Negro League Stats

[uruguru]

Originally Posted by OutS|der

The way I see it the stats are just a product of the era, the only issue is when they use small sample size like the new ERA leader of 28 games otherwise you can argue about the talent levels till the cows come home.

All past totals have to be taken with a grain of salt due to how baseball has evolved over the years. Due to various reasons stats were inflated due to subpar talent, using speed, steroids, juiced balls, or when pitchers threw low 90's and could doctor the ball.

It's a different problem.

The issues you talk about can be addressed well by using era-adjusted stats. Using OPS+ and ERA+, for example, instead of OPS and ERA.

Low talent level is a more insidious problem to solve because it allows good players to look like great players, and great players to look like all-timers. You can't just look at player-to-player variation because the left side of the curve (the bad players) are constantly replaced throughout the season.

We know the effects of thin talent levels exist but we have no way of understanding how to account for it. Josh Gibson hits 20 homers in 39 games in a league full of minor-league equivalent players and everyone wants to pretend he was Babe Ruth.

[Pelican]

So, what do you say to the (many) home runs that Gibson hit against the best MLB pitchers in head-to-head games? Or, for that matter, the ones Ruth hit against lousy pitchers for dreadful AL teams?

Not sure what your factual basis is for saying that the Negro Leagues were "full of minor league equivalent players" (if that is your view). You might be right. I am willing to admit that I don't have the answer. Unlike the stars, the worst MLB players never got to play the worst Negro League guys. So, it's all just speculation.

My guess is that bad Negro League players had a lot in common with bad MLB players. Rosters were filled out with mediocre players. Mediocre as compared to Ruth and to Gibson. Better than 99% of baseball players.

For sure, the white leagues had many more minor league players to choose from for replacements. By the 1930's most MLB teams, following the Cardinals' example, had coherent minor league organizations. But in most cases, there was a reason - related to ability - they were in the minors.

Agree with you on era-adjusted stats. The game does a good job of that, and users can fine-tune the results. I'm playing in 1911, and the difference between expected stats in that season and in 2024 is astounding, hitters and pitchers (although with a .241 MLB batting average this year, we may be moving in the Deadball direction, at least as to that outdated metric, if not as to slugging).

[joefromchicago]

Originally Posted by uruguru

To give you an idea of what I mean, I selected a year at random: 1930. In that year, there seemed to be 15 NLB teams and 16 MLB teams. In other words, 48% of what we now consider to be "major league quality" teams were composed of black players. But in 1930, they represented only 9.7% of the total US population. So clearly the talent level of the NLB was thinner, and that would have inflated the stats of the top players.

Not exactly an apples-to-apples comparison. In 1930, I would guess that most MLB teams carried around 22 players on their rosters. They were allowed to carry more (I think the limit was 25 at the time, but I don't have that information in front of me), but for cost-cutting reasons most teams didn't carry the full limit and some teams might have even carried fewer than their normal complement on road trips.

In contrast, my guess (and, with the Negro Leagues, it's almost always a guess) is that most teams carried around 15 players on a normal basis. And MLB only recognizes the NNL as a major league in 1930, so, although Seamheads lists 15 NeL teams in 1930, only 8 of them were "major." So if we calculate that MLB teams carried 22 players and NeL teams carried 15, we get 352 MLB players and 120 NeL players, which means that NeL players represented around 25% of all major-league players in 1930. And if you did an analysis based on player-games-played, the proportion would be even more heavily weighted in favor of the white MLBs, as the NNL teams only played around 80 league games that year.

[OutS|der]

Originally Posted by uruguru

It's a different problem.

The issues you talk about can be addressed well by using era-adjusted stats. Using OPS+ and ERA+, for example, instead of OPS and ERA.

Low talent level is a more insidious problem to solve because it allows good players to look like great players, and great players to look like all-timers. You can't just look at player-to-player variation because the left side of the curve (the bad players) are constantly replaced throughout the season.

We know the effects of thin talent levels exist but we have no way of understanding how to account for it. Josh Gibson hits 20 homers in 39 games in a league full of minor-league equivalent players and everyone wants to pretend he was Babe Ruth.

You're right we have no way to account for thin talent levels Babe Ruth played against.

[uruguru]

Originally Posted by OutS|der

You're right we have no way to account for thin talent levels Babe Ruth played against.

I realize that this was just a snarky attempt to own someone on the internet, but you are correct that the talent levels that Ruth faced were lower than what Williams and Dimaggio faced in the 40s, and Mantle and Mays faced in the 50s.

But if we are comparing the relative talent levels between MLB and the NNL at any particular year in history, the talent levels in the NNL are clearly going to be a lot lower simply because they are pulling players from a much smaller pool of talent (i.e. 10% of the country vs 90%)

[Syd Thrift]

Originally Posted by uruguru

I realize that this was just a snarky attempt to own someone on the internet, but you are correct that the talent levels that Ruth faced were lower than what Williams and Dimaggio faced in the 40s, and Mantle and Mays faced in the 50s.

But if we are comparing the relative talent levels between MLB and the NNL at any particular year in history, the talent levels in the NNL are clearly going to be a lot lower simply because they are pulling players from a much smaller pool of talent (i.e. 10% of the country vs 90%)

Sure, we know this is the case to some degree but it's pretty impossible to quantify the actual degree. Once baseball did get fully integrated, until the big influx of Latin American and later Asian players, black ballplayers were overrepresented in MLB (note: not of course because of silliness like "blacks are naturally better at baseball" but because athletics has historically been a way for black people to escape poverty that hasn't quite been the same for white people). To take this to an extreme, would an all-black basketball league with 8ish teams (which is about where the NBA sat at once it stabilized) in the mid-50s really be more "diluted" than the all-white NBA?

We also know that whenever NeL teams and MLB teams played head to head, the NeL was very competitive; IIRC the NeL teams in the 30s in particularl won around 2/3rds of the time. Obviously, we can't begin to say that this is a sign that the NNL or NAL were *superior* in talent to MLB - I'm sure the all-black teams were trying harder - but it sure is a point of evidence for them being *similar*, like, the NFL champions used to play a team made up of the best players in college and, even though the NFL teams classically and kind of famously didn't take those games seriously in the least they still defeated the College All-Stars so often and so badly most of the time than eventually the game was dropped.

I agree that we'll probably never really be able to get at the actual exact different levels between the two leagues. Even when we do have instances of players having played in both leagues, which of course only comes at the very, very end of the existence of the NeL, the sample sizes are miniscule. I think people, frankly, need to leave it at "yeah, we don't know for sure" as opposed to pretending to care about "sanctity of the statistics" (which you haven't done here and I'm not saying you are).

[alkeiper]

I feel confident enough in the statistics where they exist, granted you need some adjustment to the level. The biggest problem is the sample size. If you're only drawing a sample from one year you're bound to get some inaccurate assessments. Is the simulation looking at multiple seasons? Can you train it to do that?

What drives me mad are players like Pancho Coimbre and Tetelo Vargas who were clearly great and for whom data for the majority of their careers simply does not exist.

[Garlon]

We have posted an NeL Mod for the community.

https://forums.ootpdevelopments.com/...d.php?t=357028