Well, the Big Train will be a cinch to make the top of my ballot. Beyond that, I haven't a clue yet.

I'll be away from this for the extent of the discussion. Should be back by the 15th though.

1) Johnson
2) Young

1) Johnson
2) Young
3) Alexander
4) Williams
Then I have to think about Mathewson versus Nichols.

One thing to note: IP per season trended downward throughout the period. This shows up in my top equivalent FWP seasons, which skew sharply toward the earlier part of this time.

Just to make sure, we're ranking these guys as players only, right? That is, we're not supposed to figure out how to add in credit for Rube Foster's managing or Griffith's ownership.

- Brock

As players only. And that's how it should have been in the last group - I'm pretty sure no one was talking about Spalding's post-playing career.

1) Johnson
2) Young
3) Alexander
4) Williams
Then I have to think about Mathewson versus Nichols.

Sounds right to me.

Hmm, Smokey Joe is in that conversation.

I've got career RA+ Pythpat records for the MLB pitchers in this group. (I don't have that for Foster, Mendez, or Williams.) I'll show it to you below sorted by career equivalent FWP. I'll also show you a "big years score", which is simply the seasonal excess equivalent FWP above 15. There are three caveats:

1. I haven't corrected for the pitcher's own offensive quality. While none of these guys were Spalding/Caruthers level hitters, there were in fact large differences between the best and worst of them. And given how many AB they'd get in a season, that does matter.

2. Defensive backing matters. I made exactly two attempts to correct for that, with Nichols and Brown. (Also with the non-HOM Willis.) But I didn't do that systematically across all of these pitchers.

3. The "big years score" needs to be timelined. Heavily. Otherwise it's just too much of a built-in advantage for the earlier guys.

Anyway, here's the list:

Young 519-298 [255 big years score]
Johnson 427-230 [208]
Alexander 369-193 [198]
Nichols 352-210 [147] (defense adjusted)
Mathewson 332-199 [156]
Plank 303-197 [64]
--- there's a bit of a gap here ---
Walsh 210-119 [107]
Rusie 248-171 [81]
McGinnity 227-155 [77]
Coveleski 209-134 [61]
Rixie 275-224 [19]
Faber 225-199 [41]
Brown 211-143 [52] (defense adjusted)
(Willis 248-196 [44] (defense adjusted, not HoM))
Waddell 200-129 [59]
(Adams 201-132 [40] (not HoM, should be defense adjusted and hasn't been))
Griffith 216-160 [43]
(Cicotte 209-149 [48] (not HoM))
-----
(Chesbro 182-140 [50] - you know we'll hear from Happy Jack!)

Since the issue of Nichols versus Mathewson could be close: what do people do with Nichols's late-career minor league efforts? As you recall, he voluntarily left the major leagues to take up an ownership position with a minor league team, for whom he was personally the star gate attraction. It wasn't the highest level competition available to him, but it might have been his best economic opportunity. He was good enough when he briefly returned to the majors to suggest that he had been a major league quality pitcher during that time.

I'm inclined not to count it - it wasn't the highest level competition, and there wasn't anything preventing him from continuing in the majors. But what do others think?

Lists year electee got in, and rank of all HOM pitchers in that election, those who beat out an HOM pitcher in parentheses, and the top non-electee P in parentheses

Amos Rusie - 1904, 2 Rusie, 3 Radbourn, 5 Galvin, 7 Spalding, 14 Caruthers (18 Welch)
Kid Nichols - 1911, 1 Nichols (19 McCormack), 20 Caruthers (22 Welch)
Cy Young - 1917, 1 Young, 8 McGinnity, 9 Caruthers, 12 Waddell, 21 Griffith (22 Joss)
Ed Walsh - 1920, 1 Walsh, 5 McGinnity, 9 Caruthers, 11 Waddell, 18 Griffith (22 Willis)
Christy Mathewson - 1922, 2 Mathewson, 3 Brown, 5 McGinnity, 10 Caruthers, 11 Waddell, 20 Griffith (24 Welch)
Eddie Plank - 1924, 2 Plank, 3 Brown, 5 McGinnity, 10 Caruthers, 14 Waddell, 20 Griffith, 23 RFoster (25 Welch)
Mordecai Brown - 1925, 2 Brown , 3 McGinnity , 9 Caruthers 9, 15 Waddell, 19 Griffith (22 Welch), 23 RFoster (28 Joss)
Joe McGinnity - 1928, 2 McGinnity, 6 Caruthers, 12 Griffith, 13 Waddell, 15 RFoster (16 Welch)
Walter Johnson - 1933, 1 Johnson, 7 Griffith, 9 Waddell (10 Welch), 16 Mendez (24 Joss)
Pete Alexander - 1936, 1 Alexander, 2 Williams, 6 Covaleski, 11 Waddell, 13 Griffith (15 Welch), 22 Mendez (29 Joss, 29 Mays)
Smokey Joe Williams - 1936, 1 Alexander, 2 Williams, 6 Covaleski, 11 Waddell, 13 Griffith (15 Welch), 22 Mendez (29 Joss, 29 Mays)
Stan Coveleski - 1938, 2 Covaleski, 7 Griffith, 9 Waddell (10 Welch, 15 Redding), 19 Mendez (27 Mays)
Red Faber -1939, 1 Faber, 6 Rixey, 7 Griffith, 10 Waddell (14 Welch, 18 Redding), 20 Mendez (29 Willis)
Rube Foster - 1945 - 2 Foster, 4 Rixey, 5 Griffith, 8 Ferrell, 14 Waddell (16 Welch, Redding 20, Grimes 22), Mendez 23 (Mays 35)
Eppa Rixey - 1968, 1 Rixey, 4 Griffith (12 Redding), 18 Mendez (21 Walters), 22 Waddell (23 Welch)
Clark Griffith - 1971, 1 Spahn, 2 Griffith (8 Redding), 12 Mendez, 20 Waddell, 21 Pierce (22 Walters)
Jose Mendez - 1985, 1 Mendez, 4 Pierce, 5 Waddell (13 Redding)
Rube Waddell - 1986, 2 Waddell, 4 Pierce (9 Redding)

Hey, OCF -

Every once in a while I get a reminder of just how short a time it's been since I started doing the HoM. What's an "equivalent FWP"? I'm guessing that WP is winning percentage, but have no idea what F or equivalent are. Sorry to show my ignorance, but your work is usually useful, so I'd like to know what the headings mean.

