players? Is that alignment of opinion coincidental?

At the same time, ask yourself: Why did almost 100% of MLB players and managers believe that Black players were the equal of their white counterparts?

Ask yourself: Why was the divide between the owners and the men on the field so absolute?

Ask yourself: Where did the owners’ vested interests lie?

And then ask yourself: Who were the more qualified observers of the true quality of Black baseball talent – the executives and owners, or the players and managers?

But that’s just me.

Amen, brother.

Are there sources that provide this kind of detail at the seasonal level of just how many games NeL teams played against all levels of competition and how those games tended to be paced across the baseball season?

Retrosheet has prepared game logs for individual teams - start at our "Negro League Seasons" page, click the season you'd like, then the team you'd like.

Here are the 1942 Homestead Grays. We have 112 games they played(*), for which we have found box scores for 77 of them (Seamheads - and, hence, Baseball-Reference - only include games for which they have found a box score in calculating their season statistics). From that page, you can click through to the roster - and get player logs for the players (e.g., Josh Gibson's 1942 batting log) - or a team game log.

(*) Three caveats

(i) We have only made it back to 1942 (working backward in time). Our current pace is 5-6 seasons per year which would mean we're 4-5 years from finishing the entire history of the formal Negro Leagues.
(ii) We only include games played between "major-league" caliber teams. We don't necessarily limit that to members of the NAL or NNL (e.g., we have some games the Clowns played against NAL teams in 1942; they didn't join the NAL until 1943), but Negro League teams played a lot of games against semi-pro teams. We are missing these games (for now).
(iii) We have only included games that we know about. I know this seems sort of self-evident, but it was not uncommon for a newspaper to advertise that two Negro League teams would be playing a game in a few days (or that night) but would never follow up with the final score (or even whether the game was actually played). We only include games if we are reasonably certain about the date, location, and final score.

Great, high-level discussion here. This is the sort of debate that NgL analysis needs going forward.

Are there sources that provide this kind of detail at the seasonal level of just how many games NeL teams played against all levels of competition and how those games tended to be paced across the baseball season?

Just for clarity, allow me to note that the arguments I advance are based solely on NgL-vs.-NgL play, and exclude all barnstorming, exhibition play, winter ball, games vs. white Major Leaguers, etc. Doing otherwise would be akin to, say, analyzing Dizzy Dean in part on his performances with the Dizzy Dean All-Stars: interesting historically to be sure, but incompatible with the canonical data set.

I definitely look forward to seeing the new research from Dr. C and Kiko.

Oh, if anyone else is having trouble posting: it let me post my last comment on the first page of this thread, but not any of the other pages of this thread. One more thing to try/look for. It's amazing that BBTF hasn't given up the ghost given how long it's been in these throes...

It's also flatly contradicted by the results in actual play when MLB integrated

This refers, presumably, the 50th percentile MLB vs NgL player debate. I'd be surprised if many 50th percentile NgL players even got integrated? I am guessing this means that by the 70s or maybe 60s, there was a total count of black players in MLB that would have amounted to the 50th percentile in NgL if NgL had still existed at that time. If that's not what you mean, I'd be interested in an analysis of NgL players by percentile and how they performed as a group in MLB. That would be really illuminating I think. I think the difficulty will be finding enough players with play in both leagues, but maybe there's a creative way to examine it that I'm not thinking of.

Have an historical question to ask and know not where else to put it:

What's up with the BT (bases taken) stat on BaseballReference? It jumps in 1954. It pretty much doubles. I presume this is some sort of methodological change?

At cookie on ballot thread post #9.
On Bob Johnson being overlooked, the early days of the HOM were dominated by win shares and old versions of Baseball Prospectus. Johnson was on some terrible teams and was punished by win shares. The opposite held for Joe Medwick, whom was elected and someone I find lacking comparatively. Earl Averill has some of this going on, though how much credit he nets from the PCL and how to messure his defense make a large impact too. Cool Papa Bell was elected from this era as well and looks like Johnny Damon now.

BR WAR versus past flavors is huge for Thurman Munson as well.

Good call out on Babe Adams as a HOM worthy guy, though I still have Urban Shocker a bit ahead with WWI credit.

Posada is tough, even taking some air out of his awful defensive stats and he's short.

On Tinker, BR and BG WAR ahead of WS. Same for almost twin Art Fletcher.

deleted

I have been thinking about how fragmented our backlog has become with 150 years of history to cover in just 15 ballot spots. Would voters be interested in extending the ballot to 20 slots going forward? It wouldn't really affect scoring much - we would score slots 16-20 as 5-4-3-2-1 points. It would give more weight to the player with near-ballot support rather than the player who has strong support from a small faction. Seeing how our number of active voters keeps decreasing and our number of players to consider keeps increasing I think it would show more consensus in our selections. 2026 and 2027 look like heavy backlog elections.

