bump

We're talkin' 25 WS difference in a career between an OK corner OF and a horrible one--8 extra games won or not won. To me the impact seems damn small and 8 games or 25 WS seems sufficient. And if not, deduct another 2.5 WS per year from F. Howard and he remains a candidate.

Dan R.,
I believe that you calculate your standard deviations using individual player stats. Does it follow necessarily that a low standard deviation among players corresponds with a low standard deviation among teams?

Also, the difference between average OPS+ for second basemen and shortstops in Nellie Fox's time was 1 or 2. How can there be a 15 point difference in replacement level?

one more time for us all to remember: using Win Shares, hitters in DH leagues (AL 1973ff) are underrated. We may disagree on the exact amount, but two players of equal goodness get between 5% and 11% fewer batting win shares in the modern AL than in the NL. Crucial when comparing Nettles/DwEvans/Lynn/ etc to others.

Funny you should mention this.

I realized earlier this week that I neglected to include the dh adjustment. As a result, there are changes on my ballot. Both Munson and Singleton jump on my ballot and look PHOM right now. Nettles and Dw. Evans are also big gainers.

Great post, TomH:

"one more time for us all to remember: using Win Shares, hitters in DH leagues (AL 1973ff) are underrated. We may disagree on the exact amount, but two players of equal goodness get between 5% and 11% fewer batting win shares in the modern AL than in the NL. Crucial when comparing Nettles/DwEvans/Lynn/ etc to others."

I don't like WS, but if you're gonna use 'em, use 'em right!

What's your take on adj OPS+ in the DH era?

look, I make a DH adjustment and all

By why would doing so be using them right? They are tied to actual wins and if there are more real hitters there will be fewer WS per hitter so individual hitters have less of an impact. Makes sense, even if I don't like to use it.

Again, why would someone who doesn't want to take the DH into effect be using WS 'wrong'?

Cblau,

You would expect that a low stdev among players would correlate to a low stdev among teams. Since talent is not distributed equally among teams, the bigger the spread among players, the bigger the spread among the players on the haves and the players on the have-nots. But I haven't tested that empirically.

I made up the 15-point number. The difference between 2B and SS in the worst-regulars average in the 1950's AL is 0.6 wins (2.4 wins below average for 2B, 3.0 for SS). Then you have to add on the fact that the worst SS regulars from 1985 to 2005 were 0.3 wins better than the freely available level as calculated by Nate Silver, while the worst 2B regulars from 1985 to 2005 were at exactly the freely available level. So I get a gap of 0.9 wins overall. I'm not sure how many OPS points 0.9 wins per 162 games is, but you could find out pretty easily with the MlVr formula, I imagine.

I suppose the next thing is you're gonna tell me that we need to adjust for the fact that pitchers don't throw as many innings as they used to. Bunch 'o wusses, you ask me.

;-)

But seriously, the conditions of the game have changed in a lot of ways over the years. If individual pitchers had more value 100 years ago, God bless 'em. If individual hitters had/have more valuable w/o the DH, God bless 'em.

I mean, I'm not opposed to the adjustments, but how "perfect" of a test tube game can you really construct? I would guess, e.g., that "the pool" has a lot more impact on the shape of things than the DH. How much should we discount white players before integration? etc. etc. etc. I am half-facetious here. Like the man said, if your WS don't add up to actual W/3 then whatcha got?

if your WS don't add up to actual W/3 then whatcha got?

How about, A more accurate rating of the individual player, separate from his teammates' influences.

"right" and "wrong" - maybe they were not the "right" words to use :)

Win Shares divide team wins into players. So actually, if your goal is only to determine how many of the Yankees 114 wins in 1998 should be given to each player, then you should obviously NOT try to 'adjust' for the DH, since the DH is actually what was used in (most) Yankee games.

OTOH, if we wish to compare Jeter's value with Sosa's that year, we need to recognize that it lineups with 8 decent batters function diffferntly than those with 9. To be silly, if one league began using 10 man lineups (short fielders, or pitcher + DH), then obviously each batter would hit less often, and so would be less 'valuable' in terms of WS, WARP, and even traditional measures like RBI. But it sure doesn't make the player worse, it only means that he was stuck in a bad spot, kind of like being unjustly in the minor leagues for 10% of the year.

I don't actually disagree with any of that, TomH et al. To me, when all is said and done, it's less about the WS than it is a player's relative position among his peers. The 3rd best pitcher of the '90s = the 3rd best pitcher of, well, the '90s (different century) until proven otherwise. Of course, if the 5th best pitcher of one decade has 2X as many WS as the 2nd best from another, then maybe we've proven otherwise (and maybe not).

But 90 percent of the proof is big stuff. And sometimes I think we sweat the small stuff so much that we lose the forest in the trees.

That said, what is the "DH adjustment"? The short-hand, that is. As in when we compare an AL position player to an NL guy? 10 percent?

It seems the DH adds about .4 to .5 run per game, which = about 10% more runs. If it takes 10% more runs to equal 1 win, then we need to adjust (offense only) WS by 10%.

Someone may need to remind us how (if) WARP adjusts or needs to for the DH. Probably in the uberstat thread somewhere.

WARP1 most CERTAINLY does NOT adjust for the DH. WARP2/3 do, one would assume.

This is only tangentially related, but in thinking about rating guys from the 70's (such as in DH leagues), I looked at how many players are in the HOM who played in each period from 1910 to 1980. Specifically, I looked at how many players were active in five year increments that we elected. The short summary is that the period from 1925 to 1945 is severely overrepresented (both among African-Americans and White ballplayers, the modern game (70's for example) is severely underrepresented. Some of this will take care of itself as some obviously qualified candidates who came up in the 70's become eligible, but certainly not all of it. Yet when you look at many voters' lists, they are filled with old-time baseball players. There is a thread going on about Bus Clarkson, who played much of his career in a time that is already overrepresented. All of this despite the fact that there is probably a larger talent pool today in the last thirty years than ever before.

Anyway, here's the list:

Min. White Total

10 7 23 30
15 9 28 37
20 12 20 32
25 19 29 48
30 17 31 48
35 16 27 43
40 16 31 47
45 15 28 43
50 9 22 31
55 9 24 33
60 10 23 33
65 12 25 37
70 14 22 36
75 10 18 28
80 5 13 18

Some of this also feeds into Dan's work, so I'm hoping to start a discussion.

To me, when all is said and done, it's less about the WS than it is a player's relative position among his peers.

I know its our custom in the HoM to be polite to all belief systems, no matter how irrational, but to cling to this idea when its been shown a thousand times to be erroneous verges on stupidity.

'zop, I'll assume I misunderstand your point, because I completely disagree.

Who was a better pitcher: Christy Mathewson (more dominating numbers) or Tom Seaver (ranks higher among peers)? Neither line of argument is erroneous.

I know its our custom in the HoM to be polite to all belief systems, no matter how irrational, but to cling to this idea when its been shown a thousand times to be erroneous verges on stupidity.

Isn't it possible to explain why you think someone's position is wrong without saying they are verging on stupidity, 'zop?

>its been shown a thousand times to be erroneous

OK, make it a 1,001 and I'll consider it. How exactly has this been shown to be wrong? If you mean because there are differences between player A and player B (the 3rd best pitcher of the 'oughts and the 3rd best pitcher of the '90s, etc. etc. etc. etc.), well, of course. I don't disagree.

If you are appealing to the "pool," then, yes, help me. What does that show, again? Or maybe you're appealing to standard deviations? Or something else? How exactly is it wrong to care where a player ranks among his peers?

Where a player ranks relative to his peers is irrelevant to his merit, except insofar as his peers define the distribution of the environment he plays in. If I'm an 2B, and I have the fortune (misfortune) to play in a golden era for great 2B's, and there are 3 other guys better than me but any one of the 3 of us would have been the best 2B's of the generation before or after, I am not "worse" because of where I rank relative to my peers. Phil Rizzuto is not "worse" because he played at the same time as Pee Wee Reese.

A player's peers DO define their value, in that the ease of dominating an environment and the replacement level is defined by ones peers. But the random distribution of outlier talent among positions does not impact a player's contributions.

