As far as # of Negro League teams I'd use roughly this number

Before 1890 - Union Assoc and Player's League will give you plenty of team seasons.
1891 - 2 teams
1900 AL expansion will bump up team seasons from 12 to 16 so don't bother having more than 2 NGL equivalents
1910 - Chicago American Giants formed - 3 teams
1920 - NNL formed - 4 teams
1923 - ECL formed - 5 teams
1928 - ECL disbands but ANL plays - 5 teams
1929 - ANL defunct - 4 teams
1932 - NNL collapses - 3 teams (represents NSL and Monarchs)
1933 - new NNL formed, East-West ASG held - 4 teams
1936 - NAL is created - 6 teams
1942 - Negro League World Series revived - 8 teams
1945 - Robinson signs - 7 teams
1946 - Campanella, Newcombe sign - 6 teams
1948 - Last season for NGL World Series, Paige signed - 5 teams
1949 - NNL folded, teams absorbed into NAL - 4 teams
1951 - NAL shrinks - 2 teams
1952 - 150 players integrated, Aaron plays for the Clowns - 1 team
1953 - Zero teams

If you're using a 3 year moving average then you get:

0 teams in 1889, .66 in 1890, 1.33 in 1891, 2 in 1892

You can figure out the rest. It's a total of 232 team seasons or about 7-1/2 years of today's baseball over the course of 60 years.

Thanks DL. Those numbers are counting the Negro Leagues as half leagues, right?

I'm going to backtrack the 29 players that are in the Hall of Fame and see what that would roughly come to also.

Either way we end up doing it, this will give us a good idea of how to proportion them also . . .

Before 1890 I am NOT planning on counting the Union Association at all. Still working through how to count the AA, Player's League, etc..

I much prefer a 'smoothed' approach at that time to a herky jerky approach.

The quality of play did not decline in the 1890s, despite the contraction to 12 teams, for example. So we'll need to work through some of those things. We are 'generally' basing this on team seasons, but not exactly. We are moderating it with common sense as well.

Those numbers are counting the Negro Leagues as half leagues, right?

Correct.

The absolute earliest I would add a team is 1885 when the Cuban Giants started barnstorming. Maybe only 1/2 of a team to start. That would add 3-6 team seasons for a total of 235-238. If there is a number between 232 and 238 that makes the overall math easy going forward I would use that number.

We inducted exactly one Negro Leaguer whose career ended before 1921 (Frank Grant, whose career ended in 1903), and we didn't induct him until 1926. So I agree that counting the early Negro Leagues pretty lightly is appropriate.

Whoops, I forgot Home Run Johnson, whose (effective) career ended in 1913 in my post above.

OK, I've got some random numbers . . .

First, we elect players a lot faster than the HoF does. Our median (non-Negro League) HoMer waited 5 years for election to the Hall of Merit. The median Hall of Fame election time for those same players is 13 years. Average wait was 11.85 years for HoM, 22.5 for HoF. I didn't go back and remove the backlog Hall of Famers like Cy Young who waited 25 years for HoF election, so that skews the results quite a bit I'd imagine.

Evidence that we were too conservative early . . . there is a -.35 correlation with retirement year and gap between election, meaning if you retired younger earlier, you would have a longer wait.

I know what you are thinking, remove the early backlog guys and that would clear some of it up. Nope. Starting with players elected in 1906, the correlation shoots to -.44.

And this isn't because we cleared the backlog later and were electing all no brainers from say 1990-2001.

The gaps of the most recent (as of 2001 electees): 8, 5, 5, 14, 5, 6, 5, 6, 5, 90, 5, 5 5, 30, 5, 43, 64, 18 (then you have a bunch of 5s and 6s).

Median election time for the 28 Negro Leaguers we elected through 2001: 6.5, average time 14.7 years. So we were a little more cautious with them, but I think overall this shows we did a pretty good job with them.

if you retired younger, you would have a longer wait.

If you retire younger you appeal less to career value voters.

More random stuff - average career length of non-Negro League HoMers - 16.5 years (not counting token seasons on either end). For Negro Leaguers 18.9 years.

The trend is going up - career end year correlates with career length at .33. Doesn't take a rocket scientist to figure out that careers are getting longer as medical advances and money are both a lot better than they used to be.

if you retired younger, you would have a longer wait.

If you retire younger you appeal less to career value voters.

Not retired 'younger'. Retired 'earlier'. Meaning your career ended in 1923 vs. 1973. Sorry for the confusion.

OK, some ideas for how to count teams . . .

The base number (not dealing with Negro League teams just yet).

