It is not always or necessarily true that our elections are decided by 1st place votes, but...this election should be close enough that the "bonus" points for a 1st place vote could be decisive, though the bonus points don't quite compensate for a 6-point penalty for being left off of a ballot. It will be interesting to see how the 1sts end up laying out. Following is the number of 1st and 2nd place votes each carry-over candidate received in '09.

Start 11

My Prelim for 1910 has no newcomers. In fact, no newcomers crack my top 20. Maybe Top 30...

1. Pud Galvin - The only pitcher available still deserving of a spot. I won't be disappointed if he doesn't get in this year, but he deserves to, imo.

2. Joe Start - The likely winner this year.

3. Bid McPhee - Consistantly strong for me, but there are others more deserving. Will get in eventually.

4. Cal McVey - See McPhee

5. Jimmy Ryan

As for 1910 I have no real basis at this point for moving anybody around though I'm sure I will if only for the fun of it.

Must-Be HoMers

Is Start really the guy to beat, per John Murphy? A 9-point margin is not enough to establish a fave, I don't think. Make Joe and Pud co-faves.

I did indicate that I wouldn't make a bet on who wins, so I acknowledge it can go either way. However, my gut says Start.

Here's the bio of new candidate Jack Doyle from the SABR BioProject. It won't get him any votes, but it gives some of the flavor of MLB a century ago. It also covers Dirty Jack's acrimonious relationship with John McGraw.

Some thoughts on the top three.

We've had a lot of talk recently about Joe Start, so rather than say all that again, I'm going to lay out the reasons why I'm planning to rank Galvin #1 in 1910, with Start at #2, and McPhee somewhere between 4 and 8.

Reasons to take Galvin over Start:

1) His longevity is more impressive, and his career value higher. Galvin's 18-year career is the longest of any nineteenth-century pitcher, and his IP (with estimates for 76-78) are around 7500, topping the next nearest pitcher to him by 50% (that's Tim Keefe at 5056). Start played for 27 years, but is he the only first baseman of the era to play 27 years? No.

2) He has a statistically documented peak that places him as the best, or one of the best pitchers in baseball, 78-84, at a point in time during which pitchers were particularly important. Start may have been one of the best players in baseball, 66-70.

3) He has a good argument, from both peak and career, to be the best pitcher of the nineteenth century, even though others were inducted first; Start, good as he is, seems to me to be the fourth-best first baseman, after Anson, Brouthers, and Connor. It's a deep position, sure, but so are 1880s pitchers.

To put Start over Galvin, it seems to me that you must either credit Start with a great peak in the late 1860s or give him substantial career value credit for play in the early 1860s. There are reasons to do both, but I think the reasons not to do so are stronger. If Start goes in, that's fine with me -- he should go in soon. But Galvin has the stronger case.

Why McPhee is below the others. Career value is his main claim, and he has less of it than Start or Galvin, unless you believe WARPs fielding numbers for the nineteenth century are an accurate measure of value; I think they are a substantial overestimate. If McPhee goees in, that's fine with me -- he should go in at some point. But there are several other players with stronger cases.

Duke Farrell was ranked #45 all-time by Bill James?!

He's a very good fielding catcher who has over 6,200 AB's, unlike Bennett & Clements.

I assume he's going to take a hit because his best batting season was 1891 AA? Looks like me MIGHT be the best catcher of the 1890's anyway?

Duke Farrell was ranked #45 all-time by Bill James?!

He's a very good fielding catcher who has over 6,200 AB's, unlike Bennett & Clements.

I assume he's going to take a hit because his best batting season was 1891 AA? Looks like me MIGHT be the best catcher of the 1890's anyway?

_Might be_. Clements was definitely a better hitter, but wasn't as durable -- though Farrell was not much more durable _as a catcher_, but he split time at third base and in the outfield, which increased his playing time.

Zimmer was definitely a more durable catcher, but he wasn't quite the hitter Farrell (though he was close) was and couldn't play other positions.

I suspect that Deacon McGuire was the best catcher of the 1890s -- durable, a pretty good hitter at his peak, and able to play other positions -- but he won't be eligible for a few years yet.

I don't think any of them were as good as Charlie Bennett.

Prelim Ballot - Only one major change (reconsideration of Tony Mullane)

1. CHARLIE BENNETT, C -Comp is Roy Campanella. Until at least Roger Bresnahan, only Ewing was a better Catcher. Catchers may have trouble "adding up" numbers due to the nature of the position, but last I checked you can't play the field without a catcher.