Thanks in advance, Brock

BTW, I use Nichols' minor league stuff as a tiebreaker, if he's close to someone else. Part of playing at the highest level is the assumption that this is the level that pays you the best. I mean, if some independent minor league team offered Manny Ramirez $50 million a year to play, he'd probably be a minor leaguer, assuming that he thought that the minor league owner could actually afford the $50 mil. So would people who are not nearly as mercenary as Manny, and Manny is not the most mercenary player in the game. At enough money, even people like Albert Pujols would move. Essentially, Kid Nichols went to a "higher" league, in these terms of money. I don't count the play as major league, though, unless I need a tiebreaker. I also do this to Albert Spalding. His arm might have returned after a year or two. But sporting goods just paid more. There are a couple of other guys who left the bigs early for money or love (Home Run Baker). I give them all tiebreakers, but nothing more.

"FWP" is a "Fibonacci Win Points". It's an old nonlinear Bill James gimmick for thinking about and comparing won - loss records. The formula is

W*(W/(W+L))+W-L.

That is, it's wins over .500 plus wins times winning percentage.

If the winning percentage is about .618, the FWP is the same as the wins. For instance, a record of 233-144 would have FWP 233. (Those are consecutive Fibonacci numbers.) If the winning percentage is .500, the FWP are half the wins. 200-200 would give 100 FWP. If the winning percentage is about .414 (that's sqrt(2)-1), the FWP are zero. So a 10-14 season is a wash; that makes that something like the replacement level for this.

The only difference is that I'm not using it on actual records but on RA+ equivalent records.

It's a weird little ad hoc gimmick with no theoretical support - but I happen to subjectively agree with where it leads me most of the time.

I didn't put the equivalent FWP in post #9 - just the equivalent W-L and the "big years points", but I ranked the list by that. Basically, I'm willing to let you draw your own conclusions from the equivalent record. Had I put the FWP in, the list would have looked like this:

Young 551
Johnson 473
Alexander 397
Nichols (def adj) 363
Mathewson 340
Plank 289
Walsh 225 - hence my "something of a gap" statement. Of course, there's also something a gap between Mathewson and Plank.

The bottom of the list would be

Brown (def adj) 193 (But I also haven't given him anything for relief leverage and pennant-significance leverage.)
Willis (def adj) 192
Waddell 191
Adams (unadjusted) 189
Griffith 181 (Note that the campaign for Griffith argued that he was better than his RA shows.)
Cicotte 181
...
Chesbro 144

And for a couple of cases where the offensive adjustment which hasn't been made would really matter:
Joe Wood: 99-60 for 101 equivalent FWP
Babe Ruth: 80-55 for 73 equivalent FWP

OCF -

Thanks! I remember Fibonacci Win Points, but I was thinking that this was something very new, so I didn't channel the memory. And I agree with you; they do tend to lead to good evaluations, although there is no substantive reason why they should, unless you can demonstrate that baseball is an essentially organic activity and therefore subject to nature's obsession with Fibonacci numbers. And thanks for the list; I promise to use it when I make my rankings up.

- Brock

And I agree with you; they do tend to lead to good evaluations, although there is no substantive reason why they should,

Well, the zero point is .414, which is probably around what a lot of people would use as a replacement level.

Incidentally, you can also write the FWP formula as a quadratic equation in winning percentage, multiplied by decisions. That lets you solve for all kinds of fun stuff... on the off chance that anyone is interested, the equation is:

D*(p^2 + 2p - 1) = FWP, where D is decisions and p is winning percentage.

Of course, there are 101 reasons to override this, as there are many other pieces of information we know about these pitchers. The exact order of my list in #9 will not be the order of my ballot, starting with Johnson in the #1 spot ahead of Young, and with many other shuffles as I go down the list. I wouldn't expect anyone else's ballot to be in exactly that order, either.

One comment on Eric's note: for me, D is not an integer and is not the actual number of decisions; it is IP/9. (IP/8.75 would probably match slightly better, but as long as I do the same for everyone, it doesn't matter.)

By HOM electoral strength:

Unanimous or very nearly so:
Nichols, Young, Mathewson, Johnson, Williams, Alexander

Neither backlog nor nearly unanimous:
Walsh and Plank

Backlog (chronological):
Rusie, Brown, McGinnity, Foster, Coveleski,
Faber, Rixey, Griffith, Mendez, Waddell

(I know, technically Faber was not backlog, but he was very close.)

(I know, technically Faber was not backlog, but he was very close.)

I'd say approximately the same thing about Coveleski.

Among these pitchers, the best timing (as in easiest to get elected from there) belonged to Rusie, who became eligible at a very young age and hence became eligible significantly before any of the the other pitchers in this entire group - he didn't have to be compared to Walsh, McGinnity, and Waddell, and he certainly escaped comparison to Young and Nichols. (Rusie did beat out most of the previous group, however, being elected ahead of Radbourn, Spalding, Galvin, and Caruthers).

The worst timing belonged to Coveleski, who became eligible in the monster 1934 class. The only people who became eligible after 1934 who were elected before Coveleski were Alexander and Harry Heilmann. Just to review, here's the election sequence:

1933: Walter Johnson, Zack Wheat (so Wheat "snuck in ahead of the crowd")
1934: Ty Cobb, Tris Speaker
1935: Eddie Collins, Pop Lloyd
1936: Pete Alexander, Joe Williams
1937: Cristobal Torriente, Harry Heilmann
1938: Stan Coveleski, Heine Groh
1939: Max Carey, Red Faber (by then the class of '34 had moved through the pipeline)
1940: Joe Rogan, Lip Pike
1941: Babe Ruth, Rogers Hornsby

OK, so I had to go back and look: Joe Rogan belongs in the next group of candidates, right?

My point is really that although Coveleski and Faber had some of the characteristics of backlog candidates, they were both elected just about as fast as they could have been.

I'd say approximately the same thing about Coveleski.