1871-1875: 6
1876-1881: 8
1882-1900: 14
1901-1960: 16
1961: 18
1962: 20
1969: 24
1977: 26
1992: 28
1998: 30

Each of those gaps will be phased in over a 5 year period.

So while we jump to 8 for 1876, it will actually be:

1876: 6.4
1877: 6.8
1878: 7.2
1879: 7.6
1880: 8.0

I just didn't want to clutter the list above. This is sort of like DL's suggestion of a moving average, but I don't see why we'd start to phase in an expansion before it's even happened.

Then starting with 1882 we get:

1882: 9.2
1883: 10.4
1884: 11.6
1885: 12.8
1886: 14.0

The Players league is just a blip on the screen. I don't even think it should be accounted for.

By 1892-99 the NL was sitting on 12 teams, but that league was also very strong due to the contraction of the AA - it's not like all of those players disappeared. The fact that we are back to 16 teams within 9 years is solid evidence that there was plenty of talent around.

I don't really see a reason to drop back the number of teams, it's not as if play declined or there were fewer great players. I'm much more comfortable with averaging 1882-1900 than saying the game was better from 1882-1891 than 1892-1900, when we know it wasn't.

1901 the AL comes along and we are pretty much back where we were in 1889 with 16 teams. This gets phased in:

1901: 14.4
1902: 14.8
1903: 15.2
1904: 15.6
1905: 16.0

Now remember, we aren't accounting for the Negro Leagues here yet. I think it's pretty conservative to say the ML game didn't improve from 1905 to 1960, but I don't really see any alternative. So we stick at 16 teams until the Angels and Twins come along.

1961: 16.4
1962: 17.2
1963: 18.0
1964: 18.8
1965: 19.6
1966: 20.0

Followed by the next big expansion:

1969: 20.8
1970: 21.6
1971: 22.4
1972: 23.2
1973: 24.0

Then the Jays and Mariners:

1977: 24.4
1978: 24.8
1979: 25.2
1980: 25.6
1981: 26.0

The Marlins and Rockies:

1993: 26.4
1994: 26.8
1995: 27.2
1996: 27.6
1997: 28.0

And finally the Diamondbacks and Devil Rays.

1998: 28.4
1999: 28.8
2000: 29.2
2001: 29.6
2002: 30.0

Does this make sense? I assume there may be some disagreement about my suggestions for how to treat the early years, but the idea is to smooth things out, be logical, and be fair to all eras.

So we stick at 16 teams until the Angels and Twins come along.

The Twins relocated but they're an original franchise. I think you mean the Senators.

I don't see why we'd start to phase in an expansion before it's even happened.

Because generally the quality of play improvement will precede the expansion. If you're totaling up seasons then spreading it over 5 years doesn't matter. It's still the same total of team seasons in the end.

Through 2001, the Hall of Fame elected 189 players from this pool (non-Negro Leagues).

Add to this 29 Negro Leaguers elected, and our target should be 218 total HoMers through 2001.

Now we need to discuss the 'lag'. Should we using a moving average, based on career lengths of HoMers elected? Or should we just average all of history?

If we average all HoMers elected 1898-2001, we get 16.46 years for career length. The harmonic mean reduces the impact of the high end outliers and is 15.4 years. The harmonic mean of the gap between career end and election is 6.6 years.

This would give us a gap of 7.7+6.6 = 14.3 years for the lag time for when team seasons should start counting towards elections.

But I do think we should work in a moving average of career length for the first half of the lag calc. It's pretty obvious from the start

1992-2001: 18.2
1991-2000: 17.9
1990-1999: 17.7
1989-1998: 17.6
1988-1997: 17.1

That's as far as I've gone so far, but it's definitely looking like a trend we need to account for, or we'll end up with too many electees.

The Twins relocated but they're an original franchise. I think you mean the Senators.

I realize this, but it sounded weird saying the Senators came along in 1960. I know the Twins were the old franchise, but they were the new city. I guess I could have said Angels and "New" Senators or something . . .

Because generally the quality of play improvement will precede the expansion. If you're totaling up seasons then spreading it over 5 years doesn't matter. It's still the same total of team seasons in the end.

You are right that it doesn't matter, but I don't really agree with this.

I think if you build it they will come. Meaning you expand and then players appear. More demand generates more good players.

Most expansions are generated by economics, not because there are all of these great players that we don't have any room for. Right?

I just realized I should be adding one when referring to average career length.

1945-1955 is an 11-year career, not a 10-year career.

But I think for our averages, using 10 is correct. The mid-season in a 1945-1955 career is 1950, and 1955-1945, divided by 2 puts us at the correct number also, 1950.

Sorry if this is confusing. I'm still wrapping my head around it.