I suspect that Deacon McGuire was the best catcher of the 1890s -- durable, a pretty good hitter at his peak, and able to play other positions -- but he won't be eligible for a few years yet.

I like Farrell a little better, but an argument can be made in McGuire's favor. You might be right.

Prelim:

1) Bid McPhee (1)

2) Cal McVey (3)

3) Dickey Pearce (4)

4) Cupid Childs (5)

5) Joe Start (6)

6) Pud Galvin (7)

7) Harry Wright (8)

8) Charlie Bennett (9)

9)Duke Farrell (n/a): Only newbie this "year" on my ballot. May deserve to move up, but I'll keep him here for now. Many fine catchers during the nineties, but no true standouts such as Deacon White, Buck Ewing or the abridged career of Cal McVey. Best major league catcher for 1893. Best player who played a good deal of third base in the AA for 1891 (hard one to rate).

10) Billy Nash (10)

11) Chief Zimmer (11)

12) John McGraw (12)

13) Frank Grant (14)

14) Jimmy Ryan (15)

15) Lip Pike (n/a):



I'm knocking Clement off this ballot.

Thompson has more WARP3.

Because, IMO, they overrate Thompson's defense, while they give Ryan a big demerit on defense (where I think he was a little bit better than Big Sam).

Thompson has better "dependent" stats.

Overrated again, though Thompson in context was still a great RBI man.

Thompson has a much higher peak.

Arguable. The problems with WARP's defensive idiosyncracies arise again when you compare peaks.

Overall, I'll take Ryan over Thompson (not that it's a huge difference).

I haven?t done a controversial post for a while. Maybe this will be one. My apologies to all FOFG and FODP.

In any long-term project, it helps to step back from time-to-time and think about what you?re doing. Am I really going about this in the best way? Am I getting the best results possible? Questions like that.

Last year, in the long debate over the HoM rules, I expressed the idea that we wanted a structure that ensured we elect the right players. The ?right players? being those who were provably great, who have a defensible case as a great player. The case of Frank Grant has led me to a reassessment of my approach to this project, as I?m faced with a contradiction between this philosophy and my voting.

Thinking about Grant now, the case for him rests on three pillars: 1) His great performance, in a few minor league seasons in his younger days; 2) Sketchy testimonials during the rest of his career; 3) His reputation as the preeminent Black player to complete his career before the mid-teens.

As Lawyer for the HoMers, I asked myself, ?Would this evidence hold up in court? Can we be reasonably certain from this that Grant deserves a place with the ?immortals?? In my judgment, the answer is ?No?.

To briefly address each point of evidence: 1) This performance is good hard evidence, but similar feats have been performed by a great many non-HoMers. 2) These anecdotes seem to come from a limited number of sources and/or contain an obvious inherent bias. 3) The best in a subjective category is hardly evidence to point to for justifying election to the HoM. In short, the degree of assumptions and guesswork involved to reach the conclusion that Grant is a HoMer are not supported by the known facts. Grant will drop off my ballot.

Then, as an old Friend Of Dickey Pearce, I had to ask, ?How is this conclusion so different from my reasoning behind Pearce?s support?? The answer: it?s not. My comment in voting for Pearce says, ?I don?t really have a compelling argument for him.? I was forced to admit that this runs contrary to my original ideal of HoMers having a defensible case.

Do I think Pearce is a HoMer? Possibly, but the case relies a lot on imagination. The indications are he was a pioneer, an innovator, a superior fielder. Did he ever have the dominance we look for in a HoMer? Maybe, but since I don?t really think it?s defensible to say he did; he will drop off my ballot. The evidence is just not strong enough.

Which leads us, unavoidably, to Joe Start. Prior to 2002, I had never thought of Start as a guy who deserved any consideration for the HOF. I had never seen anyone tout him as such, and his stats kept him below the radar. Now I have him #1 on my ballot. Is this defensible?

In Start?s case I think it is. The body of facts concerning him is such that I think we can extrapolate from the known and reasonably conclude he was past his prime in 1871. We can reasonably conclude that he was one of the game?s top five players for most of the 1860?s.

I may change my opinion again on the case for Frank Grant. But I don?t want to base it on any affirmative action approach like, ?We need a certain quota of black players from each era so why not Grant?? Our mandate here indicates we should ultimately have at least 20 Negro league players in the HoM. Grant doesn?t make anyone?s top 20 list that I?m familiar with. Is it not possible the conditions of the game made it impossible for any Black to excel to HoMer status in the 19th century? Or maybe we should simply say we know Grant was outstanding, but it can?t be shown he was THE greatest Black player of the 19th century.