@: "Would voters be interested in extending the ballot to 20 slots going forward?"

+1.

I'd be ok with a 20 man ballot. I mentioned in another post that I also think we should have two tier voting, with the second vote being from voters on only the top 10-15 after the first round of voting. I think we'd get a more clear result.

I definitely support 20 man ballot.

@334: I assume this has to do with the availability of complete gamelogs or stat tracking for bases taken? I would not assume a methodology that is different between pre-1954 and post-1954.

I support a 20 man ballot as well.

no complaints on 20-man ballot idea

I also think 20 votes is a good idea moving forward.

I also support a 20 man ballot.

We have 11 ballots, and only Esteban Rivera has indicated an intention to vote who hasn't yet. I guess it wouldn't really be a HOM project without a deadline extension.

I expect to see ballots from Mark Schirk, Ardo, Chris Cobb, Mike Webber and Rob Wood.

I've reached out to Brent, Carl Goetz, Michael J. Binkley, Rob Wood, adarowski, Patrick W, HAWK, John Murphy, JoeD, Al Peterson, Mark A Shirk, James Newburg, Qufini, Ardo, theorioleway, Chris Cobb, Mike Webber, and kwarren. I think that's everyone who voted last year who hasn't this year (plus kwarren), exclusding Esteban since he already checked in here.

Hey there. I am planning to post a ballot this year despite missing last year (which was only because I screwed up the deadline. I have put that in giant letters on my draft ballot this year so that really should not happen again.)

2023 prelim ballot

I was hoping to upgrade my rating system this year, but it didn’t get done, so I’m sticking with the same (rWAR-based) system I used last year. I adjust for season length, military service, league quality, and post-season performance, give minor league credit for seasons that would be above major league average, and judgmentally adjusts some of rWAR’s positional factors for certain time periods. I’m a peak/prime voter and pretty much ignore seasons that are average or below.

1. Carlos Beltrán. In 65 post-season games, he had a 1.021 OPS.
2. Bobby Bonds. Over the 11-year period from 1969 to 1979, he averaged 149 games a year with a 132 OPS+ and 5.0 WAR.
3. Sal Bando. Over the 10-year period from 1969 to 1978, he averaged 156 games a year with a 127 OPS+ and 5.7 WAR.
4. Buddy Bell. Over the 12-year period from 1973 to 1984, he averaged 4.9 WAR with an outstanding glove.
5. Kevin Appier. Over the 8-year period from 1990 to 1997, he averaged 205 IP with a 140 ERA+.
6. John Olerud. Averaged 4.7 WAR over the 10-year period from 1993 to 2002.
7. Kirby Puckett. Averaged 4.5 WAR over the 10-year period from 1986 to 1995; had a .897 OPS in 24 post-season games.
8. Willie Davis. A great fielder; he struggled as a hitter in what should have been his prime due to the expanded strike zone but still managed to have 7 seasons with at least 4 WAR.
9. Jason Giambi
10. Lance Berkman
11. Bernie Williams
12. Phil Rizzuto. An excellent defensive shortstop, I credit him as about a 5 WAR/yr player for 1943-45.
13. César Cedeño. Didn’t do much after age 29, but he was a fine player before then.
14. Roy Oswalt
15. Mark Teixeira. From 2004 to 2012, he averaged 150 games a year with a 134 OPS+, good fielding, and 5.0 WAR.

Numbers 16 through 20 are David Ortiz, Tim Hudson, Fred McGriff, Thurman Munson, and Chuck Finley.

Required disclosures:

Thuman Munson – just misses my ballot at # 19.

David Ortiz – just misses my ballot at # 16.