I forgot about that. By the def'ns that Paul Wendt and I had worked out, he and Groh were technically "midlog", candidates that never mingled with the "backlog" (by finishing behind one). They "feel" like backloggers because they never finished far ahead of them either. I remember Faber and Evans (Dewey) as mid-log because they had the exquisite timing that allowed them to be 1st year electees despite having the backlog breathing down their necks. (Faber was only 9 pts ahead of Carey, who had finished behind Beckley a few years before).

Brown was also elected just about as fast as possible also. In his first election he finished 3rd behind Lajoie and Mathewson in 1922. Wagner was elected the following year (an elect-1). Crawford and Plank in 1924. Brown finally getting elected in 1925, with HR Johnson sliding in ahead of him.

So we can split the "backloggers" into:

Fast:
Rusie, Brown, Coveleski, Faber

Slow:
McGinnity, Foster, Rixey, Griffith, Mendez, Waddell

Interesing to see if there will be much intermixing between the electoral groups.

OCF #16
One comment on Eric's note: for me, D is not an integer and is not the actual number of decisions; it is IP/9. (IP/8.75 would probably match slightly better, but as long as I do the same for everyone, it doesn't matter.)

It doesn't matter that you use 9.0 rather than 8.75 but it does matter that you use official innings rather than official wins and losses to stipulate the number of decisions in the "equivalent record" --that is, use innings to credit the workload. That you contemplate giving Brown extra credit for high leverage as a relief pitcher suggests you believe that there is relatively low leverage in the work of starting pitchers with relatively low complete games and innings per start --roughly, there is low leverage in the work of modern starting pitchers.

OCF #13
"FWP" is a "Fibonacci Win Points". It's an old nonlinear Bill James gimmick for thinking about and comparing won - loss records. The formula is

W*(W/(W+L))+W-L.

It's linear in workload, W and L or their sum D.

It's nonlinear in winning percentage. A team does equally well by splitting its workload between a .500 pitcher and a .700 pitcher or between two .600 pitchers (who all complete every game). But the .500 and .700 pitcher score more points during the time period, commonly a career, if somehow the four pitchers post careers of equal length.

D is not an integer and is not the actual number of decisions; it is IP/9.

Example:
For Cy Young, 7354.7 official innings, stipulate 817.2 "decisions".
Using RA+ calculate some RA+-equivalent winning and losing rates.
Calculate "wins" and "losses" that sum to 817.2 and convert them to integers somehow. Some methods yield 817 decisions, others may be off by one or more.

OCF, What is your source for RA+? Is it based on a career ballpark-teammate factor such as 102 (integer near 100)?

A team does equally well by splitting its workload between a .500 pitcher and a .700 pitcher or between two .600 pitchers (who all complete every game). But the .500 and .700 pitcher score more points during the time period, commonly a career, if somehow the four pitchers post careers of equal length.

If you phrase it that way, then I'll claim this as a feature, not a bug. After all, the .500 pitcher is a much, much more common commodity than that second .600 pitcher.

My source for park factors has been a Stats Handbook. If I used the bb-ref pitching park factors, the numbers would shift a little. The park factors are all single-year, and the conversion of RA+ and IP to W-L is done a year at a time; only the W and L are added up over a career. So I don't use any career-averaged stats (like RA or park factor) directly.

I don't actually convert anything to integers in any intermediate calculations. I round to integers when I report them here, but the fractional parts are still there, underneath.

OCF (post #4) says, "One thing to note: IP per season trended downward throughout the period. This shows up in my top equivalent FWP seasons, which skew sharply toward the earlier part of this time"

I've been debating for a couple of days what to do about this, but I've finally decided to let you guys decide. In the Big Bad Baseball Annual 1995 (the first one actually called BBBA, and the first one produced by Don Malcolm), I did a study of historical pitcher workloads. The center of the study was a curve demonstrating what did happen to pitcher workloads and when, and what sort of historical curve is there. It's certainly not a precise study (it was 1995), but I reread it, and most of it still stands, IMO. Most important here is that it demonstrates that the curve, which looks like a parabola, "turns the corner" right after the turn of the century, just in the middle of the period we're discussing here. That is, you can get a real good idea of what sort of adjustments to make for individual pitcher time periods.

I no longer have an electronic copy of it, because that computer is long gone, but I can scan the pages in from the book and post that up. Unfortunately, the study is 24 pages long. So question #1 is whether anyone actually wants to see 24 pages on pitcher workloads from the 1860s onward. The second one is, if people do want to see it, how do I post up 24 pages of scans without breaking down the storage limit for the thread?

So, opinions? Help with the posting? What do you guys think? Do I post this monstrosity up or not? Do you think it will help? Paul Wendt may get writer's cramp correcting and adding to it, but I do think it will help in differentiating among these pitchers in this time period.

- Brock Hanke

Brock:

If you can scan it into a PDF file and E-mail it to me, I'll post it on a file-sharing site and link to it.

-- MWE

Mike -

Thanks. I've made PDFs before, so I should be able to handle that. As long as the PDF is not too big for email on either of our two ends, it will be done tomorrow or this weekend. I have no current access to a file sharing site - at least that I know of. Therefore, your offer is exactly what I needed.

- Brock

Paul Wendt may get writer's cramp correcting and adding to it,

Perhaps I will pass entirely. I may pass on the new edition of RAA and PRAA.
--

continuing re RA+-based FWP:

OCF
The park factors are all single-year, and the conversion of RA+ and IP to W-L is done a year at a time; only the W and L are added up over a career. So I don't use any career-averaged stats (like RA or park factor) directly.

In that case there it incorporates a nonlinear bonus for high-rate seasons. A pitcher with one .500 followed by one .700 season scores more points than a pitcher with two .600 seasons (and equal innings pitched all four seasons). I have missed that aspect and supposed that the explicit "big years points" complement a rating based on career aggregates alone.

If Bill James applied it to career W-L records directly, then there was be no built-in bonus for high-winning-percentage seasons in his work.

repeating myself
>>
OCF #13
"FWP" is a "Fibonacci Win Points". It's an old nonlinear Bill James gimmick for thinking about and comparing won - loss records.

It's linear in workload, W and L or their sum D.

It's nonlinear in winning percentage.
<<

It's doubly nonlinear in RA+, because nonlinear in the intermediate estimate of winning percentage which is nonlinear in RA+.

It's linear in workload, or IP in this instance. Because it is linear in workload, a pitcher with one 6-4 season followed by one 18-12 season scores the same as a pitcher with two 12-8 seasons.