To sum it up, if we?re just looking to fill the slot of ?19th century Black HoMer?, Grant seems like the guy. IMO, inductions that are based on filling categories are not defensible. To be defensible, elections must be primarily merit based. Certainly, reasonable assumptions can contribute to a guy?s case but they should not be the primary force supporting an election. The arguments for Grant thus far contain an ounce of evidence and a pound of conjecture.

Mark M,

I agree. I have Ryan 3rd and there is no way he is above Thompson

Prelim ballot is as follows....

1. Charlie Bennett (2) -

I don't really have anything to say about this year's election that I haven't already said, but I have a couple of questions I want to start thinking about for future years.

1. Clearly, there needs to be some discounting of performances in the 1901-1902 AL. But how much of a discount? And how long does it take for the leagues to come to near equality?

2. We've been wallowing in outfielders, and will be for a while. (We still have Burkett, Kelley, and Keeler to go from the generation we've been looking at.) But we're going to start having quality infielders soon. Who ranks higher, Lave Cross or Jimmy Collins? And how do we compare third basemen to shortstops (e.g., Bill Dahlen)?

>1. Clearly, there needs to be some discounting of performances in the 1901-1902 AL. But how much of a discount? And how long does it take for the leagues to come to near equality?

The AL of course won the very first World Series. This of course was '03, not '01-'02. But the AA only won 1 of 7 Temple Cups, while the AL won 3 of 7 and then 3 in a row (6 of 10). If it's "near-equality" you want, how about "right away," or certainly by '03.

Secondly, speaking of wallowing in an OF glut--this is just for fun, but you could argue that a "glut" by definition means a bunch of guys who cannot be differentiated, which in turn suggests that maybe none of them is all that special. We've had complaints about ballots without pitchers. How about a ballot without OFers! Just for fun, again.

1. Start

Mark, here's how my adjWS calc. worked out for Thompson's and Ryan's peaks--adjusted to 162 games and with a fielding bonus corresponding to the pre-'93 pitching discount.

Top 3 yrs--Ryan 118 Thompson 105

Clint:

Someone was posting a few of Bill James' rankings from the NHBA. I put together a list of all eligible players who rank in the top 50 at their position.

18 Jennings-SS

Who ranks higher, Lave Cross or Jimmy Collins?

I'm going with Collins. Much higher peak than Lave.

100% agree with posts #28 and #29. I've got Collins top three. Cross way off the ballot.

The HoMers

100% agree with posts #28 and #29. I've got Collins top three. Cross way off the ballot.

I have Cross as the best third baseman who was truly a nineties player (I don't consider Collins really from that decade). Like the catchers, there are a few third basemen that can make my ballot, but they don't really compare to Sutton.

Do I think Pearce is a HoMer? Possibly, but the case relies a lot on imagination. The indications are he was a pioneer, an innovator, a superior fielder. Did he ever have the dominance we look for in a HoMer?

Well, we know he was the best of his time and had an extremely long career (which makes any parallel with Frank Grant fall apart since there are questions where he ranks among his contemporaries). Yes, we don't know if there were better shortstops around during the fifties and sixties that have somehow eluded the history pages. However, we can say the same thing with the nineties guys. Is it reasonable to suggest there was a better hitter (or hitters) than Delahanty playing out in the West or in the South? Of course we can, but I didn't notice anybody slowing down the diesel named the Big Ed this election (with no complaints from me).

We should deal with what we know. Pearce was the best of his time. Should comparable documented players go in before him? I have tried to do that (same with Harry Wright). Should we be conservative with Pearce? I have been with Pearce, H. Wright, even Start. After all of that, Pearce (and the other two) deserve to represent the sixties. It's not a zero sum game. Other than delaying someone else's election, electing the pre-NA guys will not take a slot away from a deserving later player.

TomH-

RobC, what sort of adjustments do you make to Warp3?

Chris,

I dont do any actual changes to Warp3, I just, for certain players, adjust their Warp3 rankings a little, especially for players in very short seasons (NA mostly). The playing time adjustment they use is a very poor regression to a mean, but I havent bothered to calculate a better one so I kinda eye-ball it. Also, I dont use it as much for pitchers but that doesnt apply to the OF discussion.

I actually like a replacement level about Cleveland Spiders level so I dont make an adjustment for that (although I certainly understand that some find that level to be way too low).