On the other hand, I should be adding one to the gap between career end and election year. Hank Aaron's last season was 1976, and he was elected in 1982. So for purposes of what election year that team seasons should be effective we should be adding 7.77 and not 6.6.

Make sense?

I'd rather see the NL drop back down to 12 in 1891 but let the NGL teams keep it at 14. If we put 1 NGL team starting in 1888 (Frank Grant banned from International League) until 1891 when 2 appear then we end up with an even number of team seasons through year 2001: 2546. If you do a 3 year trailing average you get 2522 team seasons through 2001. 5 year trailing average is 2498. By doing a 5 year trailing average you're underestimating the total number of team seasons by about 2%.

DL - I'm still trying to keep the Negro League teams out for now. I want to make sure we are on a reasonable track otherwise first.

We have 28 NGL electees through 2001, the HoF had 29 (including the 2006 election) . . . so I think just leaving that part of the equation on the back-burner for now will make it easier.

I'm starting to think a 15 year moving average for the career length lag will be better . . . that's around the average career length, and there's less noise in the early going . . .

Adding in the Negro Leagues gives us this:

1871-1875 6
1876-1881 8
1882-1887 14
1888-1890 15 (could smooth this back to 14 from 1882-1900)
1891-1900 14
1901-1909 18
1910-1919 19
1920-1922 20
1923-1928 21
1929-1931 20
1932-1932 19 (could smooth this out to 20 from 1910-1935)
1933-1935 20
1936-1941 22
1942-1944 24
1945-1945 23
1946-1947 22
1948-1948 21
1949-1949 20
1950-1950 19
1951-1951 18
1952-1952 17
1953-1960 16 (coincides with post-WWII talent gap)
1961-1968 20
1969-1976 24
1977-1991 26
1992-1997 28
1998-2012 30

This list is also good justification for using 10 as the ballot length for MMP voting from 1871-1968. I could justify going to an 11 person ballot from 1936-1941. Even though there were more NGL teams during WWII the talent drain due to the war means keeping ballot length at 10 makes sense from 1942-1948.

OK, 15 year moving average of career length of electees. After thinking about it more, an avearge as opposed to a harmonic mean is appropriate here. The harmonic mean is still appropriate for the end of career to election portion of the lag.

15 years ending in:

2001: 18.1
2000: 17.9
1999: 17.4
1998: 17.2
1997: 16.6
1996: 16.8
1995: 17.1
1994: 17.0
1993: 16.8
1992: 16.5
1991: 16.5
1990: 16.5
1989: 16.2
1988: 15.9
1987: 16.0
1986: 15.9
1985: 16.0
1984: 16.5
1983: 16.6
1982: 17.1
1981: 16.6
1980: 16.9
1979: 17.0
1978: 16.6
1977: 16.4
1976: 16.4
1975: 16.4
1974: 16.3
1973: 16.3
1972: 16.0
1971: 16.2
1970: 16.3
1969: 16.0
1968: 15.7
1967: 15.6
1966: 16.0
1965: 15.7
1964: 15.7
1963: 16.0
1962: 16.1
1961: 16.4
1960: 16.6
1959: 17.0
1958: 16.9
1957: 17.1
1956: 16.7
1955: 16.9
1954: 16.7
1953: 16.8
1952: 16.5
1951: 16.3
1950: 16.2
1949: 16.4
1948: 16.9
1947: 17.0
1946: 16.8
1945: 17.0
1944: 16.5
1943: 16.5
1942: 16.2
1941: 16.4
1940: 15.9
1939: 16.0
1938: 15.9
1937: 16.5
1936: 16.6
1935: 16.3
1934: 15.6
1933: 15.2
1932: 14.7
1931: 15.0
1930: 14.7
1929: 15.3
1928: 15.1
1927: 15.6
1926: 16.1
1925: 16.1
1924: 16.1
1923: 16.0
1922: 15.9
1921: 15.6
1920: 15.3
1919: 15.3
1918: 14.8
1917: 15.6
1916: 15.3
1915: 15.1
1914: 14.6
1913: 15.1
1912: 15.3

Plotting that on a graph, it's pretty volatile with a slight trend upwards that becomes a sharp trend upwards around 1999.

Actually, zooming in on the important area of the graph, it looks like there's a huge (from around 15.25 to around 16.5) jump that takes place in the early 1930s. That holds (nearly every year between 15.5 and 17.1) until around 1997. Then there is an early 30s like jump up another year to around 17.5 from 1998-2001. Whether this will hold is yet to be seen, but it is interesting.

What's the line of best fit on 121? Can you plot the original data on a control chart?

A trend line smooths this from 15.8 in 1912 to 16.6 in 2001. I don't think I like that very much.