I can see why using a player's rank among his peers is a tempting technique. It crudely approximates Dan R.'s standard-deviation work, since it does measure a portion of the distribution curve that defines ease of dominance. But it is a highly imperfect way to do this! Trying to project back the ease of domination based upon outliers is folly.

In addition, it has nice parallels with the Keltner test and other non-quantitative measures of "Fame". But Fame is not Merit, and thus this should have no bearing in this discussion.

To refer back to Dan R.'s work for a second; Dan is a big fan of Concepcion. He supports Concepcion NOT BECAUSE HE'S THE BEST SS OF HIS ERA. Dan has gone back and looked at ALL SS of his era and shown that it was harder for each SS to earn wins for his team during that era (both the baseline AND the spread of WARP are at historic lows). If you want to incorporate generational context into your rankings, that's the only appropriate way to do this.

If we were voting for the hall of "fame", then using rank relative to peers would be entirely appropriate. But this group purports to measure "merit", and ranking among peers is not a good measure of merit, especially when there are so many other superior techniques available.

You can rank players based upon total career hits, and you'd get a list, and it would probably have a bunch of HoM hitters on the top, but that doesn't mean its a good measure of Merit, or that using such a list as a primary criterion isn't stupid.

I think people are arguing apples and oranges here. If you define peer as people who played the same position at the same time, its one thing. If you define peer as people who played at the same time, regardless of position, its another.

For example, the second best thirdbaseman playing in the 70's and 80's was George Brett. The third best first baseman playing in the mid 30's was Hank Greenberg. Does it then follow that the second best third baseman playing in the mid 30's, whoever he was, was as good as Brett? Was the third best firstbaseman playing in the early 80's as good as Hank Greenberg? Of course not. Was Dave Stieb (arguably the best pitcher of the '80s) as good as Tom Seaver or Roger Clemens? Of course not.

At the same time, the tenth or twentieth best player playing in the 1970's shouldn't be rated substantially differently than the tenth or twentieth best player in the 30's, IMO. We have put into the HOM 48 players playing in the late 20's/early 30's, and barely over 30 that played in the '50s. There's something wrong with that, IMO.

dizzy pac'ed some good logic in there

Dizzypaco--I know you don't tend to agree with me about anything, but isn't that supportive of my argument that the group isn't accounting sufficiently for standard deviations, since the stdev of the 20s/30s was far higher than that of the 50s?

Does anyone know if the following stats in Sinin's BB encyclopedia - RCAA, OWP - take out pitcher's hitting before computing them?

isn't that supportive of my argument that the group isn't accounting sufficiently for standard deviations

Not really. I think what happened was much simpler than that.

The number of selections from MLB in the the 20s/30s is pretty similar to the number of selections in the periods before (back to the 1880's) and after (up to expansion). It's just that the NeL selections were made IN ADDITION TO the MLB selections, not IN PLACE OF some MLB selections. The effect has been to treat it like a 24 team era; 16 white and 8 black (approximately, based on ratio of selections). I think that this can be justified.

What is probably not justifiable is the way that the 1940's/1950's have been treated. It is unlikely that the talent level went down that drastically when compared to the 30's. So it is arguable that a similar numbers of players should be elected from the generations of this era, the one bridging between the 30's and the 70's (by an argument similar to the one that justifies the 1930's bulge). The 1950's should not be treated in the same way with the same number of selections as the "deadball" 16 team era, just because the number of teams was the same.

Hail the co-leaders for "most HOMers in a league in the 20th century, 10 G minimum":

1964 NL (26) - Spahn, Snider*, Pierce*, Mays, Mathews, Banks, Aaron, Koufax, Clemente, Boyer, Bunning, Drysdale, FRobinson, BGibson, BWilliams, McCovey, Santo, Marichal, Torre, Stargell, GPerry, Allen, Rose, JWynn*, Morgan*, Niekro*
1969 NL (26) - Mays, Banks, Aaron, Clemente, Boyer*, Bunning, Drysdale*, BGibson, BWilliams, McCovey, Santo, Marichal, Torre, Stargell, GPerry, Allen, Rose, JWynn, Morgan, Niekro, Jenkins, Sutton, Bench, Seaver, Carlton, DaEvans*
1970 NL (26) - Mays, Wilhelm, Banks*, Aaron, Clemente, Bunning, BGibson, BWilliams, McCovey, Santo, Marichal, Torre, Stargell, GPerry, Allen, Rose, JWynn, Morgan, Niekro, Jenkins, Sutton, Bench, Seaver, Carlton, DaEvans*, TSimmons

The pre-1960s leader was 1926-28 AL, with 20.
The NL has matched or exceeded that 20 mark every year from 1961-73, while the AL is staggering along with 10 to 13 in that whole span.

Negro Leagues high was 17 in 1924.

The three NL years are now chasing only 1893 NL (27) and 1892 NL (30) - but they're almost out of steam.
Tony Perez, 6th among backloggers, would count in all 3 1960s seasons, as would No. 19 Rusty Staub. No. 31 Bobby Bonds would add to the 1969-70 counts, and No. 32 Orlando Cepeda would count in all 3. No. 33 Dave Concepcion helps only in 1970, No. 34 Lou Brock counts in all 3, and No. 35 Ken Singleton is only in 1970. No. 42 Frank Howard scores one for 1964.
Next is No. 85 Al Oliver in 1969-70, No. 93 Steve Garvey with 1970, and fellow No. 93 Bill Mazeroski with 1964, 69, and 70.



1892 (30) - Anson, O'Rourke, Kelly, Ward, Bennett*, Gore*, Brouthers, Glasscock, Richardson*, Galvin, Keefe, Ewing, Connor, Stovey, Clarkson, Thompson, Delahanty, Rusie, Hamilton, Nichols, Burkett, McPhee, GDavis, Young, Caruthers*, Childs, Dahlen, Kelley*, Keeler*, Jennings

Just made this one up, might be of interest:

1850s, HOMers per year - x-x-x-x-x-x-1-1-1-1

1860s, HOMers per year - 1-2-2-2-3-2-2-4-6-8

1870s, HOMers per year - 9-10-12-12-12-12-12-11-12-16

1880s, HOMers per year - 17-20-21-20-21-23-24-23-25-25

1890s, HOMers per year - 30-32-31-28-23-24-23-22-22-24

1900s, HOMers per year - 23-25-25-23-25-26-25-26-28-28

1910s, HOMers per year - 29-28-27-29-27-29-33-27-24-26

1920s, HOMers per year - 28-30-34-36-41-44-47-46-46-43

1930s, HOMers per year - 42-43-45-43-41-41-41-42-39-41

1940s, HOMers per year - 44-43-39-29-20-22-34-34-34-27

1950s, HOMers per year - 27-28-25-27-28-32-33-30-30-31

1960s, HOMers per year - 31-32-33-37-36-34-35-35-35-36

1970s, HOMers per year - 36-34-35-33-33-30-28-24-21-20 (increasing numbers not eligible yet)

This again is minimum 10 G

Howie,

For the way years, are you counting players in the war during those years? My guess would be no since you stipulated that there was a 10g minimum. What would those years (43-44-45) look like if Feller, Keller, Williams, DiMaggio et al were inserted? They look like a low point, but I would have to imagine that is because of the war and should not be the cause for alarm that they seem to be.

I mean we shouldnt' look into electing Snuffy Stirnweiss!

Well, here are the 1940s, I'll let you tackle that one!