Having calculated adjWS peaks (adj for season length, league strength and fielding bonus), I always like to trot them out here. I agree that there are hairsbreadths separating a lot of guys on our ballots including the CFers. But I think Ryan's peak stands out somewhat, and not just in terms of "raw" adjWS, if you know what I mean. the adjWS are to some extent products of their time--players could earn a lot more WS in the '80s than the '90s (bigger adjustments to 162 and no more fielding bonus after '93). So what is interesting to me is how they stand among their peers. So I listed every player who was within 10 WS of the best for 3 consecutive year peak and within 20 for 5 consecutive years.

Ryan--best 3 year peak 1890 (117), best 5 year peak 1891 (176), best 5 year peak 1982 (181)

Van "Halen"--second best 3 year peak 1890 (114)

Duffy--best 5 year peak 1893 (167), best 3 year peak 1894 (112), best 5 year peak 1894 (175), best 3 year peak 1895 (103), second best 5 year peak 1895 (166), third best 5 year peak 1896 (161)

Not much to choose from among those 3 year peaks, but Van "Halen" never put together the big 5 that the other two did. Ryan beats Duffy overall based on higher career value, too, BTW.

<i>What?s up with Van Haltren? People are saying he didn?t have such a great peak, but that?s not what this study is showing. Where have I gone wrong? I have his yearly WS as:

Yes, I adjust for league strength so Duffy's '91 year is so adjusted.

Joe, you're overreacting to the two comments on Jennings. He doesn't belong on Clint's list? Sure, he does, he did exactly what Clint meant to measure. We can all take in that info for what it is worth, it need not be suppressed.

And WS has him 18th? Well, of course, the system was jerry-rigged to his advantage. I don't think so. Your comment assumes that career-value is the be-all and end-all, and James is the outlier. That is your assumption. I find James' system every bit as valid as a career-only system (though I'm not fond of the ######## dump ;-)

Just so you know, everybody's got their own fielding adjustment. I used 1.5 up til '93, then zero.

Just so you know, everybody's got their own fielding adjustment. I used 1.5 up til '93, then zero.

I'll mention, on the subject of fielding adjustments, that now that I'm finished with win share translations for the NA, I've started trying to get a handle on how to reconcile pitching and fielding runs in WARP with WS more reliably, or at least to identify clearly where and how all the major discrepancies between the systems arise, so that we can make more informed judgments about which system is preferable for which numbers.

This is a slow process because the numbers have to be compared on a team-by-team basis, which for me, means getting complete team rosters from baseball-reference.com, searching each player on the baseballprospectus site for WARP info for that year, then finding each player in the win shares electronic supplement to get the ws data for the year, then comparing them. If anyone out there is interested in working on this, I'd be happy to share the methodology so that we can get some more data points more quickly.

I don't have any results that are significant yet, but the first study, of the 1891 Brooklyn NL team (which attracted my attention because I wanted a better sense of how credible Mike Griffin's out-of-sight 55 frar is), is suggestive.

Here are some of the things that look probable at this point, if this team turns out to be representative.

1) Win shares underrates the value of fielding by about 1/3 in 1891

2) WARP overvalues fielding by about 20% in 1891.

3) Win shares underrating is more regular than WARP's overrating, which is concentrated in the outfield. The two systems are most nearly in agreement for 2nd base, SS, and 3rd base. Win shares _may_ underestimate 1st base and catcher more than other positions. The first-base and catcher positions were unsettled on the Grooms, so the data is pretty muddy here.

4) The most serious discrepancies, and the ones hardest to correct for systematically, occur at the high and the low ends of the value spectrum. At the low end, WS curves towards its zero point, where WARP moves linearly into negative numbers. These negative numbers are balanced on the team level by spikes at the high end of values. So WARP and WS agree more closely on good but not great defensive players, and they differ most greatly on the worst and best.

This last, if borne out by other data and determined to be a problem with one system or the other, is particularly unfortunate for our purposes, because many of the players we are looking at are great fielders at the high end of fielding value.

A similar effect occurs, I think, in batting values, but as the two systems are _much_ more reliable on batting value, this single discrepancy seldom leads to highly divergent representations of players' values as hitters.

5) The idea that Mike Griffin was 55 frar for 1891, as WARP says he is, is not remotely credible. He was a great defensive player that year, but nowhere near that great.

>Bill James, Voros McCracken and many others have talked about fallacies in Hall of Fame arguments.

> Marc, as far as James system, it's about 80-85% peak value and about 15% career value, so yes, I

I think it is probably also worth emphasizing that the competition level after 1892 and before the advent of the AL was as high as we have seen thus far. Their was one league with complete hegemony that consisted of all of the base players and that league (NL) also had less teams (and less players) than there had previously been at the "major league" level. Also, the seasons were getting longer so that less adjusting is needed to bring the players to a 162 game benchmark.