We could just go with a 16.5 year average career length for the lag calculation throughout the project. but that doesn't seem right to me either.

That chart really is kind of all over the place. But the line going off the page from 1998-2001 is definitely something to keep an eye on I guess . . .

I wish I could show you guys this somehow . . . is there any way to do that?

Let me plot each data point, not just the 15-year moving average, maybe something will show up there . . . I'll plot it against retirement (as opposed to election) year also. Something should jump out at us somewhere.

Interesting . . . the scatter plot based on retirement year shows it all over the place. Not a perfectly even splattering of data points, but a very slight trend upwards.

The trend line shows 1876 retirement being a 15.9 year career length, 1995 retirement at 16.9. It's a trend, but it's a very slow one.

I think the trend line you just described is the correct one to use.

Kind of think sticking with baselines of 16.5 and 7.75 makes sense right?

So our team season lag will be 8.25 + 7.75 = 16 years. Basically teams will start counting towards election 16 years after they play.

8.25 of that lag is half of the average Hall of Merit career.

7.75 is the harmonic mean of election year - final season played for the average Hall of Merit career.

Does that seem reasonable? Am I missing anything? We would of course reserve the right to modify this should something crazy happen with either of these trends somewhere in the future.

DL, do you see merit (no pun intended) in running with that trendline as opposed to a simple 16.5?

I'm open to using either methodology.

If we do use the trendline, I'll go back and rerun it based on 'middle' season of the career, not retirement year. Should have done that in the first place, just wasn't thinking. It will be very similar I'm sure, but then it will be applied to the correct team season year.

OK, running the regression on center season of career, we get 1866 at 16.0, 1986 at 17.0.

So we can use 1/2 of that as the career length lag.

Also, we elected 9 players who played before 1871 and one who played as far back as 1856.

We probably should account for this very slightly in our team-season calculation. I think this can be back-tracked relatively easy. It doesn't sound like much, but there were basically 44 "Hall of Merit" player seasons that took place pre-NA.

There were 3038 non-NGL Hall of Merit player seasons total. So the pre-NA time should come to about 1.45% of the total non-NGL seasons we come up with.

1856-1861: 1 player
1862-1865: 2
1866: 3
1867: 4
1868: 6
1869: 8
1870: 9

So I think it's reasonable to use those in our final numbers. Stretching the trendline back to 1856 shows a 15.9 year lag on that season.

Based on all the data presented I would use a "team season lag" of 16 years for all years.

There were 3038 non-NGL Hall of Merit player seasons total.

Divide that by 16 year career length and you get 190 players. Is that correct?

Actually 184 (still not counting the 28 NGLs). So that works about to 16.5. Which makes sense since the trend goes from 16-17.

BTW according to my data, we were 6 behind the HoF in 2001, after accounting for NGLs.

HoF - 189 + 29 NGLs
HoM - 184 + 28 NGLs

I'll post my list to make sure we aren't missing anyone or double counting. I went based on the baseball-reference.com list which designates if they are in as player or something else.

Apologies for how long these are, I realize they are tough to read.

Both:


HoM only

Negro Leaguers

Both:



Only Hall of Merit



Only Hall of Fame

Just remember those lists are only through 2001, except for the additional NGLs elected in 2006.

Thanks for doing all this. One question on the last post(s). This is as of after the 2001 HOF and HOM elections, right? And for Negro Leaguers it includes the 2006 HOF selections?

Correct. We are also including the 2006 HoF NGL elections.

The reason for this is that:

1) They nearly match our selections

2) The Hall of Fame should have wrapped that up (save the handful of true borderline backloggers) long before 2001, like we did.

3) Not fair to the non-NGL backloggers to lose spots because the HoF took forever to get to the NGLers. If we would have known those were coming in 2001, that would have impacted our 'matching' philosophy at the time; which was to match the HoF up until they decided to change the rules so the Veteran's Committee could barely elect anyone.

Inducting 1 HoM player per 10 team seasons fits with currently electing 3/year with 30 teams. It also adds up to 290 slots available through 2012 and we've inducted 246. Throw in the 16 year lag and that drops the number by 16*30*/10=48 for a number of 242 slots. That is also equal to 189 MLB slots through 2001 assuming a 16 year lag. That's exactly what the HoF inducted. It gives 24 slots for Negro League players using my assumptions on player seasons. If my assumptions are low by 40 NgL team seasons in 60 years then we're right on target for NgL players and MLB players as of 2001. It gives us 246 slots by 2012 and 246 players inducted.

There's our formula: (Player seasons from 1871 to Current Year-16) / 10.

If we keep inducting 3/year there is no need to correct anything. It matches our current total and the Hall of Fame slots as of 2001.