If the former "6-4 followed by 18-12" pitcher is rated above the latter, that is in the big years points. Commonly we account for big years in ways that do make ratings non-linear with extra credit for high workloads. For example, we count everyone's numbers of 20-wins shares seasons and 30-win shares seasons. Indirectly that credits both high-rate and high-workload seasons --roughly, because those counts are so discontinuous.

Prelim ballot

1) Walter Johnson
2) Cy Young
3) Pete Alexander
4) Smokey Joe Williams - these are also my 1-4 on the all-time list
5) Christy Mathewson
6) Kid Nichols
7) Amos Rusie
8) Eddie Plank (may change to Plank, Rusie)
9) Jose Mendez
10) Ed Walsh
11) Stan Coveleski (was Coveleski, Mendez, Walsh before I started using Dan R's numbers)
12) Rube Waddell
13) Mordecai Brown
14) Eppa Rixey
15) Rube Foster
16) Clark Griffith
17) Red Faber (Faber > Brown before Dan R)
18) Joe McGinnity

DL,

Care to discuss/document the basis for your ranking of the bottom third or so of this group?

I know, in most of these positional elections we've cared mostly about the top, but from the top down to Stan Coveleski on this one there's not a great deal to talk about, aside from Williams and Mendez, I suppose.

12-18, however, has a bunch of pitchers of similar value, but very different career shapes and pitching styles. Ultimately, they are probably too close together for rank ordering them to be all that meaningful, but I'd like at least to see some discussion of Waddell/Brown/Rixey/Foster/Griffith/Faber/McGinnity set!

I use two different inputs - value above average and value above replacement - to answer the question of who I think should be inducted. Here's the main numbers out of my spreadsheet that I'm using. WPA is a small bonus. I'm essentially averaging the BP and Dan R numbers with more weight on Dan R. I am figuring in pitcher hitting but it's mostly irrelevant in this discussion (Brown was OK, Griffith could hit).

------ DAN R --------- BP --ME-
player PWAA WARP2 PRAR PRAA WPA
Waddell 27.0 47.8 822 304 0.0 (credit for Western League included)
REDDING 24.5 48.5 860 244 0.0 (estimated for Cannonball Dick)
Brown 20.8 48.2 781 251 1.0 (estimated Dan R 1913-1916)
Rixey 21.3 55.2 973 194 0.0
Foster 21.0 51.0 970 190 0.0 (estimated to look a lot like Rixey)
Griffith 21.7 47.5 710 235 0.0 (est 1892)
RFaber 18.8 46.6 1009 213 0.4
-gap-
McGinnity 18 40.8 716 236 0.8 (lowest scores across the board, esp Dan R numbers)

Waddell has no trouble staying ahead of this group with by far the best value above average and McGinnity is dragging behind the rear.

The hitting and postseason credit nudge Mordecai Brown ahead of the rest. Rixey's career value compensates for his lower value above average. Griffith needs his hitting to get him solidly above Faber. I have these guys between 60 and 71 all-time so it's a tight grouping. They're also all in Pud Galvin/Don Sutton territory. I wouldn't argue with anyone else's ranking of Brown to Faber, they're essentially at the same point on the bell curve. It's very possible my Dan R numbers are missing relief innings. If any of these guys have significant relief value let me know and I'll adjust.

DL from MN, remember that my numbers were based on BP's DERA, which has now been changed...I imagine the new version is much better if it's based on play by play defensive stats.

> I imagine the new version is much better

Why, do we have PBP defensive stats for this era now? I can imagine it is better for the eras where we actually have the data.

Here's the numbers for the top 3 contenders - new and old BP

Player PRAR PRAA oPRAR oPRAA
WJohnson 1383 821 1989 816
CyYoung 2013 932 1881 800
Alexander 1118 628 1671 673

So, Walter Johnson and Pete Alexander stay close to the same for PRAA (good result - this means the value of an average pitcher was pretty solid) and PRAR drops for both (good result, saner replacement value). However, Cy Young is now gaining 200 PRAR and 132 PRAR which would put him solidly into #1 all-time territory - if you go by WARP1 he leads by 50 WARP now over Johnson.

Doublechecking recent numbers:
Clemens 1275 700
Maddux 995 376 (dragged down by his career after 2002) - Clemens leads by 20 WARP1

The new PRAR numbers for Cy Young are clearly not correct: even if the new WARP1 has adopted a constant pitching replacement level, there's no way the difference in career IP between Johnson and Young justifies that difference in PRAR.

Looking at the cards, it appears to me that the system has failed to apply the new xIP system (whatever it is) to Young. He has 7356 actual innings pitched, and his xIP are listed as 7399.5 (more xIP than IP). Johnson has 5914.7 IP, 5056. xIP.

There's simply no way that Walter Johnson, strikeout king, should be earning fewer xIP / IP than Cy Young.

Someone who knows how might drop Clay Davenport an e-mail about this one.

Unfortunately, I think we will have to be very cautious with the new WARP1 numbers until a) we actually understand what they mean and b) the bugs from the change-over get sorted out.

I'm assuming Dick Redding would be in this group. Since he's a top 10 returnee I'd encourage people to include him (unofficially) in their rankings of this era.

Does Coveleski get minor league credit? He started playing in the minors in 1909 but didn't stick in the majors until 1916.

FYI - I have Urban Shocker ahead of Redding and Brown. Vic Willis would only slot ahead of McGinnity.

Trying to find info on Coveleski I noticed that he does have 65 relief appearances over the course of his career that wouldn't be credited in Dan R's database.

Stolen from SABR bio project:
"In 1912 he moved to Atlantic City in the same league and continued pitching well, finishing the season with 20 wins and a 2.53 ERA. Near the end of the 1912 season Coveleski caught the eye of manager Connie Mack from the nearby major league Philadelphia Athletics. Signed to a contract and given a couple of late-season starts, the 23-year-old Coveleski acquitted himself well, but Mack believed he needed additional seasoning. Accordingly, the manager sent Covey across the country to Spokane in the Northwestern League, apparently believing that he had a gentleman's agreement with Spokane to retain the rights to the promising young pitcher.