This does not apply with the quite same force to either the NA and the NL prior to the AA. Those leagues had shorter seasons and (at least as I see it) were not as universally recognized as the highest league as the post-AA National League after 1892. Finally, apart from rapant dirty tricks the game of the NL after 1892 was closer in terms of rules to what we consider Major League baseball today.

Tip O?Niell played only 9 seasons but his average WS per season is 33.98. Is his short career the reason he gets little or no support from this group?

That's my problem with him. Four of five more seasons of above-average play would have made a world of difference to me.

I'll mention, on the subject of fielding adjustments, that now that I'm finished with win share translations for the NA...

I'm in the process of calculating the actual WS for the NA (rather than translations from WARP). There are some glitches, because I don't have reliable park factors, passed ball data or home/road W-L records. I'm using half of the park factors from baseballreference.com (i.e, if baseballreference says 103, I'm using 1.015 as a park factor). I believe the baseballreference park factors come from Total Baseball, but I'm not sure. The WS numbers should be close though. Maybe when I finish, we can compare them with Chris' translations from WARP.

Anyway, I've finished 1871 and I'm halfway through 1872. So far the batting WS are pretty close to the WARP hitting translation formulas. The fielding translations are more hit or miss.

This is a slow process because the numbers have to be compared on a team-by-team basis, which for me, means getting complete team rosters from baseball-reference.com, searching each player on the baseballprospectus site for WARP info for that year, then finding each player in the win shares electronic supplement to get the ws data for the year, then comparing them.

A very cumbersome process. I feel for you. Have you been using the Lahman database to at least get the team rosters?

>And what you call 'rate' is basically another peak measure, WS per 162 games, which is very biased

I have to agree with Marc. Rate is not really the same thing as peak. Now, if James had used the 3 and 5-year peak rates for each player in his rankings (as a companion to 3 and 5-year peak WS), that would be different.

I do agree with Joe that James' rankings lean a little too heavily on the peak side, however.

Tom and John say that WS leans a litte bit more (or, even, a little too heavily) toward peak. I agree.

You said it was 85/15. That is what the discussion is about.

As to Tom's example, it is a handpicked sample not unlike Clint's list showing that Hughie Jennings is among the 40 or so players who were among the top 5 5 or more times. Show me a real example of such an outrageous ranking, other than Hughie Jennings whose "true" ranking we don't agree on.

Joe repeated:

And what you call 'rate' is basically another peak measure, WS per 162 games, which is very biased against people that stick around past their prime.

But career WS is biased against people that don't stick around past their prime. The combination of the two counteract their cons nicely, IMO.

Here is a great example: Who was better in '23: Rogers Hornsby or Jigger Statz? Some would argue that they were equal because they have the same amount of WS. However, Hornsby created his 26 Win Shares in 47 less games than Statz. How can they be comparable? The answer is they can't. Hornsby's impact on the games he played was unquestionably greater than Statz. To suggest they helped their teams equally is silly.

BTW, my "great example" from above doesn't really have anything to do with my first paragraph from post #67.

Moving everyone up one notch...

1. Joe Start - Continuing to be pushed for election; best of the pre-NL/NA players.

2. Cal McVey - Again, one of the best pure hitters in baseball in the 19th century. Deserves recognition.

3. Bid McPhee - A glut of legitimate candidates at 2B is forming; McPhee is the best of the bunch.

4. Harry Stovey - undervalued because he lacked the glovework.

5. Charlie Bennett - how can we ignore the best (remaining) catcher of baseball's first half-century?

6. Jimmy Ryan - i've always thought he merited election to Cooperstown; time now to give him a "higher" honor. :-)

7. Cupid Childs - an extraordinary run producer at 2B.

8. George Van Haltren - just slightly less impressive than Ryan.

9. Mike Tiernan - the best of the rest; moved higher than Thompson this year.

10. Lip Pike - great hitter; great peak...earned his place here.

11. Frank Grant - had to put him SOMEWHERE, but thought I'd start conservatively. Negro 2B who combined the best qualities of Childs and McPhee. Would like to find out more - definatively - about him.

12. Sam Thompson - still plugging Thompson in here.

13. Pud Galvin - lower than he ought to be because of the new people on the ballot.

14. John McGraw - most people forget he was a helluva player too.

15. Hughie Jennings - peak years too high to ignore; better competition in 1890s than in earlier years. (I put Jennings here instead of Ned Williamson.)