Bresnahan also became a HOMer in 2004, according to his page....

Thought about this a little more on the ride home from work. If you divide by 16 instead of 10 you get a nice "small hall" line.

No reason to change the number of players inducted unless MLB expands. No reason to track to the Hall of Fame anymore. No reason to have a special election.

DL, that's close, but I'm not trying to 'match' it to fit what we've done.

I want to come up with the correct theoretical numbers and see where that leaves us.

Later tonight or tomorrow I'll run through all of the numbers with the correct lags, slowly increasing over time. It's kind of a pain to easily do in excel.

I don't want to back of the envelope it. I want to get it right. I am not convinced 3 per year is some magic number. And if we skew low, compared to where we should be, we go against our charter of being fair to all eras and it will start to add up over time.

Bresnahan also became a HOMer in 2004, according to his page....

Right Howie, those lists only apply through 2001.

I thought the idea was to match the HoF in 2001 and use that as the correct number to that point in time. Player seasons / 10 matches that number if you throw in the 16 year lag, which happens to match the career length.

It is - to match the HoF in 2001 + NGLs.

Here's a question though - I think we should be using 6 years as the lag time related to eligibility not 7.75 (the harmonic mean of the average election wait for HoMers).

As we said before, the 6 comes from a player who retires in 1976 being eligible for the 1982 election. After stewing on this some more, I think this should be used for the lag, not the average election time. The lag is to decide when those seasons become 'eligible'.

Does that make sense?

So we'd have a moving lag range of half of 15.9 (for 1856) through 17.2 (for 2011) plus 6 as the lag factor.

Make sense?

So the lag for 1856 would be 7.9 + 6 = 13.9 seasons. By 2011 the lag is 8.6 + 6 = 14.6 seasons.

Why half of a career? The number could just as easily be justified as the 10 season minimum for eligibility plus the 6 year self-imposed waiting period. That's when it makes sense to count a team season - when the players who were rookies that year are first eligible for consideration. That would mean a player who started in 1871 (and you can't really start competing for a pennant any earlier) would be eligible in 1888 (not 1887) because that is when enough team seasons would be compiled for a voting slot to open up.

Might be a lot of work for marginal benefit. Could just go with a 14.25 year lag throughout?

Why half of a career?

Half a career because among the Hall of Fame seasons in any given season, some are from players with 16 years left and others are from players with 1 year left.

That's when it makes sense to count a team season - when the players who were rookies that year are first eligible for consideration.

I disagree there - most of those rookies won't be eligible for consideration then - and those that are, the ones with 10 year careers, aren't usually very good candidates.

That's why we went through all of the trouble to figure out what the average Hall of Merit/Fame career length is.

Give me a chance to work up the numbers, I think they'll make sense by then.

If anyone is looking for something "fun" to do and would like to verify that my lists above are correct - mainly that I've slotted people into the 'player/contributor' bins appropriately, that would be greatly appreciated.

Could just go with a 14.25 year lag throughout?

Or round it to 14 years

Clark Griffith was inducted into the HOF as an Executive.

Thanks Rob.

B-R lists him as a player. His Hall of Fame plaque speaks mostly of his playing, including his over 200 career victories. His Hall of Fame web page lists his batting stats, LOL.

Why do you say he was inducted as an executive - was he on a special ballot for executives or something like that?

Check this 1946 Hall of Fame election link . . . Clark Griffith looks like he's lumped in with all of the other players.

Looking at that list and who is on it, all of guys nominated made it in eventually.

Also, this detailed description of the 1946 elections seems to show Griffith as being lumped in with the rest of the players.

Here's another from the Hall of Fame website, titled the 1946 Induction Ceremony and it lists Griffith in with "dominating pitchers in Jack Chesbro, Clark Griffith, Joe McGinnity, Eddie Plank, Rube Waddell, and Ed Walsh."

The previous link talked about how the 1946 committee specifically went back and looked at pitchers from the pre-1910 era.

Citing all of that, do we think it's fair to include Griffith in with the players unless some other 'evidence' emerges?

Maybe it is best to consider Griffith a combo selection (player, manager, owner). Anyway, the Baseball Hall of Fame page on Griffith says he was selected as an "Executive".

http://baseballhall.org/hall-famers/members/searchable-data?lname=griffith&cat=All&negro=All&pos=All&team;=&state=All&class;[value][year]=&class;[value][month]=0&class;[value][day]=0&class;[value][hour]=0&class;[value][minute]=0&class;[value][second]=0

No big deal either way I guess.

Yeah maybe Rob. Since we did end up electing McGraw too, maybe we just combine them as one player and one executive for "HoF" purposes.

OK here we go on the Negro Leaguers and their team season accounting . . .