NL
1940 (11) - Hartnett*, Waner, Ott, Hubbell, Vaughan, BiHerman, Hack, Medwick, Mize, Slaughter, Reese
1941 (12) - Hartnett*, Waner, Ott, Hubbell, Vaughan, BiHerman, Hack, Medwick, Mize, Slaughter, Reese, Musial*
1942 (12) - Waner, Ott, Hubbell, Vaughan, Foxx*, BiHerman, Hack, Medwick, Mize, Slaughter, Reese, Musial
1943 (8) - Waner, Ott, Hubbell*, Vaughan, BiHerman, Hack, Medwick, Musial
1944 (6) - Waner*, Ott, Fox*, Hack, Medwick, Musial
1945 (4) - Ott, Foxx, Hack, Medwick
1946 (9) - Ott*, BiHerman, Hack, Medwick*, Mize, Slaughter, Reese, Musial, Kiner
1947 (13) - Vaughan*, Greenberg, BiHerman*, Hack*, Medwick*, Mize, Slaughter, Reese, Musial, Kiner, Spahn, JRobinson, Snider*
1948 (13) - Vaughan*, Medwick*, Mize, Slaughter, Reese, Musial, Kiner, Spahn, JRobinson, Snider*, Campanella, Ashburn, Roberts*
1949 (11.9) - Mize*, Slaughter, Reese, Musial, Kiner, Spahn, JRobinson, Snider, Campanella, Ashburn, Roberts, Irvin*

AL
1940 (19) - ASimmons*, Lyons, Grove*, Foxx, Ruffing, Gehringer, Cronin, Dickey, Averill*, Appling, Greenberg, DiMaggio, Feller, Gordon, TWilliams, Boudreau, Keller, Doerr, Newhouser*
1941 (17) - Lyons, Grove*, Foxx, Ruffing, Gehringer, Cronin, Dickey, Appling, Greenberg, DiMaggio, Feller, Gordon, TWilliams, Boudreau, Keller, Doerr, Newhouser
1942 (14.7) - Lyons, Foxx*, Ruffing, Gehringer*, Cronin*, Dickey, Appling, DiMaggio, Gordon, TWilliams, Boudreau, Keller, Doerr, Newhouser, Wynn
1943 (10) - ASimmons*, Cronin*, Dickey, Appling, Gordon, Boudreau, Keller, Doerr, Newhouser, Wynn
1944 (5) - Cronin*, Boudreau, Doerr, Newhouser, Wynn
1945 (7) - Ruffing*, Appling*, Greenberg, Feller*, Boudreau, Keller*, Newhouser
1946 (14) - Ruffing*, Dickey*, Appling, Greenberg, DiMaggio, Feller, Gordon, TWilliams, Boudreau, Keller, Doerr, Newhouser, Wynn*, Lemon*
1947 (13.8) - Ruffing*, Appling, DiMaggio, Feller, Gordon, TWilliams, Boudreau, Keller*, Doerr, Newhouser, Wynn, Lemon, Doby*, Berra*, WBrown*
1948 (15) - Paige*, Appling, DiMaggio, Feller, Gordon, TWilliams, Boudreau, Keller*, Doerr, Newhouser, Wynn, Lemon, Doby, Berra, Pierce*
1949 (15.1) - Paige*, Appling, DiMaggio*, Feller, Mize*, Gordon, TWilliams, Boudreau, Keller*, Doerr, Newhouser, Wynn, Lemon, Doby, Berra, Pierce

NeL (and foreign)
1940 (14) - Mackey*, JWilson, CPBell**, Stearnes, Suttles, Wells**, Dihigo**, JGibson**, RBrown, Leonard, WBrown**, Trouppe**, Campanella*, Irvin
1941 (14) - Mackey,*, JWilson, CPBell**, Suttles, Wells**, Dihigo**, JGibson**, Paige, RBrown, Leonard, WBrown, Trouppe**, Campanella*, Irvin
1942 (12) - JWilson, CPBell, Wells, Dihigo**, JGibson, Paige, RBrown, Leonard, WBrown, Trouppe**, Campanella, Doby
1943 (10) - JWilson, CPBell, Wells**, Dihigo**, JGibson, Paige, RBrown, Leonard, WBrown, Trouppe**, Doby
1944 (9) - CPBell, Wells**, Dihigo**, JGibson, Paige, RBrown, Leonard, Trouppe**, Campanella
1945 (11) - JWilson*, CPBell, Wells*, Dihigo**, JGibson, Paige, RBrown, Leonard, Trouppe**, Campanella, JRobinson
1946 (11) - CPBell, Wells*, Dihigo**, JGibson, Paige, RBrown**, Leonard, WBrown, Trouppe**, Campanella, Irvin, Doby*
1947 (7.1) - Wells*, Paige, RBrown**, Leonard, Trouppe**, WBrown*, Irvin, Doby*
1948 (6) - Wells*, RBrown**, Leonard, WBrown, Trouppe**, Irvin

The way I counted it was any season in between when a player played his first and last game, in order to avoid the war issue. I excluded the single game novelty appearances. When you do that, its obvious that the 1940's are not underrepresented.

Dizzypaco--I know you don't tend to agree with me about anything, but isn't that supportive of my argument that the group isn't accounting sufficiently for standard deviations, since the stdev of the 20s/30s was far higher than that of the 50s?

Dan, I actually agree with many of your general theories, its the execution that I often disagree with. I agree with you that the 70's were harder to put up dominating numbers than the 30's, and therefore we should consider that when considering the merits of individual players. I'm just not sure I agree with your method of adjusting for it.

I agree with jimd - the talent pool rose from the 1860's to 1930's, as did the number of teams (when combining the major leagues and Negro Leagues). There was a contraction in the number of teams beginning in the 40's - but no corresponding drop in the talent pool. Furthermore, there was a drop in the number of leagues - meaning the number of players leading their league in something, or being the best in the league at something also dropped. Finally, I'm a believer that periods of high offense are much easier to dominate than other periods. All of these combined to create illusions in the statistics.

One of the problems I have with many voters is that while they pay lip service to the generational imbalance, they often then ignore it in their voting patterns. Despite the fact that we know the 30's are severely overrepresented, many voters are still pushing for even more electees, whether it is someone like Bob Johnson or yet another Negro League star.

Diz,

The one thing you also have to remember is that a voter may really think that Bob Johnson relaly belongs in the HOM while also thinking the 1930's are overrepresented. Maybe they disagree with a number of players but they have voted for Bob JOhnson since the 1950's and this doesn't effect their view of him. What it shoudl effect is their view of the guys who are on the outside but haven't been favorites.

In other words, I am not going to stop voting for a guy because I think his era is overrepresented if I really believe he is a HOMer. Maybe I don't mind overrepesentation, maybe I think we made 3 or 4 mistakes in that era, but if I really like a guy I am not going to stop voting for him because of era overrepresentation. For instance i am voting for Dizzy Dean. I am not going to stop voting for a player that I believe belongs in. If it were up to me we would take out Red Ruffing and put Diz in. However, I may drop someone like Wally Berger, Chuck Klein, or Tony Lazzeri because of this.

Wow, that was long and confusing. Did it make any sense?

Wow, that was long and confusing. Did it make any sense?

It makes sense, as long as one acknowledges that many of the statistics compiled by players active at this time should not be taken at face value, including WARP, win shares, OPS+ or anything else - which most advocates of some of these players don't do. To me, it affects Dizzy Dean a lot less than Bob Johnson.

Dizzy Dean may have been, for a brief period of time, the best pitcher in baseball. If not the best, certainly in the top two or three, even counting the Negro Leagues. If you place a lot of emphasis on peak, its hard not to support Dean, even given the number of players from this era already elected. I can understand thinking that Dean should have gone in long before pitchers already elected, if you place a lot of emphasis on peak.

Bob Johnson, on the other hand was never one of the best players in the league. His career was relatively short, and his peak performance was miles short of Gehrig or Foxx or Greenberg or DiMaggio or several others - and that's not even counting the Negro Leagues. He was certainly very good - if you take his adjusted numbers, and put them in the 50's, 60's, or 70's, I'd say he had a very good case. But he didn't - he played in an era where there were a bunch of guys just like him, and several more who were clearly better. To me, Bob Johnson is the perfect example of a guy who's numbers are very impressive only if you ignore the context in which he played.

So if you do support Johnson, or someone like him, I think you need to acknowledge one of the following: Either that there should be far more players in the HOM from the 1930's than any other era, or that there are several players (not just one or two) that played at the same time as Johnson, are less deserving that Johnson, and should not have gone in. And then name them. I'm not, by the way, suggesting that either belief is crazy, just that I disagree with both statements.