Ben Taylor – I’m going to discuss a couple of disagreements I have with Dr C’s MLEs for Taylor:
1) At the beginning of his career (1909-12), Taylor was primarily a pitcher, but the MLEs evaluate him as if he were a first baseman. Taylor record as a pitcher was near or below replacement level, so had he been in organized baseball during that period, I think he would have stayed in the minor leagues. It’s not that unusual for a player to switch to or from pitching early in his career, and we don’t (for example) give major league credit to Stan Musial as a hitter for his three minor league seasons spent trying to learn to pitch. Dr C credits Taylor with 9.8 WAR from 1910 to 1913 (1913 being the year he transitioned from pitcher to first baseman). I think MLEs should be based as much on possible on the actual careers and contributions of the players. I think it would be more appropriate to show no MLE WAR for Taylor for 1910-12 and about 1.6 WAR for 1913 (for a bench/part-time role during his transition season).
2) Toward the end of Taylor’s career (say ages 36 to 39, or 1925-28), in my opinion the MLEs inflate his value. (I think this also happens with other players at the end of their careers; it’s not just a problem with Taylor’s MLEs.) There are two issues going on. First, for some reason they show Taylor as a better hitter in MLE terms than I think he was. For the four seasons in question, my back-of-the-envelope formula converts his Seamheads NeLg OPS+ of 116-126-120-101 to MLE OPS+ of 92-101-96-79. In other words, over this 4-year span I estimate that Taylor was a below-average hitter with an average MLE OPS+ of 92, whereas Dr C’s MLEs show him as still an average or above-average hitter in 3 of the 4 seasons. The second issue is that as an older and (in my opinion) below average hitter, I think Taylor’s MLE plate appearances should drop off more rapidly than shown in Dr C’s MLEs. Compare, for example, Gil Hodges, who from ages 36 to 39 averaged just 161 plate appearances per season with an 84 OPS+. With my estimated MLE OPS+ of 92 for Taylor over that age period, I’d expect more than 161 plate appearances per season but many fewer than the 538 PA/season assumed by Dr C. I’ve adjusted the MLE WAR for those 4 seasons down from Dr C’s 7.8 to 3.2.

The net effect of my adjustments is to reduce Taylor’s career WAR from 62.5 to 51.7. I have him ranked below several recent first baseman (Olerud, Giambi, Texeira, Cash) and about the same as his near contemporary, Chance.

Vic Willis – just misses my ballot at # 22.

Bob Johnson – His raw statistics are inflated by weak war-time league quality. He doesn’t do well in my system and is not really close to my ballot.

Tommy John – He pitched for 26 seasons, but in 19 of those seasons, his WAR didn’t exceed 3.0. That’s the kind of career that my system doesn’t think much of.

Love seeing all the prelims, let's get some of those moved over to the ballot thread so our counters can get the election counted on time. If people need help because of the site I can usually post.

I have been thinking about something to do now that MMP voting is finished. We last ranked the players by position in 2008-9. Since then we will have added 50 players to the Hall of Merit including some inner-circle players. Are people interested in re-ranking players by position in 2023?

Previous rankings

catchers
first basemen
second basemen
shortstops
third basemen
left fielders (we moved Monte Irvin after the election to LF)
center fielders
right fielders

pitchers

We also ranked pitchers by era before combining them.

@351 I'd be down for it

Players added and positions

C - Mike Piazza, Ivan Rodriguez
1B - Jeff Bagwell, Rafael Palmeiro, Frank Thomas, Jim Thome, Todd Helton
2B - Roberto Alomar, Craig Biggio, Jeff Kent
SS - Barry Larkin, Derek Jeter, Alex Rodriguez
3B - John McGraw, Edgar Martinez (no point in ranking DH), Chipper Jones, Scott Rolen
LF - Rickey Henderson, Barry Bonds, Manny Ramirez
CF - Ken Griffey Jr, Jim Edmonds, Andruw Jones, Kenny Lofton, Carlos Beltran (assuming)
RF - Reggie Smith (CF?), Larry Walker, Gary Sheffield, Vlad Guerrero, Bobby Abreu, Sammy Sosa
P - Kevin Brown, Rick Reuschel, David Cone, Roger Clemens, Greg Maddux, Tom Glavine, Randy Johnson, Pedro Martinez, Curt Schilling, Mike Mussina, John Smoltz, Roy Halladay, Mariano Rivera, Dick Redding, Luis Tiant, Johan Santana, Andy Pettitte

+2 more from this year (Berkman, Ortiz, Bell, Munson, Bando?)

Looks like the only question is whether Reggie Smith is a CF or a RF

I'd definitely be up for a re-ranking!

I think a re-ranking sounds great. I wasn't around for the first one. No idea how I'll fit in with the consensus (if at all!), but I'm game to find out.

I'm not getting a window at the moment on the Ballot thread, so I am posting this here with ballot-counting short version first. I will try to get it into the ballot thread myself tonight or tomorrow, but if I can't get it there, please count this submission.

2023 Ballot Summary

1. Carlos Beltran
2. Buddy Bell
3. Tim Hudson
4. Jason Giambi
5. Bobby Bonds
6. Kevin Appier
7. Brian Giles
8. Lance Berkman
9. Urban Shocker
10. Orel Hershiser
11. Vic Willis
12. Robin Ventura
13. Sal Bando
14. Bucky Walters
15. Babe Adams

See next post for explanations of non-balloted returning top 10 players.

Detailed 2023 Ballot

SYSTEM REVIEW

I cast my first ballot in 1903 and voted in each election through 2010. I missed 2011 and 2012, voted in 2013 and 2014, then missed 2015-19. I returned to voting in 2020, so 2023 will be my 114th election.