Our total number of eligible, lagged, team seasons through 2001 to 1913.3. Don't ask me how I got there just yet, I'll explain a little bit later. Just go with that assumption for now.

There were 189 non-NGL Hall of Fame players elected through that time.

This amounts to 10.1234 (not making that up, cool decimal) team seasons per HoMer.

There are 29 Negro Leaguers in the Hall of Fame for our purposes.

This would mean that we would need to add 293.6 team seasons to the total number of team seasons.

This works out to more than DL's estimate of 232, but it's not outrageously off and seems entirely reasonable.

How to distribute those teams by season. Well really, there are three possible methods.

First - proportion based on NGL HoMer season through 2001 (long chart alert):



That totals to 557 player seasons. Take 293.6/557 and you get .53 teams seasons per HoMer season . . . or:



Method two would be to regress that slightly so we don't get the big peak on 1926. Basically I'd run a regression from 1887-1925 and from 1926-1956 and use that to smooth it out a bit.

This differs pretty significantly from DL's suggested distribution (I've prorated his proportions to get to our 293.6 total), or method 3:



His is centered much more on the early 40s than the mid 20s.

I suppose we could go with a 4th method, averaging the two approaches.



This shows a peak in the mid-20s, followed by another peak in the late 30s and early 40s.

I'm not a NGL expert by any stretch, so I think I favor the last approach. Definitely open to being convinced otherwise on this. But we did elect a bunch of HoMers centered on the mid-1920s, and I assume we did that for a reason.

We are getting close. Once we've got this figured out, I'll present the entire thing. It's pretty interesting. I still haven't gone past 2001 to see if we are ahead or behind through our 2012 election. Want to nail the methodology down first.

I did just think of one additional way to do it.

Use my methodology, but chop off the first and last 2 seasons of each player's career. This would smooth out the overlap of career ends and beginnings, might make it take a slightly different shape, not as focused on the mid-1920s.

The peak you see in the 1920s makes sense. I conservatively only added 1 team for the Eastern Colored League. There were 6-7 teams in that league from 1923-1928. It would add 2 more team seasons per year (total of 7) if we treat it as a full league. That's 10 additional team seasons right there. The odd thing is seeing the peak happen for player seasons in the gap between leagues. It looks like the talent level was still there for 6 teams even though the economic conditions weren't. That gets me up around 250-260 team seasons.

Your number is close enough to 10 team seasons / HoM slot that I wouldn't budge. The actions of the Hall of Fame making the VC much more difficult starting in 2001 indicates that they were trying to correct for an overshoot. Even if you keep that number it means we elect one additional guy every 10 years.

Edit - one _fewer_ guy every 10 years. Not one additional. Messes up my VC theory.

How is this . . . took two years off of the front and three off back of each HoM career.

Sorry DL, hadn't seen your comment when I posted mine.

So of these, which do you think is the most appropriate way to distribute the NGL teams . . . I could tweak and run some other ways if you have ideas also. I'm not too hung up on the 293.6 or 250 or 232 teams right now - I'm more concerned about getting the relative distribution correct.

Add the ECL teams, then average/smooth it out.

This is an update of a post I made two years ago.

RE: Player Count, Hall of Merit vs. Hall of Fame – We’re still catching up.

With the 2012 election, the Hall of Merit has elected 246 players. At the same time, Larkin's election will bring the HOF official player count to 236 (207 MLB, 29 NeL). Despite these counts, the HoM still has not caught up to the HOF’s number; we are still more exclusive.

Once you account for the HoMers that the HOF does not count as “players”, our number drops below theirs. Start with the banned players: HoMers Pete Rose and Joe Jackson would certainly be in the HOF if there were no “character clause”. A couple others are classified by the HOF as “managers”: John McGraw and Rube Foster. Add Joe Torre to that list, whom the HOF has rejected for his playing but is a shoo-in for election as a manager. Then we have three HoMers whom the HOF lists in the “executives/pioneers” category: George Wright, Al Spalding and Clark Griffith. Along these same lines, we see four players in the HoM who did not play ten years in MLB from 1876-on, implying the HOF would classify them as “pioneers” if they were elected: Dickey Pearce, Lip Pike, Cal McVey and Ross Barnes. (None of those four appeared on the lists of 200 candidates for the four VC elections from 2003-09.)

That’s 12 HoMers who are not in the HOF consideration set of players. Thus, an apples-to-apples comparison shows the HOF player total of 236 as still being more than our adjusted total here of 234.

Dan, that's a good point - we definitely need to account for all of the oddities you mention . . .