Joe, you sound as if you're not aware that James counts up career WS (though that was John's comment about career WS that you quoted, not mine). Unlike TPR, there are no negative numbers, no penalties for playing below the average or even below replacement. Like raw numbers (hits, HR, pitching wins), WS accumulate as long as a player is active.

When a player is "hanging around" in the sense we mean it, he is NOT accumulating any peak value, of course, nor is his peak diminishing.

His rate of course diminishes during a decline. The vast majority of players experience this and so their rate numbers are comparable. Those few who do not decline (Delahanty, Joss, Joe Jackson, etc. etc.) need to be evaluated accordingly. Those who hang around a little "too long" (PeteRose?) can be evaluated accordingly. We are all smart enough to do that.

Ironically, the rate stat is the one that gives 19th century players their appropriate due, so I'm not sure why you dislike a rate stat. On the other hand, I agree that an adjWS number is an even better way to understand players pre-154 game seasons.

In short, the problems with WS are known and can be dealt with more easily than the problems with TPR or WARP, IMO, because it is a more simple and straightforward system. If you don't like the weighting of career, peak, rate and the timeline, you can use Bill James' numbers to rework the output. What a handy tool!

With all due respect, Joe, you sound like a dad whose daughter is just beginning to date. You are deathly afraid that she (we) will do something stupid. Trust us. (Hughie was #13 on the last ballot and I had him #10. He had a great peak, but I don't think he is a threat to your daughters.)

I think it is useful to distinguish Win Share figures from how they are "combined" in the New Historical Baseball Abstract to form the basis of James' rankings of the all-time greats.

Many people have critiqued the underlying win shares formulation, including me. In addition to questioning the implicitly very low replacement level, Joe and others have questioned the validity of the James-calculated win shares for 19th century players. Thus we have different weights being placed on fielding vs pitching, the different fielding positions, etc. On the whole, though, the win shares system is fairly robust and useful, even for 19th century players (the adjusted win share figures make a good thing better).

On the other hand, I lead the parade in criticising how James combined a player's seasonal win share figures into a single number in the NHBA. He did an uncharacteristically poor job and I wrote up my critique in a By The Numbers article soon after the NHBA came out. He wound up weighting peak (together with rate stats) much too highly compared to career value. One reason which he essentially admits to is that the replacement level is so low it is easy to accumulate career win shares without adding much more to one's resume (helping your teams win games and pennants).

When the NHBA came out, the Baseball Survivor group was in the midst of our deliberations. I proposed that we largely ignore James' combined number (I cannot remember what he calls it) and essentially calculate what Joe is doing here in his pennants-added calculations. I recommend that we do the same thing here.

So the case for Jennings rests squarely on the assertion that he was among the top 5 position players in the NL for each of his 5 good years, 1894-98. Probably he was, but there are two things that nag at me a little - maybe not doubts, exactly, but a little buzzing in the back of the mind as I try to think about it:

1. The "grand statistics" - WARP or WS - by which we make that assertion have in the specific case of Jennings unusually large inputs from two unusual columns: on the defensive side, PO, and on the offensive side, HBP. I don't doubt that he was a very good defender, but he was leading leagues in PO and not in A. And, yes, a HBP is a time on base, same as a walk, so I shouldn't question it.

2. His greatness is tied not just to a particular time (94-98) but to a particular team (the Orioles). I know, his sudden decline has mostly to do with his own injury history, but it's striking that he wasn't that good with other teams, before or after. I wouldn't raise an eyebrow, except that the Orioles are one of history's most notorious teams, with a reputation for skirting the rules and intimidating opponents and umpires. Again, real wins and real championships resulted, so I shouldn't question it - or should I?

Career votes points leaders

For additional historical perspective, I've edited and added to Clint's excellent lists to show how they stood in 1910, with NA play added. It's possible that my top 5 counts for the NA are not entirely accurate, since I don't have win shares for everybody, but I looked over the league leaders pretty carefully to see if there were any seasons that might be comparable to the top 5 among the players for whom I did have win shares. I found one case of a one-year wonder breaking into the top 5, but that was all.

10 Honus Wagner

Anyone who has used Win Shares or Runs over replacement level should read the excellent article just out here:

http://gosu02.tripod.com/osusaber/id77.html

Players that don't stick around aren't helping their teams, and those that do are, so I don't see a system that rewards the later as overpenalizing the ones that leave early (again, assuming WS or widgets or whatever you use aren't faulty in the first place). No way this offsets the flaw in WS, it very much over-offsets it (if that's a word).