I rank players within decade-by-decade cohorts, using a metric that combines career value as measured by wins above replacement, prime value as measured by wins above average on a seasonal basis (roughly seasonal WAR minus 2, negative values treated as zero), and peak quality, as measured by a player’s rate of WAR over a period of 5 or more consecutive seasons. I integrate the decade-by-decade rankings by considering primarily the players’ rank within their cohort, re-scaled to put all decades on a scale of 30, where 30 marks the in-out line. For players in nearby decades, I also compare their raw scores, but playing conditions often change so much from one decade to the next that the contextual value achievable by top players often varies considerably in ways that go beyond quality-of-competition changes, so I place most weight on each player’s ranking against their contemporaries.

The Lie of the Land

I am in the process of re-examining all players with 30 or more WAR or the equivalent. I was hoping to complete this process before this election, but I still have the pre-1890 and post-2010 periods to do, and I need to do a second pass through the Negro Leagues to fine-tune the analysis based on what I learned from the first pass. As I see things now, there are 32 eligible players who are at or above the HoM in-out line. I’ve ordered the ballot following my scaled rankings closely but not formulaicly. There are another 50 players or so who are within 5% of the in-out line, so except for a few at the very top, everyone is very tightly packed. Squinting at the list a little can result in players moving 10 or more spots in the rankings.