Coveleski played two years in Spokane, hurling over 300 innings each year. In his stellar sophomore campaign Coveleski won 20 games and led the league in strikeouts. After the 1914 season, Portland of the Pacific Coast League traded five players to Spokane to acquire Coveleski, now considered the Northwestern League's top pitcher. According to one story Mack had forgotten about his rights to Coveleski and failed to take up his claim for the player. Later, upon inquiring about Covey, Mack was sent a box of big red apples as thanks and told his rights had expired."

"While at Portland Coveleski developed the spitball for which he later became famous.... With his newly developed spitball, Coveleski led the Pacific Coast League in games pitched and finished the 1915 season 17-17 with a respectable 2.67 ERA."

It looks like he was pitching at the level of an average innings-eater from 1913-1915.

Quick, somewhat raw prelim:

1. W Johnson
2. C Young
3. Alexander
4. J Williams
5. Mathewson
6. Nichols
7. Rusie
8. Walsh
9. Plank
10. Mendez
11. Waddell
12. Coveleski
13. Griffith
14. McGinnity
15. M Brown
16. R Foster
17. Faber
18. Rixey (not pHOM, the only one here who's not)

A lot could change, especially after Coveleski. I explain Mendez/Waddell by my peakiness. Ditto not having McGinnity last, though really everyone at 13 on could show up somewhere else on my final ballot. (Well, except Rixey, probably.)

Ed Walsh Best 6 years (Dan R) 45.6
Eddie Plank Best 6 years 32

I guess I can see how you have Walsh ahead of Plank. Unfortunately Big Ed doesn't have much outside of that and Plank has 6 more pretty good seasons (>=3 WARP). Plank has 9 wins on Walsh from a career perspective.

Yeah, I still weight peaks pretty heavily--not as much as I used to, but enough that Walsh's peak advantage keeps him ahead. (Probably all I need to say is that I still vote for Al Rosen every election....)

I need to add WWI credit and relief credit (180 relief appearances) for Red Faber. He'll slide back above Brown with that credit added.

Just FYI, I have just sent Mike Emeigh an email telling him that I finally have a copy of my 1994 pitcher study from the 1995 BBBA. It will take at least a few days for Mike to get it emailed to him and then posted up, but don't blame him. I just had one of the worst colds in my life and was unable to think at all. I wasn't willing to trust myself to make a PDF in this condition. I'm finally over it, but I've lost 2 weeks. I have no idea whether anyone is still interested, but I'm sending the thing to Mike, and wanted to make sure that no one blamed anyone other than me for the lateness.

Sorry, Brock

I'm not a doctor but that sounds like influenza. Glad you're on the mend.

The pitching numbers I have made available (based on a now-outdated version of DERA) DO include relief *appearances* in seasons where the majority of his appearances were starts. They do not include *any* statistics for seasons where a P had the majority of his appearances in relief.

DL -

Thanks for the sympathy; I appreciate it. I think it was a cold for three reasons: First, I've had a flu shot, although there are always some strains that bypass that. Second, I didn't have the symptoms of the flu, but of a really really bad cold. And third, I get these from time to time, due to a childhood spent in the house of a schoolteacher. If you don't want to be subject to really bad two-week colds, don't live in a house where someone brings in everything like clockwork, spring and fall. It was worst on my mother, who was the schoolteacher. She'd bring it home and we'd take care of her for two weeks. Then my dad, my brother and I would all come down with it at the same time and she'd have to take care of all three of us plus go back to teaching every day. - Brock

Phew, this has been a tough one.

First there's the Big 6, who lap the field but not each other by any means.

Then Walsh, McGinnity, TF Brown, Plank and Rusie have interesting paths to success - peak vs prime vs career vs high-volume in-season vs lower-volume in-season. Beauty marks and warts for each.

Finally, Covaleski and Faber and maybe Rube Foster battling.
Or does Foster battle Rixey, Griffith, Mendez and Waddell at the bottom?

Those lost ballots don't help matters any, either; I couldn't even find most of mine via archives.

I'm about half-done; glad for the extra time in this case.

DL in the ballot thread #8
10) Stan Coveleski - arguably deserves minor league credit
11) Ed Walsh - Lots of peak value but a short career.
12) Rube Waddell - Give him credit for the Western League, strong value above average
(Urban Shocker)
(Dick Redding)
13) Red Faber - lots of spitballers in this era, deserves credit for WWI

Coveleski, Shocker, and Faber were among the spitball grandfathers, permitted to use the pitch in the major leagues after 1920. Quinn, Grimes, and Clarence Mitchell were three others and with Faber they were the four longest users, until 1932-34. Of course the regulation limited the number of spitball pitchers and made their work less common, more distinctive than it would have been under outright ban or under the status quo.

Probably this form of regulation was quite favorable to Faber, et al.: not outright ban (certainly worse for them) but low ceiling on the number of competitors permitted to throw "the same" pitch. There must have been differences among the spitball pitchers but I suppose that their pitches moved in similar ways. Given outright ban in the minor leagues, it's plausible that the spitballs were a novel family for batters who were new to the major leagues. I feel it must be true at least that new major league batters needed some time to adjust; at best, from the grandfatherly perspective, all major league batters handled spitballs worse in 1930 than in 1920 because the regulation severely limited their learning. How likely is the latter? I don't have a good guess whether the pitches were similar enough to each other and different enough from other breaking pitches --that is, no good guess whether the spitball pitches were an important class from a batting perspective.

For the baserunners: be alert for passed balls.

Pitchers 1893-1923 Preliminary Ballot

Well, I have a lot of numbers floating around on these guys. I had recently shifted to using a pitcher eval system derived largely from Lee Sinins’ RCAP and WARP1’s PRAA and FRAA, with a peak measure from my own old, home-grown win shares system, all scaled to a replacement level in which an average pitcher is 27 PRAR/200 ip.

Then the WARP numbers changed drastically, including some systemic changes whose meaning I don’t yet understand and of whose validity I am dubious. So for now, I have decided to build my rankings around the average of my old home-grown system, my hybrid using using the old WARP numbers, and my hybrid system using the new WARP numbers.

The main values of adding in my old system, as I see it, are that it is win based and that I know how the adjustments for fielding support were calculated from team defensive efficiency. The WARP numbers are undoubtedly more sophisticated, but they are derived from component stats (so they may diverge significantly from support-neutral wins) and they are black box.