Of course players who do not stick around do not help their team. Where have I suggested otherwise? My point was that two people can have the same amount of WS, but the value of each player's WSs are different.

If we're not looking at WS per 162 Games, then you have to conclude that Jigger Statz was as valuable to the Cubs as Hornsby was to the Redbirds in '23, which is ludicrous. Looking at the WS rate stat, Hornsby was at 39.39, while Statz was 27.37 WS per 162 Games. Hornsby was forty per cent better than Statz per game he played. Of course, Hornsby did cost his team 13 WS because he missed those 47 games (which I account for), but what he did in those games he did play will not be recognized properly by stating he only had 26 WS.

BTW, I should ask you if you feel Statz had the same value as Hornsby (instead of assuming it)? Forget about any problems with the WS formula that you might have. Just assume it is correct. You would only go by the actual WS number and nothing else?

"But career WS is biased against people that don't stick around past their prime. The combination of the two counteract their cons nicely, IMO. "

I disagree, you need to adjust WS first on the season level, then throw them into the system that squares up peak and career, not try to use that system to fix the flaws with the initial system (WS), because that will never work. You have to assume that WS are correct (like Bill does) and just his system for balancing career and peak separately.

I should have stated earlier that I do adjust WS on the season level. By this, I mean that I look at the WS and the WS per 162 Games per Season for each season for each player in question. I'm not sure that I'm following your point, though. I'm not sure if it's my tired state or that you are missing some words in your post. :-)

>>"My point was that two people can have the same amount of WS, but the value of each player's WSs are

it seems likely that Bill Craver or Jimmy Wood or someone like that snuck into the top five one of those years, I just left the NA alone. But that clearly understates some guys' figures -- most notably Cal McVey.

My v. quick calculations (a little more thorough than eyeballingg, but all done in my head) found that Bill Craver made the top five in 1874, just missed in 1875. Jimmy Wood just misses the top 5 in his best years.

In the NA, the players who happen to be on the teams that play more league games have a significant advantage in accumulating win shares. Wood and Craver have arguments to be among the top five players in several seasons, but they were usually appearing with teams that played fewer games, so they're not among the top 5 in win shares. Pike tended to play fewer games also, so his regular appearance in the top 5 is a pretty impressive achievement, in my opinion.

Andrew,

Im being a little absurd here (but just a little): Taking your "peak" argument to its extreme, Tom Browning is a no-brainer HoMer. At his peak (September 16, 1988) he was among a handful of the best pitchers in the history of the game.

Tuffy Rhodes for the HoM?

Joe:

Thanks for responding. For the most part, we're on the same page. We're basically going in the same direction, but using different paths.

RobC, that is way beyond "a little" absurd. Now you sound like Joe, the anxious father whose daughter has just started dating. All of us who are not "pure" career/aggregate evaluators are to be closely watched because we are dancing on the razor-edge of doing something really stupid. In your paranoid heart you cannot quite get over the idea that somebody is going to submit the following:

1. Mark Whiten

Andrew, don't try to define it. You can't satisfy them. Peaks of any kind are illusions.

I like to use a combination of peak measures. I use Best 3 consecutive years, Best 5 non consecutive years, Number of seasons over 25 WS(adjusted for league strength and length), and a measure I call Peak points(you get a peak point for every adjWS above 25 in a season-25 or less is 0). I admit 25 WS is an arbitrary figure, but I figure if a player needs to use seasons of less than 25 WS as part of his peak, he's going to need strong career numbers to get my vote anyway. Its just a number that feels comfortable to me.

Andrew wrote: ...you privilege people who coincidentally meet the intelligent yet arbitrary mechanism you developed to quantify something inchoate

This phrase screams lawyer. :) :) :)

I look at peak, extended peak and career when evaluating WS -- which is only one component of what I evaluate. I use 3 best years, 5 best consecutive years, 7 best years and a career total, and a separate WS defensive rating. I assign A through F grades to each of those categories based on where the player's results place him vis a vis other very good players from the same position. I use this method even in the defensive category -- in other words, I don't use Bill James' letter grades for defense, because those are given in comparison to all very good players...I'm trying to compare to good players.

I then develop a WS "grade point average" by weighting the five categories. I give a little more weight to the career category because three of the five categories are peak categories (3 year and 5 year) or quasi-peak categories (7 year) and I don't want to get unbalanced. I also give more weight to defense for all positions except 1b, rf and lf.

This phrase screams lawyer. :) :) :)

I couldn't resist checking up on that. Sure enough, from Andrew's post on the "Let's get to know each other ... " thread:

I'm a 32-year-old professor of Constitutional Law and Legal History at the University of South Carolina Law School.