THE BALLOT

1. Carlos Beltran. Newly eligible. (2000s. System score: 140.10. In-Decade Scaled Rank: 12.5). While Beltran is not an All-Time Great, he’s an easy #1 on this ballot. His in-decade rank is well ahead of everyone else, and his system score also is also tops among eligible players. DL mentioned that he sees Beltran as slightly ahead of Jim Edmonds: my analysis concurs with that assessment.
2. Buddy Bell #3 in 2022. (1980s. System score: 136.8. In-Decade Scaled Rank: 17.1). Excellent third baseman in a period in which third basemen excelled.
3. Tim Hudson #10 in 2022. (2000s. System Score: 115.3. In-Decade Scaled Rank 25.5). Hudson’s well-rounded game places him above Mark Buerhle, Roy Oswalt and the now-elected Andy Pettitte in my pitcher rankings for 2000s decade. He’ll be the best remaining eligible pitcher from that decade until CC Sabathia reaches the ballot. I nudge him ahead of Giambi because of the amount of hang-around value included in Giambi’s career.
4. Jason Giambi #9 in 2022 (2000s. System Score: 116.35. In-Decade scaled rank 24.5). His batting peak was great, but not Frank-Thomas-great, so the fact that he was not good in other aspects of the game leaves him still some distance from election, though he should make it eventually.
5. Bobby Bonds #5 in 2022. (1970s. System Score: 123.9. In-Decade Scaled Rank: 29.4). Complete player with a fine prime but short career. Looks increasingly likely to be elected. Bonds gets a little bit of a boost in the rankings for being a top hitter in a period in which it was very tough to accumulate hitting value compared to pitching value. His scaled rank among 1970s position players is 17.6, slightly ahead of Giambi’s 18 among 2000s position players, even though his scaled rank among all players trails Giambi’s by 5, 29.4 to 24.5
6. Kevin Appier #4 in 2022. (1990s. System Score: 125.6. In-Decade Scaled Rank: 26.2). Very similar in value to the now-elected Johan Santana. Late 90s expansion helped to extend his career a bit after his prime. That and his relatively low pitchers-only ranking for the 1990s of 9.3 (lowest of balloted pitchers) brings him down a little bit from his 2022 spot.
7. Brian Giles #7 in 2022. (2000s. System Score: 117.72. In-Decade Scaled Rank 26.5). My system likes him a little bit better than Lance Berkman, who has been getting somewhat more attention from the electorate, because of Giles’ stronger consecutive peak and higher defensive value. I’m not sure why Giles is so far under the radar. Will he become the Bob Johnson of the 20-aughts?
8. Lance Berkman #14 in 2022. (2000s. System Score: 114.65. In-Decade scaled rank 28.5). Very similar to Brian Giles. Fielding value pushes Giles ahead.
9. Urban Shocker #24 in 2022. (1920s. System Score: 126.3. In-Decade scaled rank 22.0). Like Willis above, I’ve found that Shocker is too far above the in-out line to leave off of my ballot on period representation or competition-level considerations. If electorate of 15 years ago had been voting based on the current metrics, Shocker would probably have been elected, and Eppa Rixey would probably not have been elected. Now we have to deal with the fact that putting Shocker in would mean leaving out a deserving candidate from later eras. If Shocker hadn’t had a fatal heart condition and had lived to pitch longer, he’d have been elected easily, like Stan Coveleski. He was good! However, he also had a unusual competitive advantage in being allowed to throw a high-value pitch—the spitball—that was denied to most other pitchers. If he were elected, the NL-AL pitchers from the 1920s would be 50% grand-fathered spit-ballers: Coveleski, Faber, and Shocker (the others being Alexander, Vance, and Rixey, as I divide the decades). I don’t think there would be anything wrong with that, but it’s a telling example of how significant even an apparently small competitive advantage can be in enabling some players to create more value than others. Shocker’s pitchers-only scaled rank is 6.6 for the 1920s.
10. Orel Hershiser #8 in 2022. (1980s. System Score: 118.5. In-Decade scaled rank 28.8). Outstanding peak in a decade that was hard on pitchers. My decision to bring three overlooked pitchers onto the ballot has pushed Hershiser down a little bit, but his high pitchers-only ranking for the 1980s (5.9) keeps him in a mid-ballot spot.
11. Vic Willis #27 in 2022. (1900s. System Score: 121.84. In-Decade scaled rank 20.3). Moves up in my rankings and back onto the ballot after many years of absence due to my thorough overhaul of my calculations. There are reasons to suspect that an RA/9 evaluation of Willis overrates him, but even when I make sure I am careful to dock him for everything I can document, he still ends up far above the in-out line. His career profile is very different from those of Rube Waddell and Joe McGinnity, but his value ends up being very similar to theirs. Unless we are only going to see Mathewson/Plank-level deadball pitchers as HoMers, which I think would be too high of a standard, especially given our liberality in electing pitchers from the little-dead-ball era of the 1960s and 1970s, then Willis belongs—at least until Retrosheet-level analysis finds to the contrary. His in-decade scaled rank of 20.3 would put him in the top 5, but his lower pitchers-only rank of 7.9 brings him down a bit
12. Robin Ventura #21 in 2021. (1990s. System Score: 118.7. In-Decade scaled rank 27.3). Comprehensive review of system to remove only partially applied DRA defensive numbers raises Ventura onto my ballot for the first time. The best third baseman of a very power-heavy period that strongly favored first base over third base: a rapid reversal of the 70s & 80s status quo.
13. Sal Bando. #19 in 2022. (1970s. System Score: 127.5. In-Decade scaled rank 25.7). Removing DRA from my system also raises Bando’s standing. I’m not yet putting him as high yet as the numbers indicate, but he’s definitely ballot-worthy.
14. Bucky Walters #15 in 2022. (1940s. System Score: 116.1. In-Decade scaled rank 26.8). When I revisited the 1940s and 1950s in 2021, I found that pitchers were pretty thinly represented from the 1940s—the war appears to have been especially hard on them. Walters pitched through the war, had an excellent peak, and just enough value outside it (helped, like a surprising number of top pitchers from this period, by his hitting) to move over the historic in-out line. The HoM doesn’t need another player from the 1940s, but it’s not over-represented, either, so Walters is a slightly better choice than any of the half-dozen unelected, pre-integration pitchers sitting just above or on the historic in-out line. His pitchers-only scaled rank for the 1940s is 5.5, the highest on the ballot, but as the pool was thinned out by the war, I am not placing as much weight on that as I might.
15. Babe Adams #25 in 2022 (1910s. System Score: 114.4. In-decade scaled rank 21.3). Revised analysis puts him farther above the in-out line and onto the ballot. I’ve given him a bit of minor-league credit: in a later system, I think he would have been back in the majors faster, given how well he was pitching in the top minors. Just above the in-out line from a weak decade, in which pitchers are already well represented. Pitchers’ only in-decade rank of 7.9 puts him in line with Vic Willis and ahead of all non-balloted pitchers.

Chris Cobb 2023 Ballot Explanation Continued

Top 10 returning players not on ballot

David Ortiz -- #64 in my rankings. (2000s. System score: 104.9. In-decade scaled rank 35.5) Wouldn’t be a bad choice. I think the DH adjustment is fair, and with that adjustment, Ortiz had a very nice career but not much of a peak. He was similar but not as good offensively as players like Willie Stargell and Harmon Killebrew. They are lower-tier HoMers for their decades, and they have more defensive value than Ortiz as well as more offensive value. From my way of looking, Ortiz is a bit short.

Thurman Munson -- #74 in my rankings. (1970s. System score: 113. In-decade scaled rank: 36.7). As with David Ortiz, still wouldn’t be a bad choice. The borderline position players of this decade are hard to evaluate appropriately because the era is so pitcher-heavy. If I were to make independently scaled rankings of position players, Munson would be much nearer but still not on my ballot. His upside ranking for me would be the mid-30s.