Although the numbers feeding into the system are different in each case, the factors on which I evaluate the pitchers are the same in every case: (1) career wins above replacement, with below replacement seasons dropped from the beginning and end but not the middle of the career (2) career wins above average, for all seasons counted for career wins above replacement, (3) peak rate of wins/ 250 IP X 5, which is further modified by a durability factor, which equals the ratio of their IP during their 5-year consecutive peak to 120% of the IP an average full-time starting pitcher would have thrown during those same five seasons.

1) Johnson – Greatest pitching peak of all time + second greatest pitching career of all time equals #1 on this ballot.
2) Young – Doesn’t match any of the other big four in peak value, but he was so effective for so long that his total merit far exceeds Alexander’s and Mathewson’s, and it nearly makes up even for Johnson’s peak advantage in my system. If Johnson hadn’t also been one of the best hitting pitchers of all time (what an athlete he must have been), Young might have surpassed him.
3) Alexander – Tremendous pitcher. Great stuff, great durability, great smarts.
4) Mathewson – Same package as Alexander, but a bit less durability.
5) Williams – A monster of a pitcher in the teens. I’m not convinced he held his value well through the 1920s and into the early 1930s, though he continued to be very successful in an NeL context, so I rank him behind the big four, but well ahead of Nichols, who was no slouch.
6) Nichols – Steadily excellent. Less durable in context than he appears in all-time terms: over his best seasons, he only put up about the typical innings pitched for an ace, unlike folks like Johnson and Alexander, who were throwing 20% more innings than a typical ace at their peak.
7) Plank – A career rather than a peak candidate. His consistent excellence narrowly edges Walsh’s massive peak in my system. It could easily go the other way.
8) Walsh – The greatest innings eater of his era (on a seasonal basis), edging Johnson, McGinnity, and Alexander, because he put up McGinnity-like IP totals in a context of quickly declining IP after the shift to the 154-game schedule. He was more effective in those innings than McGinnity, as well, so that his peak is the best of the era, after the big four at the top.
9) Coveleski – I probably have him higher than most, but he was really effective for his whole career. His IP seem to lag behind the others’ but he had his two best seasons shortened due to WW1 and his whole career fell in a period of declining IP.
10) Rusie – overrated because of all the strikeouts and the big seasonal IP totals, which, as in the case of Nichols, were not as exceptional as all that for the 1890s. Rusie was notably more durable than a typical ace, however.
11) McGinnity – The Iron Man rates as highly as he does largely because of his fantastic durability, but he did couple it with exceptional effectiveness, too. Both he and teammate Mathewson appear to have been exceptionally good at working with their fielders. Nothing special outside of his big peak, but that’s definitely enough.
12) Foster – I evaluate him as Clark Griffith plus. He had more athletic ability and more peak durability than Griffith, but like Griffith he won as much with smarts as with talent, and his on-the-field value was undoubtedly truncated by the shift of his energy and attention to management and the business of building a sport.
13) Griffith – The one pitcher elected from this era who was never an innings eater. He has one of the lowest “durability peaks” of all pitchers in the Hall of Merit, and his peak effectiveness is also on the low side. He held his effectiveness much longer than most of his contemporaries did, however, which enabled him to put together a HoM career.
14) Waddell – One of the hardest pitchers in this group to rank, though everyone from McGinnity on down is closely bunched. Probably the best stuff at his peak of any pitcher here except perhaps Walter Johnson, but his peak was not all that long, and everything else about his career pulls his value down from where his stuff pushed it to.
15) Rixey – Overall a career candidate, but he did have a good peak. Often ranked in lock-step with Faber. I agree that they are very close, but Rixey’s longer career and longer period of effectiveness as a starter give him a slight edge.
16) Faber – A career candidate, but his case depends upon two (and really one) spectacular season.
17) Mendez – Undoubtedly great during his peak. How much he pitched, and how much value he would have had outside of his peak, are awfully hard to estimate. Definitely enjoyed a surprising renaissance with the Monarchs in the early 1920s, before succumbing to tuberculosis later in the decade.
18) Brown – I long thought his election was a mistake. I have gone back on that view now, but I still find him the weakest of this group. His effectiveness was aided by extraordinary defensive support during his peak; he was not an innings eater; and his career was not especially long. He’s very similar to Eddie Plank in terms of peak and overall effectiveness, but with 1300 fewer IP.

In 1919 and 1920 the seventeen pitchers who would be permitted to continue using the spitball worked more than 3400 and 3600 innings or 201 and 215 innings apiece. Only Ray Fisher left the majors after 1920 but their number of innings decreased to 3000. This table gives the annual numbers of innings through 1935 when Burleigh Grimes threw the last permitted spitballs.

aggregate major league innings, 17 spitball grandfathers
1916-20: 2216 3210 2314 3420 3653
1921-25: 3037 2130 1933 1854 1753
1926-30: 1298 1027 1058 _809 _592
1931-35: _651 _365 _185 __53 ___0

Each of the sixteen major league teams played about 1300 innings annually, the aggregate workload of the spitball pitchers in 1926. Their one thousand innings pitched in 1927 or 1928 was about 5% of major league innings.

Deadball Era Resources provides some more information about the spitball pitch and its pitchers, including links to encyclopedia pages for the 17 grandfathers.

Through 1927 the grandfathers worked more in the American League than in the National, then vice versa. Here is the annual share of all American League innings.

share of AL innings, 17 spitball grandfathers
1916-20: 13.0 17.4 15.2 18.9 19.0
1921-25: 16.6 12.5 10.5 _9.2 10.2
1926-30: _7.7 _4.7 _4.3 _3.6 _2.3
1931-35: _1.6 0096 0079 0017 0000 (1931-35, Red Faber essentially alone)

At that time one pitcher in a big season worked about 3% of all league innings. By the 1930s none of these pitchers worked 2% of all league innings (about 220 ip).

Perhaps it is because the HoF has been more generous with pitchers than us but we haven't had a HoM not HoF guy come up yet. If elected, Dick Redding would be the first from this group.

Dan Greenia led a project at baseball-fever called "the Ultimate Quest for Candidates" covering major leaguers including the Ross Barneses. Officially we finally ranked the top twenty, led by Bert Blyleven but he was the only pitcher. In the unofficial results Bob Caruthers finished close to the top of the second twenty in which pitchers were a little closer on the ground, but not thick on the ground.