"What is probably not justifiable is the way that the 1940's/1950's have been treated. It is unlikely that the talent level went down that drastically when compared to the 30's."

I disagree with this - I think talent was down in the early 1950s. Integration wasn't enough to offset the loss of a significant chunk of the 18-25 year old population in WWII, as well as the increasing loss of top athletes to other sports, like the NBA and NFL.

If we were voting for the hall of "fame", then using rank relative to peers would be entirely appropriate. But this group purports to measure "merit", and ranking among peers is not a good measure of merit, especially when there are so many other superior techniques available.

This makes no sense at all to me. Merit IS how well you performed vs. your peers. That's the guys you played against. If Player A provided more wins vs. his peers in his era than Player B did in his, then he is more "meritorious".

"look, I make a DH adjustment and all

By why would doing so be using them right? They are tied to actual wins and if there are more real hitters there will be fewer WS per hitter so individual hitters have less of an impact. Makes sense, even if I don't like to use it.

Again, why would someone who doesn't want to take the DH into effect be using WS 'wrong'?"

Because the AL artificially adds offense to the game. WS does not account for this. It sounds dumb, but the wins don't add up. Since the AL added offense via the DH, but wins didn't increase accordingly, you have to add them to the offensive WS yourself. If you don't you are underrating AL players from that era, severely (I think 11%) underrating them.

If you don't add them, you are basically penalizing a player for playing in the American League 1973-present. It might not look pretty to just add them out of nowhere, but that's what the AL did when they created the rule. If you don't adjust for this, your system is wrong, IMO.

This makes no sense at all to me. Merit IS how well you performed vs. your peers. That's the guys you played against. If Player A provided more wins vs. his peers in his era than Player B did in his, then he is more "meritorious".

Of course. What I mean is that saying that a player was, hypothetically, "3rd best DH of the 1940's" is close to meaningless, because it doesnt distinguish between the following two scenarios:

(imagine an 8 team league with 8 DHs)
10WARP-10-9-3-3-3-2-2-2-1
8.2-8.2-8.0-2-2-2-2-1-1-1

To know the value any player had, you need to know his value relative to the guys at the top, AND in the middle and on the bottom.

Right, the point is that mere rank order among positional contemporaries isn't prima facie evidence of Merit.

(I think 11%)

I wanted to take a quick look at this. I put both leagues into a 4.50 R/G frame of reference so that low or high scoring years didn't skew things too much, along the way adjusting the non-p RC, the DH RC, and the P RC for both leagues. (The SBE here, of course; it's by-position data isn't perfect, it's approximate since it's not PBP based---those no NL DH totals despite interleague play). Then I simply figured the % of RC AL DHs contributed and the % of RC NL Ps contributed.

Over the period 1973-2005, the percentages are 10% and 2%. Actually 9.93% and 2.04% to be overprecise. DHs go between 8% and 12%. Pitchers are consistently 2% from the mid-70s onward.

Let's just say that 10% and 2% is accurate (might not be, of course). Does that mean that we should be saying that the players in DH leagues are getting an 8% haircut rather than 10%? (Or a 9% haircut if 11% is correct as Joe suggests.) If we say the full 10%, then we are effectively saying that the DH replaces only PHs in the lineup, and at a team or league level, that's not true. We might also be suggesting that WS does not account for pitcher batting, when that's also not true (even though WS deals with it in a not-quite straight forward manner).

Given the narrowness of the backlog, I think even this small potential difference of interpretation could be of vital importance to Singleton and Dw Evans among others, so it seemed like a good time to query it.

To adjust for the DH, I calculated how much lower the replacement level would have to be for AL position players so that a team of replacement hitters in the AL would be the same number of wins below average as a team of replacement hitters in the NL. The answer was 0.6 wins per position per full season.

Dan,

As an overgeneralized rule of thumb I'm getting a DH adjustment of 7-8% from the information you supplied in post 238. Is that accurate?

If so, we've independently arrived at essentially the same conclusion---maybe for the first time ever. ; )

Dr. Chaleeko,

I'm really not sure. I find Win Shares so inscrutable that I haven't spent any time thinking about what adjustments to apply to them. In my system, the DH adjustment is applied to the replacement player, rather than to the player being measured--a replacement player's performance in the AL will be 0.6 wins per season further below average than the same replacement player's performance in the NL. So a replacement third baseman might be 1.5 wins below average per season in the NL, but his offense would be 2.1 wins below average in the AL, since the average he is being compared to is higher due to the DH. A third baseman who is 8 wins above league average in the AL would thus be 10.1 WARP, while a third baseman who is 8 wins above league average in the NL would be only 9.5 WARP, a difference of just about 6%. By contrast, a third baseman who is 1 win below league average in the AL would be 1.1 WARP, while a third baseman who is 1 win below league average in the NL would be only 0.5 WARP, a 60% difference. So I don't think you can use a multiplier--it's straight addition.

To know the value any player had, you need to know his value relative to the guys at the top, AND in the middle and on the bottom.............Right, the point is that mere rank order among positional contemporaries isn't prima facie evidence of Merit.

OK, this is clearer - I thought you were impying something slightly different. I agree 100% that relative value vs. positional contemporaries is the correct place to be looking.

Since the AL added offense via the DH, but wins didn't increase accordingly, you have to add them to the offensive WS yourself. If you don't you are underrating AL players from that era, severely (I think 11%) underrating them.

We've been over this ground before, but once again I'll raise my disagreement - AL players are being 100% fairly valued. If offensive levels in one league are higher than the other league, whether it's the DH or something else (the NL could conceivably have higher runs/game in any given year) then that's the environment the player is playing in, and that's what context their value should be measured in.

looks like no one has tried the new Edd Roush book yet.
Here's another one from an early era:

Crazy '08: How a Cast of Cranks, Rogues, Boneheads, and Magnates Created the Greatest Year in Baseball History (Hardcover)

supposedly Mathewson, Chase, and Cobb figure prominently.
from Publisher's Weekly blurb:
"It's been almost a century since the loopy shenanigans of 1908 that produced what Fortune magazine editor Cait Murphy calls "the year that baseball comes of age," but the resultant drama has hardly faded with time. Although baseball books tend to sag with nostalgia, Murphy's wisecracking yarn digs right into the era's brawling, vivid ugliness with little regard for such niceties, and is all the better for it. Her book is so rife with corruption, greed, stupidity and downright weirdness that it makes today's sport of sanctimony and clean behavior look positively sleepy in comparison."

KJOK, BP WARP1 most definitely do *not* accurately measure the value of the environment DH league players are playing in. (WS just don't measure anything accurately, so I couldn't say how they handle the DH issue). However, attempting to show how has made me question whether my DH adjustment is too large.

Here's an example:

Let's take a guy who generates 100 runs. The presence or absence of the DH has no effect on his hitting, so assuming the quality of pitching is equal, he'll generate 100 runs regardless of whether he is in the AL or the NL.

BP defines offensive replacement level as two wins below league average hitting per 400 batting outs. Since 1973, the NL has averaged .169 runs per batting out (9.46 runs per win) and the AL .180 runs per batting out (9.89 runs per win). Thus, according to BP, a replacement hitter during that time has produced (.169*400) - (2*9.46) = 48.7 runs per 400 outs in the NL, and (.180*400) - (2*9.89) = 52.2 runs per 400 outs in the AL. This is *WRONG*--there is *no reason* why a replacement player would generate more runs in one league than in the other, assuming the quality of pitching is equal. The presence of the DH has no effect on a replacement player's hitting. But it's how BP does it.

As a result, BP would say that our 100-runs created guy is 51.3 runs above replacement in the NL, which is worth 51.3/9.46 = 5.4 wins, and just 47.8 runs above replacement in the AL, which is worth 47.8/9.89 = 4.8 wins, a gap of 0.6 wins, the same as the figure in my post #238.