Bob Johnson -- #76. (1930s. System score: 119.2. In-decade scaled rank 37.3). I now see Johnson as better than the elected Medwick and Averill, but he’s still below the in-out line. The 1930-45 period was a bonanza for mid-tier sluggers. On the one side, they didn’t have Ruth/Hornsby or Williams/Musial pressing down from above, but power-hitting was still a rare skill among position players, so the mid-tier power hitters were able to grab a lot of value. On the other side, the advent of World War 2 extended their careers by pulling the younger stars who would have replaced them into military service. As a result, they look better when compared on raw scores to position players 10-30 years before them and 10-30 years after. If you attribute all of their value to their ability rather than to their context, then we should have elected and be continuing to elect more position players from this period than almost any other. In this respect, they are like the pitchers who began their careers in the late 1960s and early 1970s. They benefited from the “big-stadium” little deadball era during their primes, rolling up huge numbers of innings pitched, and then when conditions changed in the 1980s and pitching staffs transitioned from four starters to five, their careers were extended as well. In sum, I find that when I ask questions about why there are so many position players from the 1930s whose careers are 5-10 wins better than their counterparts in the decade-by-decade rankings from 1900-1960, I find a lot of explanations that are contextual. Those findings lead me to view Johnson as being comparable to players like Kiki Cuyler and Bobby Veach, even though his career value and system score is higher than theirs.

Tommy John – Not in my top 76, which is as far as my all-period sequential rankings extend. (System score: 111.2. In-decade scaled rank: 39.2.) The Bob Johnson comment explains a number of the reasons that I don’t rank Tommy John highly and that I think he would be a problematic choice for the Hall of Merit. He had an immensely long career during one of the best periods of all time for pitchers to have long careers. On a seasonal basis, he did not tend to throw a lot of innings, nor did he tend to be remarkably effective in the innings he threw. He was a somewhat above average pitcher for a very long time, and he had many contemporaries who were better. He is the #14 pitcher in my 1970s rankings. We have elected 11 pitchers who earned more WAR in this decade than in any other, which is a very high number, probably a couple more than we should have. I don’t see the 12th best pitcher, much less the 14th best pitcher, of any decade as being well qualified for the Hall of Merit. Other ways of looking at John might place him more highly among his contemporaries, and a system that finds him to be better than, say, at least two of out of the four of Don Sutton, Luis Tiant, Rick Reuschel, and Jim Palmer, would be providing compelling evidence that John has been unfairly overlooked. But I suspect that in some cases John’s accomplishments are not being contextualized, and that leads to him being overrated.

Ben Taylor – Not in my top 76. (System score: 92.4. In-decade scaled rank: 52.9). I have been a longtime supporter of Taylor, but when I finally did full MLEs for him this year using the Seamheads data, he dropped significantly in my rankings. On his own, he looks more like Ed Konetchy than George Sisler, and both Hurley McNair and Bill Pettus look substantially better among his Negro-League contemporaries. If the conversion factors in my MLEs turn out to be too low (which may well be the case), Taylor could move back up, but even then he would still be behind McNair and Pettus, and my conversion factors would have to be very low indeed for the correction to put Taylor back in the top 20 for his decade. So he is out of my close-to-the-ballot rankings for now. I see him as a top 50 first-baseman all time, but closer to 50 than to 30.

Man, lot of login/posting problems. Guess I should sign in more often...

Prelim ballot, always pushing the time limit. Appreciate y'all keeping the trains running.

2023 ballot – A new no doubter, then various old-timers that have been knocking on the HOM door (please someone answer!).

Methodology in brief: The system used for my ranking entails a little bit of everything including WS, WAR, OPS+/ERA+. Ratings include positional adjustments, additions to one’s playing record for minor league service, war, and NeL credit and for our real oldtimers some contemporary opinion thrown in. The HOM discussion threads are gold, I appreciate all who have devoted much time to their chosen research.

The results of this work tend to favor prime/peak players over career types. Last year’s placement is in parenthesis.

1. Carlos Beltran (new). Top shelf CF with 9 All-Star Games and 3 Gold Gloves. 4th all-time in SB%, successful over 86% of the time while swiping 312 bases. Post season slashline: .307/.412/.609 in 65 games, widely known for called 3rd strike taken vs Adam Wainwright to end the Mets/Cardinals 2006 NLCS.

2. Phil Rizzuto (2). Holy Yankee shortstops Batman!! I’ve done my minor league & WWII absence calibration so Scooter scoots to ballot position. Glove first but the offense during prime years was nothing to sneeze at either.

3. Lance Berkman (3). He’s the type of extended prime player my system tends to favor. Bob Johnson of his era, likely to have the same fate where 30 years from now his name will be unfamiliar to many casual baseball fans.

4. Tommy Leach (4). Combination hot corner/centerfielder could field a little, hit a little. Second all-time in inside-the-park home runs to Wahoo Sam Crawford. Someone else stated he was uniquely valuable in his particular era and I agree he meant more in the particular era he performed in – you don’t get to play 900+ games at 3b & CF without having a good defensive skill set. Useless trivia: Still holds World Series record with 4 triples in a single series.