With a Clark Griffith here and a Goose Gossage there, counts of HOF players and members of shadow halls of fame may vary. (In other words, I don't have details at hand and I'm not checking.) Last year comparing Cooperstown and several of its shadows, I found the Coop at the high end in its number of pitchers. The shadows commonly include six to ten pitchers fewer, either in total or when they get to number 230 in a complete ranking.

We can't explain the entire difference by reference to Bruce Sutter.

Prelim. Probably my final ballot.

1-Walter Johnson
2-Cy Young
3-Pete Alexander
4-Christy Mathewson
5-Smokey Joe Williams
6-Kid Nichols
7-Eddie Plank
8-Ed Walsh
9-Rube Waddell
10-Three Finger Brown
11-Stan Coveleski
12-Red Faber
13-Rube Foster
14-Amos Rusie
15-Jose Mendez
16-Clark Griffith
17-Eppa Rixey
18-Joe McGinnity

Perhaps it is because the HoF has been more generous with pitchers than us but we haven't had a HoM not HoF guy come up yet. If elected, Dick Redding would be the first from this group.

"yet"?
We do have Bob Caruthers, who came up last month, iiuc. (February, please tell me last month was February!)

They beat us to the punch by electing Jose Mendez late in February 2006
(1971 and 1970 rankings, Hall of Merit)
but they corrected a couple of outstanding oversights then, too.

It's April 2. Last month was March. How seldom do I get to correct Paul Wendt?

- Brock

True, I never think of Caruthers as strictly a pitcher so he slipped my mind. Still, the HoF electing more pitchers through these eras is worth noting. It may be that the statistics look more impressive compared to modern candidates.

Regarding Ed Walsh and also the 24-page study I did back in 1994, which hasn't gotten to this thread yet (NOT Mike Emeigh's fault; I sent him the scans two weeks late because I was sick).

Since the study isn't up yet, and it's Friday, here's a quick and dirty table I made to see what the relationship is between debut dates and a pitcher's highest seasonal workload, just for the pitchers we are considering. The columns represent Name, Debut Year, Age at Debut, Highest Workload in any Single Season, Career IP, and Career ERA+, all from BB-Ref. It's sorted by Debut Year.

NAME DEBUT AGE HIGH IP ERA+

Amos Rusie 1889 18 549 3770 129
Cy Young 1890 23 488 7355 138
Kid Nichols 1890 20 488 5056 140
Clark Griffith 1891 21 353 3386 121
Rube Waddell 1897 20 383 2961 135
Joe McGinnity 1899 28 434 3441 120
Christy Mathewson 1900 19 391 4781 135
Eddie Plank 1901 25 357 4496 122
Mordecai Brown 1903 26 343 3172 138
Ed Walsh 1904 23 464 2964 146
Walter Johnson 1907 19 372 5915 147
Pete Alexander 1911 24 389 5190 135
Eppa Rixey 1912 21 313 4495 115
Stan Coveleski 1912 22 315 3082 127
Red Faber 1914 25 352 4087 119

The major feature of the big study was to demonstrate that pitcher workloads have dropped over time in a reasonably smooth curve from 1884 down to today (before 1884, the staff sizes and schedule lengths produce results that are actually lower than 1884, which is the high-water mark). I computed the curve by averaging the 3rd through 10th highest workloads in a season, which represents solid, uninjured top starters, but excludes outliers like Denny McLain in 1968. With Year as the X axis and IP as the Y, the curve resembles a first quadrant parabola, starting high left, dropping down quickly, "turning the corner" right after the turn of the century, and leveling out by 1910. Since then, it's dropped about 100 IP, from about 320 to about 220. But it dropped about 200 IP from 1884 to 1900.

One of the conclusions that I drew about early pitching is that you can't really compare pitchers whose debuts are more than 5 years apart. You can do that now, but not until about 1920 or so. This particular group we're dealing with here is the case group. You can't really compare, say, Amos Rusie to even Rube Waddell. They pitched in completely different eras for workloads. But you can compare Rusie to Cy Young, Kid Nichols, and Clark Griffith.

As you can see, there is a general relationship even in this small sample size with no checks for outliers; as debuts get later, the highest year's workload gets less. The dropoff is pretty dramatic at the beginning and less at the end, which is what the study's curve said would happen. Joe McGinnity and Ed Walsh are the outliers, with Ed really out there. Griffith is a minor outlier in the opposite direction.

Also of importance is that the table divides into three reasonably tight groups with a couple of inbetweeners. 1889-1891 is a group, with Clark Griffith as the light workload outlier. Waddell is in between that one and 1899-1904, or you can include him and expand the group to 1897. The outliers in this group are McGinnity and Walsh, who are the two highest outliers in any group, which fits with the idea that workloads are "turning the corner" at the very beginning of the century. Johnson is all alone in 1907, and then there is a group between 1911-1914. This last group has settled down a bit in terms of range; Alexander is the main outlier, but also the earliest. Again, this is what my curve suggested would happen: Things would level off after century's first decade.

Please remember that this chart lists the highest workload that the pitcher ever received. That is, it's composed of outliers that I discarded in the big study. The big study curve is smoother and also lower than these numbers.

One thing I decided was that Walter Johnson was perfectly placed to be the best pitcher in history. The curve has just turned, so it's the time where you get the highest workloads that do NOT break your arm. Not taking anything away from Walter; the best player at almost anything is usually placed at the best time in history for that skill.

I found other stuff in there. For example, do you have McGinnity ranked higher than Waddell? Well, if you added 500 IP to Rube, you'd just about get Joe's career IP. Do you think that if that had happened, Rube's career ERA+ would have dropped 15 points from 135 to 120? If so, why? Because Joe didn't debut until age 28? Well, is that a hindrance or a help? After all, light early workloads do correlate with long careers, and Joe's early workload was not in the majors, while Rube's was.

Anyway, the 24 pages of study cover a lot of what I found, and extends down to 1994, so only about the first 10 pages are germane here. And no one really has time to digest it all in the two days left to ballot. So here's the quick and dirty version, in case if helps anyone sort things out. I know it helped me, but that was back in 1994.