However, KJOK is right to point out that runs are worth fewer wins in the AL than in the NL, which I don't think I was accounting for. So what is the "right" answer? It doesn't matter what you choose for replacement level, as long as that level is the same (in runs) for both leagues. So let's just take the average of 52.2 and 48.7 and say that a replacement player would actually produce 50.45 runs per 400 outs, making a 100-RC guy 49.55 runs above replacement. This means that in the NL, a 100-RC guy would be 49.55 runs/9.46 = 5.24 wins above replacement, and in the AL, a 100-RC guy would be 49.55 runs/9.88 = 5.02 wins above replacement. This is a much smaller adjustment of just 0.22 wins.

I need to think this through some more, but I may adjust my WARP to use 0.22 wins instead of 0.6 for the DH adjustment. That might be enough to bump guys like Toby Harrah and Buddy Bell out of my PHoM, which would put me closer in line with the consensus.

WS just don't measure anything accurately

if something is "inscrutable" to you, it does not logically follow that it is inaccurate. In fact, by your own admission you claim that you don't know :)

It may be daunting to deciper; afer all, it took James how many hundreds of pages to write the book? I find quantum physics quite daunting, and I have yet to claim that it is inaccurate; the problem may not be the concepts themselves, nor the people who try to explain the concepts to me.

What I mean is, the results of Win Shares just seem so wildly off base to me that it doesn't seem to be worth the time to determine how he gets there. It would be like poring over scholarly work arguing that the earth is flat. Since, empirically, it's round, why waste one's time figuring out why some people thought it was flat?

Since, empirically, it's round, why waste one's time figuring out why some people thought it was flat?

Because in order to find the truth you must ask why before you ask what.
And after.

Thus, according to BP, a replacement hitter during that time has produced (.169*400) - (2*9.46) = 48.7 runs per 400 outs in the NL, and (.180*400) - (2*9.89) = 52.2 runs per 400 outs in the AL. This is *WRONG*--there is *no reason* why a replacement player would generate more runs in one league than in the other, assuming the quality of pitching is equal.

I don't really want to defend WARP either, but sure there is a reason why a replacement player would generate more runs in one league than in the other league - the replacement players available are better, or the run environment is higher! As I've argued before replacement is actually a 'chaining' effect - if a player is replaced, he'll be replaced by someone off the bench, who's spot is replaced by another bench player, whose spot may be replaced by a AAA callup, etc. If one league is better, or plays in a higher run scoring environment, than another league, then that replacement will likely produce more runs.

if something is "inscrutable" to you, it does not logically follow that it is inaccurate. In fact, by your own admission you claim that you don't know :)

It may be daunting to deciper; afer all, it took James how many hundreds of pages to write the book? I find quantum physics quite daunting, and I have yet to claim that it is inaccurate; the problem may not be the concepts themselves, nor the people who try to explain the concepts to me.

But I think Win Shares have been analyzed enough to show it's many inaccuracies:

Win Shares Walk Thru

KJOK, that's true if you're comparing players across eras--one would presume a replacement player would produce more runs in 1894 than in 1968, because it was easier to hit in 1894 than 1968. But it is no easier to hit in the AL than in the NL (except in eras of different league strength, such as today)--AL second basemen don't outhit NL second basemen, and AL CF don't outhit NL CF (over the entire time period since the DH was implemented, I mean). The difference is due to the presence of the DH vs. the presence of pitchers in the lineup, none of which makes it any easier for a replacement player to hit in the AL. If the league strength is the same, one would expect a guy who hits .300 in the NL to hit .300 in the AL, no more and no less. But that .300 in the AL will be lower relative to the league average than it is in the NL, since the AL average includes DH's while the NL average includes pitchers. The value is almost the same, since the replacement will drop by the same number of runs relative to league average. But not exactly the same because, as you point out, the same number of runs above replacement buys fewer wins in the higher AL run environment.

It would be like poring over scholarly work arguing that the earth is flat. Since, empirically, it's round, why waste one's time figuring out why some people thought it was flat?

So now I and many others here are Flat Earthers because we feel there are fewer (at least more manageable) problems with WS than WARP?

Well, I clearly think that the best thing to do is to discard them and do your own analysis.

But if you are going to use them, the big advantage of WARP, I think, is that you get the page with all its components, so you can pick and choose. Basically, BRAA is very good--if there are any strong criticisms of it, I haven't seen them--so it's a good place to start. Then you can add in any sort of modified FRAA (or other defensive statistic), add in your own replacement levels, and come up with your own values. With WS, you just get the final BWS, FWS, and PWS, and attempting to back-calculate them to extract the usable components is a major (and unnecessary, I think) pain.
I guess I just strongly disagree that WARP's problems are less "manageable" than WS's--I think they're much *more* manageable. The final WARP1 and WARP3 numbers are just as batty as Win Shares are, but the nuts and bolts are there for you to use at your leisure.

Another way to make the point is to say that you don't have to "pore over" WARP to reverse engineer it to the same extent you do with WS. But I would have even stronger criticisms of FRAA and the WARP1-WARP3 difficulty adjustment than I have of WS--not only is their derivation not shown logically on the page, it's not explained *anywhere*. You just have to take them on blind faith. They're the weakest links in the WARP equation. I ignore WARP3 and begrudgingly use FRAA, averaging it with other metrics. If Michael Humphreys gets DRA for every player-season in MLB history I'll base my conclusions on that.

Well, I clearly think that the best thing to do is to discard them and do your own analysis.

Of course, Dan. I don't take any analytical system at face value. If I did with WS, I would have only 19th century pitchers on my ballot. :-)

I guess I just strongly disagree that WARP's problems are less "manageable" than WS's--I think they're much *more* manageable.

...and I have no problem with that, especially since I like DERA and incorporate it into my system, so I feel BP's stats do have merit. I just don't think we are at the point that any one side can claim "victory."

That's exactly my point--that you can extract the useful components from WARP (FRAA in my case, plus I used their FRAR to calibrate my LF vs. RF adjustment for pre-1918 players, DERA in your case), while with WS you just have to eat the whole enchilada.

while with WS you just have to eat the whole enchilada.

But I have the breakdowns for batting, defense and pitching for each player, so it's not a whole enchilada. Plus, I use WS/162 equally with the career/season totals. I can also correct for the problem it has with horrible teams, as well as adjust WS totals for players on teams with high Pythags if need be. Besides, I trust James' logic when it comes to fielding more than I do with WARP.

At least you HAVE James' logic when it comes to fielding, which is more than WARP can say.

A few thoughts on this and that:

1)Does anyone know why WARP thinks so poorly of Edd Roush's defense? I'd like to have a reason to reject data.

2)Did folks see David Gassko's article on measuring league quality? Interesting stuff, and intuitively his results seem better than BPro's to me.

3)As far as the differences between WARP1 and WARP3 for recent players, it gets much worse. Look at Ripken and Trammell sometime. It's crazy.

4)The Ben Taylor/Keith Hernandez comparison reminded me of one of my own, which I'll try to flesh out: If you're voting for Nellie Fox, why not vote for Bill Monroe? They both have an excellent defensive reputation. Longevity? Fox was an MLB regular for 15 1/2 seasons, Monroe was definitely playing for the top black teams of his era from 1899 to 1914, and may have had a few years before that. Offense? Fox had a couple of good years, but was for the most part an average hitter. We don't really have much stats for Monroe, but they certainly don't indicate he was a bad hitter. Not as good as Grant, Johnson and Hill, probably, but not bad. I don't see any reason to think he's worse than Fox as a hitter, and he may indeed have been better. Baserunning I'll give to Nellie.

The major difference is that we have all the numbers and can quantify Fox's contributions much more clearly. There is uncertainty around Monroe, but I really don't see any evidence that tells me Fox was better.

5)Dan, when I made that crack about you and Cincinnati shortstops in the Templeton thread, I hadn't looked at your rankings yet. Wow! You probably ought to run Tommy Corcoran and Billy Myers through the system just to check.

.... and back among the flat-earthers, I'm intrigued to notice something as I start to re-sort the 1B-OFs for the final decade of voting, alongside Wynn and KHernandez - two sets of players, the 8000 PAs and the 10,000+s.