I'll give my thoughts one-by-one:

Pete Rose: Definitely should count on the Hall of Fame list
Joe Jackson: ditto

John McGraw: I don't think we should count him. His skill set is such that I don't think he'd be in the Hall of Fame as a player. I mean if Denny Lyons isn't in, I doubt McGraw gets in as player.

EDIT: Just looked up McGraw's placement on the 1936 Veteran's Committee ballot - he actually showed up pretty impressively, between Ed Delahanty and Hoss Radbourn in 7th place. Herman Long, who finished 9th, is the only player among the top 15 that isn't in the Hall of Fame. I'd say let's count him as Hall of Fame.

Rube Foster: I don't know enough about him. Reading his plaque it talks about him as a player. Maybe count McGraw/Foster as one Hall of Famer?

Joe Torre: I think the Hall of Fame pretty clearly rejected him as a player. As of 2001, he was off the BBWAA ballot (peaking at 22%) and wasn't yet eligible for the Veterans Committee - he's out.

George Wright: His plaque makes it pretty clear he was elected as a player, no matter how they've classified him now.

Al Spalding: Seems like he would not have been elected for just his playing, based on reading his plaque.

Clark Griffith: We discussed him earlier, but he almost definitely would have been elected as a player, based on his performance in player based votes. And the evidence doesn't exactly show that he wasn't elected as a player anyway.

So I would say add 5 to any Hall of Fame list that doesn't include the players mentioned above:

Rose
Jackson
EDIT: McGraw
Foster (?)
Wright
Griffith

As far as Pearce, Pike, McVey and Barnes go - they are simply 'misses' by the Hall of Fame, in terms of comparing to the Hall of Merit. They are part of what the Hall of Merit was designed to 'fix'.

Is this reasonable?

EDIT: Still open to thoughts on Foster - do we think he would have been elected without his career as an executive/manager?

Including the information above, I'm seeing 194 traditional Hall of Famers through 2001, plus 29 or 30 NGL's, depending on how we count Rube Foster.

Am I missing anyone further?

Here's how I've counted anyone 'controversial':

Candy Cummings - Pioneer

Rube Foster - ??? leaning counting him as a player.

Joe Jackson - Player

Tommy McCarthy - Player

John McGraw - Player

Pete Rose - Player

Al Spalding - Executive - note, he only received 4 of 78 votes in that 1937 election. Granted, several eventual Hall of Famers received fewer, including Monte Ward, Billy Hamilton and Dan Brouthers. But he was also behind guys like Bill Lange and Jerry Denny. Any thoughts?

George Wright - Player

Evidence that we were too conservative early . . . there is a -.35 correlation with retirement year and gap between election, meaning if you retired younger earlier, you would have a longer wait.

I know what you are thinking, remove the early backlog guys and that would clear some of it up. Nope. Starting with players elected in 1906, the correlation shoots to -.44.

Uh... there's an enormous sampling bias in this. It's possible for John McGraw to have gotten into the HOM over 100 years after his retirement as a player, because he retired over 100 years ago; it is as yet impossible for this to have happened to Kevin Appier (or whoever your choice is for moderate backlogger who retired recently, and could theoretically be inducted into the HOM at some point within the next 100 years).

Good point Eric - how do you suppose I should try to test for that.

Either way, even if that's not the way to go about showing it, there was definitely a bias towards backloading the elections. We actively discussed it.

Good point Eric - how do you suppose I should try to test for that.

Either way, even if that's not the way to go about showing it, there was definitely a bias towards backloading the elections. We actively discussed it.

I'm not sure how you would adjust for it - maybe look at median wait time, putting the players in buckets by decade of retirement? It might not completely resolve the issue, but it would help.

Regardless, backloading the elections to some extent seems like a sensible thing to have done.

open to thoughts on Foster - do we think he would have been elected without his career as an executive/manager?

He may not have been elected in 1981 but he would have made it by the time the last group made it.

Thanks DL . . . Eric, I'll take a look if I get a chance. I really kind of stumbled onto that and saw it as evidence, wasn't really actively looking for it or trying to prove anything.

So I think the only question mark left at this point is whether or not to count Spalding as a player.

I'm kind of leaning more towards counting him as a player. His 252-65 record would have eventually got him in as a player. Sure he was behind a few obscure players in the that 1936 vote, but he was also ahead of some no-brainers and most of those guys are in. His plaque does mention his playing prominently, though not exclusively.

Any thoughts one way or the other?

I am very comfortable with Spalding considered a HOF player.

OK, so going forward, I'm going to consider 195 Major League Hall of Famers and 30 Negro League Hall of Famers through 2001. If anyone has an issue with this, or a question as to who is in each bucket, please speak up ASAP. Thanks!