Enjoy! - Brock

Note that the trio surrounding Matty debuted at ages 28, 26, 25. Faber and Alexander later debuted at 25 and 24.

Here are the differences between debut dates and debut ages in Brock's table.
(errorprone)
=1867
67 Young

70 Nichols
70 Griffith
71 Rusie
71 McGinnity

76 Plank
77 Brown
77 Waddell

81 Mathewson
81 Walsh

87 Alexander
88 Johnson
89 Faber
90 Coveleski
91 Rixey
=1891

birth years
79 Foster

86 Williams
87 Mendez
89 Rogan
90 Redding (niHOM)

brock on Brown
He debuted at age 26, which is the second-oldest in this group, and he'd been a semi-pro third baseman. You'd think that, with that lack of young workload, he'd have an Iron Arm, but he didn't. That's the weakness of his career; the lack of total IP fueled by moderate workloads. Why? Well, I have a guess. My guess is that Brown did not have what we think of as a real top major league arm. What he had was a weird hand that generated weird curves. But that meant that sooner or later, his arm was going to lose enough steam that even the curves would not save him. That's what appears to have happened. Best guess I can muster up.

He debuted at 26 and his annual workloads were moderate. So he would have wracked up high career innings only by pitching to age 46, and that is not much residual for explanation.

Why were his workloads moderate? Why lighter workloads for pitchers on the West Side and heavier workloads on the South Side of Chicago? Both Brown and Walsh worked a lot in relief but Walsh also carried a heavy starting load and Brown did not. (No Cubs carried heavy loads.)

brock on Faber

14. Red Faber
He was already 25 when he hit the bigs, so early workload was not a problem. And still, they found a way. Faber had a real good year in 1921 at the age of 32. Pitched 331 innings; led the league in ERA. Not satisfied, the Chisox decided to see what he could do with 352 IP. He led the league in ERA again. But then, in 1923, at the age of 34, the IP go down 120, and the ERA goes up .6. He would pitch for 10 more seasons, but only reach as many as 200 IP three more times. Essentially, after 1923, he's a semi-starter, and that goes on for a whole decade, piling up bulk. The odd thing, though is that he never got bad, just a little above average. That's why his ERA+ stayed up at 119. Red never put in those four lousy years that drop it down. He just played out the string as a very acceptable 3rd or 4th starter.

The Chisox in 1921 and 1922 were unusual, a marquee franchise nearly ruined. Did Comiskey invest in the team?

Red Faber is sometimes overlooked in the black sox, clean sox story. Following military service in 1918 he pitched poorly in short innings during 1919. We have the Dickie Kerr story because Faber was not available for World Series work. He made a strong comeback in 1920 and he had been very good in 1916-17, capped by three wins in the 1917 World Series. He was one of the resources available to Comiskey and Co. Anyone looking for someone to blame for egregious overload of Red Faber should make sure to consider Comiskey. (Note, Comiskey under Von der Ahe was a leader in moderating the workloads of starting pitchers.)

Paul did decide to comment, "Note that the trio surrounding Matty debuted at ages 28, 26, 25. Faber and Alexander later debuted at 25 and 24.

Here are the differences between debut dates and debut ages in Brock's table."

On the ages surrounding Matty, yes. I include Walsh in the "cluster" debuting between 1899 and 1904. Ed's debut age was 23. So what you have in that subgrouping is a kid in Mathewson and four guys with some years behind them when they hit the bigs. I'm not at all sure what to make of that. Maybe the expansion of the majors when the AL got going caused a lot of top minor leaguers to make the bigs, and some of them turned out to be better than their minor league records had suggested. Just the laws of chance would suggest that might happen.

About the "differences." Do you mean "between BIRTH YEARS and debut ages"? That seems to be what you've charted, and a very nice helpful chart it is, especially including the Negro Leaguers. I had remembered Rogan as being later than Willliams, but you have them born only 2 years apart.

Paul also asks, "Why lighter workloads for pitchers on the West Side and heavier workloads on the South Side of Chicago?"

I don't know for sure, but I'd start by assuming that Frank Selee and Frank Chance were better pitchers' managers - or at least better at seeing the drift down in workloads - than Fielder Jones and Hugh Duffy. Part of the manager's job at that time was to find ballplayers. The Cubs of this time generally had four or even five viable strong starters, even if they were guys who weren't that good anywhere else. The Chisox had Ed Walsh and maybe a #2.

At ages 26.6 to 29.1 in 1915 to 1917, Faber pitched only 299, 205 and 248 innings (mean 251).

Here are some more famous pitchers of the 1920s, selected for their relatively old ages.
birth years
83 *Quinn

90 *Shocker
90 Luque
91 *Mitchell
91 Vance
92 Cooper
93 *Grimes
94 Pennock

* grandfather spitballists

Here the quintet from bottom of #59 are listed by birth year. Otherwise #59 lists quasi-birthyears derived from the socalled baseball ages in debut seasons.

86, Williams
87, Mendez
87, Alexander
87, Johnson
88, *Faber
89, *Coveleski
89, Rogan
90 Redding
90 *Shocker
90 Luque
91 *Mitchell
91, Vance
91, Rixey
92 Cooper
93 *Grimes
94 Pennock

* grandfather spitballists
, Hall of Merit members

In talking about Brown, Paul writes, "Why lighter workloads for pitchers on the West Side and heavier workloads on the South Side of Chicago?"

Brock already answered this, and I'm in agreement with what Brock said. To reemphasize, Selee and Chance had a plan (perhaps also embraced by Clarke in Pittsburgh) that was essentially new to baseball: spread the load more evenly among more pitchers than anyone else. And it worked spectacularly well - with perhaps a large part of the underlying reason being the defensive support those teams could give to a pitching staff. The real story in the success of the Cub pitchers (and Brock alluded to this as will) isn't so much how well Brown pitched - it's how well Overall and Pfiester and all the other short-term successes pitched.

that was essentially new to baseball: spread the load more evenly among more pitchers than anyone else.

I'm not sure about this innovation.
Comiskey spread the load more than Anson.
Hanlon spread the load more than Ward.

Link to previous results

https://www.baseballthinkfactory.org/hall_of_merit/discussion/election_results_johnson_and_young_are_1_and_2_on_every_ballot_among_the_hu