GROUP A
OrlaCepeda..133 OPS+ in 8695 PA
KHernandez..129 OPS+ in 8553 PA
Edd Roush...126 OPS+ in 8156 PA
BobbyBonds..130 OPS+ in 8090 PA
ReggiSmith..137 OPS+ in 8050 PA
Bob Johnson.138 OPS+ in 8047 PA
JimmyWynn...128 OPS+ in 8010 PA
FredLynn....130 OPS+ in 7923 PA
NormCash....139 OPS+ in 7910 PA

Cepeda's "edge" comes from his 700 or so PA as DH - otherwise he's right at the magic 8000 mark, too, with nearly identical numbers to RSmith and Johnson.
Roush should get a 154-G adjustment, Smith and Wynn a lotta CF, Lynn even more, Johnson high OPS+ but a little war demerit, Cash has fewest PA but highest OPS+ - no wonder it's so hard to choose from among them!
And I'll be looking further, but I don't notice right off that there's a massive peak bonus for any of these guys - generally 1 or 2 seasons in the 160s, then some 140s, and the dropoff.

GROUP B
RustyStaub.....124 OPS+ in 11229 PA
Tony Perez.....122 OPS+ in 10861 PA
DwigEvans......127 OPS+ in 10569 PA
JakeBeckley....125 OPS+ in 10470 PA
GVanHaltren....121 OPS+ in 8979 PA
Ah, the 120s.
Staub has the most PA, but Beckley is remarkable for his era. With schedule adjustments, Van Haltren really belongs more here than Group A. Evans has arguably the best prime to go with a good peak-career combination, plus his Gold Gloves. Beckley's value as an early 1B has been oft-debated - both the position in general back then, and his D in particular.

Now just throw Jones, Browning, Duffy, Cravath, and Oms into the mix (and FHoward, 142 OPS+ in 7353 PA?), stir, and....

Whew.

Joe Dimino
the AL artificially adds offense to the game. WS does not account for this. It sounds dumb, but the wins don't add up. Since the AL added offense via the DH, but wins didn't increase accordingly, you have to add them to the offensive WS yourself. If you don't you are underrating AL players from that era, severely (I think 11%) underrating them.

artificially adds offense? 11% = 1/9 Isn't that estimate based on the point, unique to the Win Shares system, that a practically fixed number of batting win shares per game is divided among nine regular batters rather than eight?


DanR #244
As a result, BP would say that our 100-runs created guy is 51.3 runs above replacement in the NL, which is worth 51.3/9.46 = 5.4 wins, and just 47.8 runs above replacement in the AL, which is worth 47.8/9.89 = 4.8 wins, a gap of 0.6 wins, the same as the figure in my post #238.

However, KJOK is right to point out that runs are worth fewer wins in the AL than in the NL, which I don't think I was accounting for.

Why do you divide BRAR(?) by 9.46 for NL and 9.89 for AL?

Devin McCullen hates talk of sports--since my data starts in 1893, I don't have the 1890-92 seasons for Tommy Corcoran. But from 1893-1907, I have Corcoran at $21 million. BP WARP and Win Shares both have Corcoran with 21% of his value from 1890-92, which suggests that my WARP would have him at 21/.79 = $26.6 million. The HoM in/out line to date is around $87 million, so my system has Corcoran at only 30.6% of the value of a HoM player. Billy Myers only played 7 years, none of them outstanding. I have him at $16 million.

Paul Wendt--those aren't BP BRAR, where 1 run = 1/9 of a win, regardless of league context. Those are real, actual runs scored. 50 runs in a 4.50 R/G league (like the AL) "buys" fewer wins than 50 runs in a 4.24 run league (like the NL).

Ack, it's late so I'll be quick. Bill James talked about the whole DH thing in Win Shares, and had 2 main points so I'll try to summarize: 1)He thinks the way to adjust for the effects of the DH on Offensive Win Shares are in the ranking system you use them with, not Win Shares itself, which is what we're doing. 2)He references the Roy White/Jim Rice comparison from the NHBA, and says that you go through the numbers and White comes out ahead people say "Wait, you have to adjust for the DH!", but when you look at Jim Rice's career, you have to say the DH was helping him, because the role suited his skills, and he got more playing time than he would have otherwise. In that case (though not in everybody's), he certainly has a point.

Dizzypaco, 231: Bob Johnson, on the other hand was never one of the best players in the league. His career was relatively short...... he played in an era where there were a bunch of guys just like him, and several more who were clearly better.
So if you do support Johnson, or someone like him, I think you need to acknowledge one of the following: Either that there should be far more players in the HOM from the 1930's than any other era, or that there are several players (not just one or two) that played at the same time as Johnson, are less deserving that Johnson, and should not have gone in. And then name them.
--
reasonable argument. And I have Indian Bob on my ballot so it's time for a bit of soul-searching.

I have 4 HoMers who played in the 30's ranked lower than Johnson: Mackey, Medwick, Averill, and Trouppe. That is only among non-pitchers.

One could say that Trouppe doen't "count" as his prime was in the 40s; but a counter to this is that 6 of Johnson's 13 years - almost half - were ALSO in the 40s.

Johnson's career was 'short'? Short?? 13 prime years, plus at least one minor league year that he by all rights should have been in the majors. If Tony Perez hadb't been brought to the majors until he was ready to play well, and if he had quit when he was beginning to NOT play well, his career would be the same length as Johnson's. Lou Gehrig's career was 14 full seasons - same as Johnson's.

Yes, his peak was not superb. His 13-year prime (pick 12 or 14 if you wish....), however, is the best of anyone on the ballot. Unless someone would like to show that the above claim is incorrect, that is a persuasive argument to me.

But having said all of that... I DO need to compare Johnsons' stading among his peers, so dizzy's point in general is a good one. Altho given Johnson's ballot finish last election (13th), I doubt he's going in to the HoM next week.

re post 259:
of course add Parker to Group B

GROUP B
RustyStaub.....124 OPS+ in 11229 PA
Tony Perez.....122 OPS+ in 10861 PA
DwigEvans......127 OPS+ in 10569 PA
JakeBeckley....125 OPS+ in 10470 PA
DavParker......121 OPS+ in 10184 PA
GVanHaltren....121 OPS+ in 8979 PA

I was rereading parts of The Politics of Glory last night. One thing I noticed, now that I have 4 years of the HoM under my belt, is that I'm not as upset with the Vets Committee choices from the Forties. I was also amused to read James' describing Hughie Jenning's lack of dominance during the 1890s.

Johnson's career was 'short'? Short?? 13 prime years, plus at least one minor league year that he by all rights should have been in the majors.

I sort of regret calling Johnson's career short for two reasons - first, I'm generally a supporter of players who have 10 to 12 very good years (Sandberg, Palmer, Puckett), even if they didn't have a long tail to their careers. And second, it obscures my main point (which Tom understood fine) - that Johnson's statistics are impressive out of the context of his era, but when Johnson is compared to his peers, there are a lot of players who were as good or better. And the same is true of many other players not yet elected who played from the mid 20's to mid 40's.

Yeah, I'd much rather elect a guy whose career is short than one whose peak is low (just comparing the two axes on the career chart).

> I need to think this through some more, but I may adjust my WARP to use 0.22 wins instead of
> 0.6 for the DH adjustment.

Would I prorate the 5.4 BRAA/BRAR per 675 PA for post 1973 AL players that you gave me previously to 2/675PA?

As a Bob Johnson supporter I'd throw out Averill, Willard Brown and Medwick from his era. I'd also rank Johnson ahead of Keller, Kiner, Bell, Slaughter and Doby among contemporaries.

OWP, min 3000 PA, by decades. Top, 10th, 15th, 20th, 25th in MLB listed, with more listed in times of expansion and longer schedules.

I used xxx3-xx12 to match Johnson's non-WWII prime to others. I skipped over the WWII era.