5. Tommy John (12). I’m going to downgrade my 19th century pitchers in relation to this innings eater. I had him close to Pettitte, another Yankees hurler close to HOM level. Nothing quite as soothing as watching hitters pound his sinker into the dirt over, and over, and over again.

6. Tony Mullane (6). Old time pitcher who threw plenty well, a good hitter to boot. Had some playing time issues since he missed seasons due to being blacklisted. He’s amongst the best of his era when accounting for the time outside of baseball due to conflicts with different leagues. Goes on the all-Nickname team: come on, “The Apollo of the Box” has that certain something.

7. Mickey Welch (7). 300 game winner in the house. Was it due to luck, run support, bad opponents? Still a feat to accomplish, sometimes I need to remind myself that and not totally overlook Smilin’ Mickey. Seemed to pitch well against the other front line starters of his day.

8. Bob Johnson (8). Always a bit underrated in Win Shares due to quality of teams he played on. His career has war years that need discount. But also a couple years at the beginning of his career were in the PCL where he was more than major league quality. The tail of his career is nonexistent since the 1946 avalanche of returning War players pushed him back to the minors.

9. Bobby Bonds (9). Even with the constant trades, drinking problem and whatnot his combination of speed/power made him a very valuable player. He wasn’t the next Mays, or as good as his son, but we’re talking about a RF who could steal bases and field his position. All five tools on display.

10. Ben Taylor (11). Outstanding NeL analysis and metric updates from this group have convinced me that the first sacker is right in the hunt.

11. Buddy Bell (13). The gap between top-tier 3Bmen is not large for the position when he played in the 70s and 80s. Body type didn’t really look the part of a great glovemen but few would deny he was outstanding.

12. Vic Willis (14). Another pitcher this one from the turn of the century..the 20th, not the 21st. His era probably is owed another slot. HOM worthy? Eh, no strong feelings.

13. Tim Hudson (16). Neck-and-neck with Pettitte and we just let in the NYY lefty. Extended prime, thought of decently by media/fans at the time with 4 All-Star appearances and 4 Top 5 Cy Young finishes.

14. David Wright (17). Sometimes life isn’t fair. Great player but the body balked too early and that was that. Seeing him play in Double-A on his way to the Mets there was little doubt Mr. Wright was going to be a star.

15. Sal Bando (18). Gap between him and Buddy Bell is razor thin despite different career shape. Much of the backlog can say that though. Bando had impressive 5 year peak 1969-1973,

Next up, but off ballot:

Next group –Jack Clark, Luke Easter, John Olerud, Mark Buerhle, Urban Shocker, David Ortiz

Disclosures:

Top 10

Thurman Munson: Not there putting him at top of catcher ranking of eligible. My order is Posada/Tenace/Munson/Schang, Munson in 50-60 range.

David Ortiz: Just off, my DH discount drops him a little but he’s lingering.

Other newbies: John Lackey
Jered Weaver
Jacoby Ellsbury
Jhonny Peralta
Matt Cain
Jayson Werth
J.J. Hardy
Mike Napoli
R.A. Dickey

whoa, yeah, that's a no for me dog...

Chris Cobb - As usual, I learned a lot from reading your comments. I have a couple of questions for you.

In your Sal Bando comment, you mentioned that you've removed DRA from your system. Ironically, as I've thought about making upgrades to my own system, I had tentatively decided to start including DRA defensive data for the period before BIS data became available. So, your comment makes me curious about why you decided to remove it, and whether you see problems with the metric. (I thought Humphreys book was fairly persuasive, but I may not be familiar with some of the pros and cons)

In your Ben Taylor comment, you mentioned that you "finally did full MLEs for him this year using the Seamheads data." In my prelim ballot (#350 above), based on my own perusal of the Seamheads data and simple back-of-the-envelope conversions, I reached a similar conclusion about Taylor. I wonder if you'd be willing to post your latest MLEs on the Taylor thread?

Thanks

Brent,

I removed DRA for two reasons, neither of which offers a particularly compelling reason for anyone else to do the same. First and foremost, I lost access to DRA fielding values when the Baseball Gauge went under, and I had incorporated DRA for only a fairly small subset of players. Since my system involves ranking players against all of their contemporaries and then integrating eras, the system really needs consistent data sources for its integrity, and I didn't feel that I was achieving that with DRA. Second, I have been concerned about the wider range of fielding values provided by DRA compared to the range of TZ as interpreted in BWAR. I think it needs some kind of systematic narrowing that I don't have the capacity to apply. Given these concerns, I decided I would have more confidence in my own results if I removed DRA from the system. Overall, my reasons have to do mainly with my inability to incorporate DRA effectively and not with any substantive objection to its representations of fielding value.