Combining Hall of Merit NGL player seasons with DL's estimated distribution, then regressing to smooth things out . . . we get gradual improvement from 1889-1922, a spike from 1923-28 with a dropoff in 1929, then gradual improvement through a second peak in 1943 with a final decline therafter.

Does that actually describe Negro League quality accurately? If not, how would you describe it, in terms of improvement, peaks, declines, etc.? I've got the underlying data set, it's just a question of how to regress showing a curve the moves realistically.

That sounds like a reasonable approach. One worry is that in using the HOM election data, the quality curve is biased against players the HOM has been biased against--sunnyday's been a pretty forceful advocate for integration-era black players (the "Lost Generation") and there does seem to be a dip in the 1940s and 1950s.

I looked at the election results to see if the current top candidates are biased towards one era or another (suggesting that era may be underrepresented in the HOM): Redding (1910s) is 6th, Newcombe (1940s/1950s) is 11th, Taylor (1910s/1920s) is 14th, Hilton Smith (1940s) is 17th, Bus Clarkson (1940s/1950s) is 19th, Bill Monroe (1900s/1910s) is 29th, Elston Howard (1940s/1950s) is 47th, Carlos Moran (1900s/1910s) is 51st, Luke Easter (1940s/1950s) is 76th, Ray Dandridge (1930s/1940s) is 84th. Leroy Matlock (1930s) has also received votes recently. These guys heavily cluster around the 1910s and the 1940s, suggesting that both eras are "underrated" slightly by just using HOM results. Maybe we should count these guys as partial electees in some way (perhaps incorporating votes received)?

We could do something like that Alex.

Do you have career dates for those guys. I'm looking for realistic MLE type career dates. If you have them all and could easily post them that would save a lot of time for me having to look them up and make judgements.

One thing to note though . . . that won't change the number of 'slots' we allocate - that's fixed at the 30 the Hall of Fame has. All this will do is determine what years we allocate those slots to.

Maybe we count the players in the Hall of Merit at 100% and the players who received votes at 80%?

Actually, ideally, since we aren't really rushed - does anyone have our Hall of Merit consensus MLE's accumulated somewhere all in one place, for all of the HoMers, Hall of Famers and those still receiving votes? Would actually be great to use WAR or Win Shares as opposed to 'seasons' when accounting for playing time. I don't think I have the time to go searching for all of them though.

Spreadsheet - if it exists would be fine, my email address through the site works fine. I wouldn't expect anyone to post them all here - although I might at some point make a thread for them.

MLE career lengths (excluding major league careers):

Redding 1911-1927
Newcombe 1946-1948
Taylor 1910-1929
Smith 1935-1948
Clarkson 1938-1956
Monroe 1899-1914
Howard 1948-1950
Moran 1899-1913
Easter: unclear where his career starts, I've seen 1941-1949, 1939-1949 and 1937-1949
Dandridge 1934-1952
Matlock 1929-1942

I have a spreadsheet of current candidates only (so that covers those receiving votes as well as HOF/not HOM). I'll clean it up a little, pick out the relevant players, then I can send it over some time later tonight.

Good stuff, thanks Alex!

OK I wasn't sure how to send it as an attachment so I decided to post it to Google Docs. I wouldn't stand behind the WAR numbers as gospel: I made a number of assumptions, such as arbitrarily setting fielding numbers and using the shortcut wOBA = (1.8*OBP + SLG)/3 approximation. If anyone's interested I can also post the full ugly spreadsheet with all the calculations and approximations.

Excellent discussion.

Joe, the totals look reasonable.

Agreed on classifications of players and that candy cummings is a pioneer, while the others can be viewed as players.

It is a shame that the pre national league players have been ignored, but yes, the hall of merit has eased those slighted.

If anyone, including alex, can post negro league mle war totals to the hall of merit yahoo group, that would be awesome.

I apolgize for the inactivity, missplaced, hopefully didn't lose my revised rankings incorporating dra defensive values.

Bleed, you can use the Google Docs link posted in 181 for some WAR estimates. There are other players that I would consider good candidates who I should probably add to the Google Doc at some point, pending some discussion of those players.

I just want to speak for the HOM voters who find this stuff over their heads, but who greatly respect the attention to detail. It matters.

Joe has had that attribute from Day One....

Thanks Howie, that's nice to hear.

I haven't forgotten about this guys . . . this was a travel week and as an added bonus, I'm snowed in in Seattle, I won't get home until midnight Saturday heading into Sunday . . . and I haven't seen (or from my wife's perspective, "helped with") the baby since Sunday night . . . so figure it will be some time Monday/Tuesday before I get a chance to dive back in.

Joe - did you reach any conclusions from all this data? If so, can you summarize them?