1903-1912
1 Ty Cobb 0.814
10 Fred Clarke 0.669
15 Roy Thomas 0.64
20 Jimmy Sheckard 0.608
25 Mike Mitchell 0.601

1913-1922
1 Babe Ruth 0.861
10 Ross Youngs 0.678
15 Sherry Magee 0.638
20 George Burns 0.615
25 Jake Daubert 0.598

1923-1932
1 Babe Ruth 0.859
10 Hack Wilson 0.698
15 Goose Goslin 0.666
20 Kiki Cuyler 0.642
25 Gabby Hartnett 0.609

1933-1942
1 Lou Gehrig 0.796
10 Joe Medwick 0.661
15 Paul Waner 0.644
20 Gabby Hartnett 0.621
25 Pepper Martin 0.601

1948-1957
1 Ted Williams 0.837
10 Al Rosen 0.668
15 Gene Woodling 0.624
20 Gil Hodges 0.611
25 Hank Bauer 0.601

1958-1967
1 Mickey Mantle 0.796
10 Orlando Cepeda 0.671
15 Billy Williams 0.632
20 Bob Allison 0.619
25 Joe Cunningham 0.609
30 Leon Wagner 0.599

1968-1977
1 Joe Morgan 0.74
10 Pete Rose 0.678
15 Reggie Smith 0.658
20 Rico Carty 0.645
25 Cesar Cedeno 0.64
30 Tony Perez 0.634

1978-1987
1 Wade Boggs 0.737
10 Keith Hernandez 0.67
15 Kirk Gibson 0.645
20 Dale Murphy 0.625
25 Jason Thompson 0.614
30 Cecil Cooper 0.608
35 Steve Kemp 0.602

1988-1997
1 Frank Thomas 0.805
10 Fred McGriff 0.674
15 Tony Gwynn 0.659
20 Paul Molitor 0.651
25 Danny Tartabull 0.641
30 Darryl Strawberry 0.629
35 Kent Hrbek 0.613
40 Paul O'Neill 0.607

avg of 10th+15th+20th+25th:
1903-12 .630
1913-22 .632
1923-32 .654
1933-42 .632
1948-57 .626
1958-67 .633
1968-77 .655
1978-87 .639
1988-97 .656

We probably should use data thta reflects proportionately more teams in later decades, but it doesn't move the avg that much; about .010 to .015.

For those who use OWP (on which WS is based), I can't see an "easier to dominate" argument for Johnson's time period. If anything, it is true of the 20s thru early 30s.

Still, there IS an obvious argument to not elect many more late 1920s they mid-late 1930s stars.

<i>1968-1977
1 Joe Morgan 0.74
10 Pete Rose 0.678
15 Reggie Smith 0.658
20 Rico Carty 0.645
25 Cesar Cedeno 0.64
30 Tony Perez 0.634<i>

With 20% expansion in 1969, you should extend the list to 35 for this decade, same as next one. Only 8% expansion in 1977. (Tony Perez OWP is also out of line with the tail ends of other decade leaders.)

Hey, everyone, whether permitting I'll be at The Coop this weekend. Anyone want me to convey a message to Dale Petrosky for them? Actually I'm going mostly because want to see the new NgLers plaques. If any of you happens to be there, look for me in the gallery...

...or at the Oomegang brewery. Mmmmmm, Belgium.

David Gassko's article in THT on league quality with age adjustment published today in THT is, I think, quite pertinent to your decision-making. It suggests a relatively steep incline in league quality between 1871 and 1931 (and particularly prior to 1900) and a relatively gentle incline in league quality between 1931 and the present.

In other words, the classic S curve with the exponential growth gone by 1930.

I'd like the answer to this question: How well does Gassko's league-quality adjustment matchup for the Dan R's SD work?

Gassko's work looks interesting. I'm a bit surprised that data shows that quality immediately post-WWII shows a definite uptick, but that most of the 50s were lower than the late 40s.

I also think the results for the past 20 years would be different if he used pitchers instead of hitters, given that we've added almost 20% personnel to pitching staffs (with less workload on the big starters), and are giving fewer at-bats to bench hitters.

Can I vote for Felix next year? :-)

Gassko only gives the graphs, not the raw data. The scale is virtually identical: from a little over 0.8 in the 1890s to 1 today. But his graph is so compressed it's hard to see the ups and downs after then besides the war dip. My system does not adjust for the war, because while the NL stdev went up, the AL stdev went *down*, since Williams, DiMaggio etc. left to fight while a bunch of mediocrities of similar talent level remained. My stdev data also shows a *massive* spike from the 1900 to the 1901 NL, with the 1900 NL having the same stdev as the 2002-03 NL while the 1901 NL had the highest stdev (both actual and regression-projected) on record. Just eyeballing Gassko's chart, I don't see such a pronounced spike. (1901 is like WWII--the stars stayed in the NL while the AL was full of homogenous lesser lights apart from Lajoie, so the NL was a tougher league with a higher stdev. The teens are the reverse, where the stars were in the AL so it had a higher stdev). If I could get my hands on Gassko's raw data I could compare his league quality and my stdev adjustments directly, which would surely be enlightening.

last minute plea

Are we really about to elect Edd Roush?

Bobby Bonds, Reggie Smith, Bob Johnson - all decent OFers who hit as well or better than Roush. Similar career lengths.

League quality? Roush's was the wekaer of the two MLB leagues at the time, and that doesn't include the black players who weren't allowed to play. Some would argue that Indian Bob, the best hitter of the group, played in a weaker league due to ease of dominance reflected in how many of the 1930s guys we've elected. He should get some Minor lg credit tho.

Defense? Some systems have Roush ahead, others rate them all about the same.

Speed? Roush was fast (not liek Bonds tho), for in the years we DO have CS data, he was
successful at stealing only 54% of the time. I'm not impressed.

Durability? Roush doesn't fair well there, either.

BP career tranlsated stats:

.......... PA .OBA .SLG .SB .CS
Smith 8246 .367 .548 130 .70
Bonds 8005 .361 .536 447 147
Johns 8173 .373 .553 110 .58
Roush 8435 .358 .493 276 133

Roush was simply not the run producer the others were. And we've elected lots of OFers already. The other three guys are not near being elected. Not sure why Roush is.

1997 ballot from this highly career voter (with a fairly low replacement level):

1. Jake Beckley - luv the career, though peakless
2. George Van Haltren - deserving star of the underrepresented 1890s
3. Graig Nettles - super fielder; I am surprised by his lack of support
4. Bob Johnson - solid hitter, solid career (w/minor lg credit)
5. Bobby Bonds - good combo of peak and career (where's the luv?)
6. Nellie Fox - very good second baseman for a long time
7. Tony Perez - good career though he was barely an adequate 3B defensively
8. Dwight Evans - good career and good defensive right fielder
9. Rusty Staub - good peak + good career (similar to Perez)
10. Tommy Bridges - luv the strikeouts & win pct with minor league and wwii credit
11. Bob Elliott - mired with woeful Pirates and Braves
12. Edd Roush - underrated, very good centerfielder
13. Charley Jones - great player, with lockout credit
14. Reggie Smith - boost from center field play and japan
15. Chuck Klein - great peak even park-adjusted
----------
16-20 RMaranville, HWilson, LAparicio, PTraynor, BClarkson

Not voting for Browning (around 100th), Fingers (around 50th), and Redding (around 50th).

I am not voting for Roush but I do have him above those three, here are some reasons why.

1. He played more CF than the other three. Even if his overall defense was about equal, the offensive level in CF is going to be lower, giving him more value.

2. He missed time at his peak due to WWI

3. Bob Johnson's best years came during WWII and definitely need adjusting

4. Career rate stats never really impressed me

You are right about the in season durability, however. Roush would have been in my PHOM long ago if it were not for the holdouts, nagging injuries, etc. that cost him about 15-20 games a season during his prime.

Rob Wood: this is the discussion thread. Did you mean to put that on the ballot thread? If so, you should re-post it over there.

If you're looking for comments: OK, you look kind of career-ish and you've got a lot of OF on your ballot. In other word, no odder than anyone else's ballot in this very odd year.

TomH: My personal order among the four is Smith > Bonds > Johnson > Roush. Smith makes my ballot this year for the first time, at #15. He's not likely to last on my ballot.

Ken Singleton appears considerably ahead of Dwight Evans on at least one ballot. They are such comparable offensive players from the same era, that I would have thought that Evans' defensive and baserunning advantage and career length (particularly his additional time in right-field) would make this comparison straightforward. They seem to be about equal at their peaks. Am I missing something?