Regarding the Ben Taylor MLEs, I don't think they are at a point where they are sufficiently authoritative to be posted. They are MAYBE slightly beyond back-of-the-envelope but not by very much. As I think about it, I think that I could post about what the calculations suggest about the relative values of Taylor, Pettus, and McNair, but I wouldn't want to argue for the conversion factors I've been using, at least not yet. I could also explain the approach I'm taking, which is quite simple and could be duplicated by anyone with a bit of time on their hands.

Thanks, Chris

I'm fine with 20 player ballot, although I'm not sure there's enough consensus at this point that it would add a lot (since points added for those spots are not significant). And is it fair to someone like Rizzuto who used to have a lot of support and maybe would have been elected earlier if we did 20 player ballots to change now?

Definitely up for a re-ranking of positions!

I am fine with going to a 20-slot ballot starting in 2024, although I don't see such a change as urgently needed.

I'd be interested in doing an updated positional ranking project. It would actually dovetail nicely with my own efforts to construct my own "top 50" lists at each position. If I hadn't been working on that, I don't know that I could keep up with a positional ranking project, but I am far enough along in my own work already that I think I could contribute.

Are we doing an extension this year? Or are we transferring prelim ballots over this evening? I am up to date on ballot count, but won't be available to count more until closer to 6 pm tonight

I can do an extension but I don't want to go past Friday. I always struggle with giving people an extension when they have a year to do the ballot.

We have 20 votes, plus whatever hasn't made it over from here (I think only one). So I'm not sure we need an extension. Either way, I should have the count to you by 630 if we don't extend.

Which prelims are there that haven't been transferred over to the ballot thread? Because even though I never confirmed with you DL (sorry), since I've counted ballots the last few years, I've done so again this year. I just didn't want to miss any. Including my ballot that I just posted, I have 21 in my spreadsheet.

And since I started voting soon after the previous positional rankings, I would be very eager to participate in new ones.

Sorry for the delay, I'm finalizing my ballot and should be posting it within the next hour. Took advantage of the holiday break I had and looked through and incorporated a lot of information from the different rating systems as well as my own evaluations. Basically, I looked at rWAR, fWAR, gWAR, Tom's pwins and ewins, win shares (specifically win shares above bench), heck I even looked at what Prospectus has been doing with their WARP (still heavy with the timelining I see). So, lots of information that was sorted through, and the end result is that there are some changes to my ballot. There's one player that surprised me in terms of actually moving up once I looked at all of those different metrics. It also further muddled the ranking of certain sets of players, most notably that all of the third basemen are now even more grouped together than they were before. I'm finishing up the final order I'm deciding on so I'm hoping to get the ballot in within an hour.

Which prelims are there that haven't been transferred over to the ballot thread?

Just Ardo - comment #72

Should we transfer and count Ardo's ballot? If not, I'm done and will send to you

I sent Dan my tally without Ardo’s ballot. I saved a separate copy that included his, but if I counted right, it wouldn’t have changed the elected, anyway.

I sent mine w/o ardo's too

Seeing how we are likely to elect a C, 3B and 2B in the next election I'd like to leave those re-rankings for 2024. I'll start with outfielders, then we can do SS, 1B and the pitchers.

...and Theorioleway mentioned it in the LF discussion thread, Doc C is working on unveiling MLEs in Q1 of 2023, would be great to work on positions with fewer Negro League electees.

Looks like:
Right Field - none
Left Field - one mostly with Irvin + partial Minoso
Second Base - one + partial Robinson
Third Base - one if we count John Beckwith with SS
First Base - two
Catcher - four + partial Campanella
Shortstop - six
Center Field - ten! if we include Rogan and Dihigo here instead of SP

My request is to pick right field next.

I would suggest we all have Mauer, Beltre, and Utley ready to slot into our rankings as well.

Been trying to post in the 2024 thread, but the site won’t let me. Could someone pls repost this there??? Thanks!

ChrisCobb: Two answers for you.
1) The QoP figures will run tomorrow on my site
2) Re the feedback loop you mentioned between MLE and QoP ratings: I was aware of that when I calculated the STARS scores. I did this work while the MLEs were still based on a QoP rating that made the NeL about as strong as AAA. I took all the players I had an MLE for that came out above replacement and noted how many seasons the MLE covered and what ages. Virtually everyone I have talked to thinks that AAA is a minimum baseline of quality for the NeLs, and some feel it is a hard undersell. So I felt pretty confident that I was getting a reasonable answer and not an inflationary spiral.

James Newberg, the MLEs do NOT include any segregation adjustment. You would need to do that yourself, and that adjustment does, in fact, need to be applied to NeL and MLB players both.