Roush was simply not the run producer the others were. And we've elected lots of OFers already. The other three guys are not near being elected. Not sure why Roush is.

Neutralized Stats per Baseball Ref:


The non-durable Roush has more adjusted games than any of the other 3, which surprised me a little.

Name - Games Played LF/CF/RF (no adjustment)
Smith - 3/808/874
Bonds - 65/285/1472
Johnson - 1592/162/24
Roush - 81/1754/13

Really Roush is a centerfielder, Reggie Smith is kinda sorta, the others are corner outfielders. Bonds had exceptionally poor taste in teammates if he wanted to play centerfield before he turned 30, first Mays then Gary Maddox. However at age 30 the yankees played ELLIOT Maddox in centerfield and Ron Bladt (Dimino, Repoz, who the heck is that?)out there more often than Barry's dad.

Ken Singleton appears considerably ahead of Dwight Evans on at least one ballot.... They seem to be about equal at their peaks. Am I missing something?

I wouldn't say their peaks are equal--by WS, Singleton is a clear notch better, to my eye, even after adjusting up Evans's great strike season. (Of course, I don't have them THAT far apart, with Singleton at #14, Evans at #21, so perhaps you're not talking to me...)

Actually, it was Rick Bladt. I remember.

The 1975 Yanks played at Shea Stadium, due to the massive renovation at Yankees Stadium.
Elliott Maddox blew out his knee slipping on the shoddy Shea turf (made shoddier by having two teams play there at once) after he had gotten off to a very good start that year. One of the panic moves the Yanks made was trying the immortal Bladt.
Funny thing is, in 1977 the Yankees trade Maddox AND Bladt to the Orioles for a washed-up Paul Blair, who managed a 90 OPS+ that year before hitting a laughable .176 in 125 1978 ABs.
The Yankees finally sobered up after a 1 for 5 start to the 1979 season and released Blair, who then hit .150 with a 21 OPS+ (!) in 140 ABs with the Reds.

In Blair's defense, I'm not sure anyone ever looked more graceful patrolling the OF grass.

Also, Bonds played most of 1975 with pulled leg muscles, so center field wasn't really an option a lot of times.

Hey, in response to #278 somebody has to say:

Are we really about to elect Dwight Evans? Couldn't say that over on the ballot thread, but it has to be said.

Are we really about to elect Edd Roush?

Bobby Bonds, Reggie Smith, Bob Johnson - all decent OFers who hit as well or better than Roush. Similar career lengths.

Defense? Some systems have Roush ahead, others rate them all about the same.

Assessments of Edd Roush as a fielder do not differ so much as those of Earl Averill, but their range is wide. It appears that Roush is "below the line" for most voters on his batting and playing time record but, given that, his fielding is decisive in his favor (electorally decisive, maybe not a majority view).

In a way or two, this makes Roush the anti-Pete Browning. I believe Browning who will remain outside because the harsher assessments of his fielding are credible enough. Of course, revisionists who call Roush or Browing an average fielder are revising from opposite ends of the spectrum.


washed-up Paul Blair, who managed a 90 OPS+ that year before hitting a laughable .176 in 125 1978 ABs.
The Yankees finally sobered up after a 1 for 5 start to the 1979 season and released Blair, who then hit .150 with a 21 OPS+ (!) in 140 ABs with the Reds.

In Blair's defense, I'm not sure anyone ever looked more graceful patrolling the OF grass.

I'm sure the late paychecks were welcome. He didn't get rich in Baltimore. Bill James rates Blair 183 and 35% fielding; Curt Flood 221 and 34% fielding (Garry Maddox 28%, Bill Bruton 26%, Bill Virdon 34%). 35% is routine for shortstops and catchers, achieved by many 2Bmen.

Jimmy Piersall 37%, Marquis Grissom 34%, Brian McRae 32%, Cesar Geronimo 30%, Tommy Agee 29%, Willie Wilson 29%, Devon White 28%, Lloyd Moseby 28%; Rick Manning unlisted

--
skimming 1950s in Win Shares something caught my eye and investigation yields this for the six decades of 16-team mlb.

Number of Win Shares, ranks 10 and 20 and 50
1900s 230 202 157
1910s 208 183 140
1920s 201 180 144
1930s 243 196 142
1940s 203 174 125
1950s 234 182 137
What caught my eye was the apparent domination of the 1950s by its top players. The drop between ranks 10 and 20 is greatest in the decade but maybe insignificant.

It's only 5 more than the 30s, which was higher overall, so I don't see a big difference.

I've asked this before, but does anyone have an idea why some people (WARP, at least) have a lower opinion of Roush's defense? (Anti-Edds, my ballot isn't done yet and right now he'd be near the bottom - this is your chance!)

Because he played right field, I can't resist adding the old rules base thief, Hugh Nicol 37% (OPS+ 78). As the only Scotland-born mlballplayer, he is sponsored at baseball-reference he is sponsored by Cheap Scottish Hotels!

The left-handed thirdbaseman Hick Carpenter 34%

More pertinent here, George Van Haltren 344 win shares: 74% 14% 12%. That is just over 40 for pitching.

Paging through "Win Shares by Decade" (p484-538, thru 1990s), here are those with at least 30% of win shares attributed to fielding whom I noticed and recognized as outfielders --a fan of Paul Blair and Jimmy McAleer
47 McAleer Jimmy
37 Nicol Hugh (rf)
37 Piersall Jimmy
36 Kreevich Mike
35 Blair Paul
34 Virdon Bill
34 Flood Curt
34 Grissom Marquis
33 Myers Hi
33 Cramer Doc
32 Douthit Taylor
32 McRae Brian
30 Welch Curt
30 Moore Terry
30 Geronimo Cesar

Five of them were regulars on multiple-championship teams of one club, Welch and Nicol in the 1880s, Douthit in the '20s, Moore in the '40s, Flood in the '60s. Alas Willie McGee 1980s is nowhere to be seen and Jim Edmonds 20-aughts is much too good a batter.

Totally out of left field, but were MLE's ever done for Sadaharu Oh? Is he eligible for the HoM? If Japan is between AAA and the majors, 868 dongs sure sounds Meritorious to me.

We discussed his eligibility a few times, and the consensus voted in favor of being fair only to North American players.

We discussed his eligibility a few times, and the consensus voted in favor of being fair only to North American players.

Well, at least to any player who was on a MLB or NeL roster at some point in his career.

We have talked about setting up another wing that would handle guys like Oh for the future.

As for Oh's MLEs, I believe they were presented in one of SABR's annual publications a few years back.

I'd love to see them; is there a link anywhere?

I'd love to see them; is there a link anywhere?

I don't know, Dan.

IIRC, the MLE compared Oh to McCovey.

The Oh MLE's are here:

The Case for Sadahara Oh and the Hall of Fame

(moving Fox discussion from ballot thread to here)

1. Rizzuto averaged 22, not 25 BRAR per year. FEWER if you only include surrounding seasons. You can get to 23 if you include his beautiful 1950 but not the declining later years.
2. Bancroft played when blacks weren't allowed. If their BRAR are similar, Fox ought to rate a pretty good edge.
3. Yes, BRAR does add in basestealing, so Concepcion's value in that regard is accounted for.

> Bancroft played when blacks weren't allowed. If their BRAR are similar, Fox ought to rate a pretty
> good edge.

Except Bancroft played SS and Fox played 2B. The RCAP (not the same as RCAR) is going to favor Bancroft. Pardon me for stating this but in the 1950's blacks were allowed but they weren't exactly encouraged. I'm still trying to figure out if Artie Wilson was as good as Nellie Fox. I think Bancroft is the top of this heap. His glove was a notch ahead of anyone else in the game at the time (possible Dick Lundy could match him) and his bat is just slightly ahead of the rest. What I'm really struggling with is how Fox could be so FAR ahead of these comparable players. I would expect splintered support among the backlog. I think Fox is jumping ahead of the others because he played 2B. If he had put up those numbers as a SS he would probably do worse here which doesn't make a whole lot of sense.