'Fly by the seat of his pants voter' here.

Bennett ... I?m moving him up despite his difficulties staying in the lineup ...

But there's a 5-year stretch in the heart of Bennett's career in which he was playing in 88% of his team's total games. Things do drop off outside of those 5 years, so you can downgrade him for having a relatively short effective career. But I would never, ever, use the phrase "difficulties staying in the lineup" in referring to Bennett. How common is it for modern catchers to play in 88% or more of team games? That said, you have him placed higher than I do. (I'll have him 5th.)

On Bennett's durability:

Bennett was a durable catcher, but his high percentage of team games played early in his career is apparently partly due to easier conditions during those years. Average percentage of team games caught by the 1st string catcher looks like this in the NL from 1881-90, which is the heart of Bennett's career:

70.5, 71.7, 66.1, 62.5, 53.2, 46.4, 47.0, 53.4, 56.2, 60.

Like most catchers, Bennett has a period of higher durability early in his career, but his peak of durability is somewhat exaggerated by the change in the physical demands of the position between the early 80s and the late 80s.

In addition to being durable, Bennett was defensively flexible early in his career, playing at other positions when he wasn't catching. He never led the league in games caught, but he led all catchers in games played in 82, 83, and 85. He was above the league average of games played for catchers from 81 through 85. After 85, he played almost exclusively as catcher, so his rate of overall appearances for the rest of his career drops below the league average for catchers, since guys who were spending time at other positions as well tended to garner more playing time. However, he continued to be above the league average in number of games caught every year through 1890 except 1887.

So he had less trouble staying in the lineup than most catchers, but he wasn't quite the iron man that Chief Zimmer was in the early 90s.

Just so people know, I am not a pure career voter. One of the lists I make each year is W3 career + W3 peak (best 5 non-consec). So, in this list I double count each players best 5 seasons. For most players, it makes maybe 1 one position difference at most. However, it gets guys like Jennings into ballot contention.

Anyway, if you look at my past ballots, you will notice that there is quite often a pure peak guy somewhere in the bottom third of my ballot: Meyerle in '98, Browning for a while (is he a pure peak guy???), Jennings at 13 for 1910, etc. This is my acknowledgment that pure career value isnt all there is to MERIT.

However, my absurdist point above does illustrate something. If you take the peak argument to an extreme it becomes silly (nothing is sillier than Tuffy Rhodes in the HoM). Taking the career argument to an extreme leads to Bid McPhee being a mortal lock, absolute no brainer. Not really all that silly, is it?

Anyway, you use this test all the time: politics, HoM, etc. It at least lets you know where you need to tread carefully.

just noticed this:

Carl,

I started doing the same thing with the last ballot. I've been using 23 WS to measure peak, though and 15 WS to measure average, and then finding the percentage of a players career WS that is peak and the percentage that is above average. I've also used average peak value (average WS of all seasons over 23 WS) and total career WS.

I've also been bothered by the arbitrariness of the 23 and 15 WS cutoffs. I'm thinking of trying to find the average adjusted WS values for each league for each year,and use either 1 or more standard deviations (not sure yet what the value should be) above that average to find peak.

Duke Farrell, average major league hitter.

Farrell OPS+ totals for the 16 of his 18 seasons in which he had > 200 PA:

Any opinions on the quality of the Western League during the 1895-97 period? It obviously was not yet up to the level of quality that would make it last as the AL. But does Mullane get any credit for this additional three years of tail kickin' performance?

I wouldn't. It's not as if he left the majors at the top of his game. He was washed up in the NL in '94. Besides, his Western League stats weren't extraordinary.

The recent discussion about James? ranking system, which emphasizes peak over career, got me comparing Sandy Koufax to Don Sutton. Although I would still take Sandy over Sutton, the exercise gave me a new appreciation for Sutton?s career. Did you know he is 5 shutouts short of being the all-time leader in the post-deadball era? I wouldn?t have guessed that one.

The name right below Sutton on the shutout list is Pud Galvin, who is far ahead of all other 19th century pitchers. I know it isn?t very sabermetric to look at a counting stat that can be influenced by ballpark effects and quality of defensive play, but it has me contemplating the move from EOPG to FOPG.

Grant: Grant didnt get stuck in the IL because he had a lousy attitude or because people thought his minor league numbers were misleading. He got stuck there because of Jim Crow.

No question about it. I doubt that there is anyone here that would disgaree with this statement. Hell, ole Cap Anson would agree with it!

There is a difference between that and guys from the 1860s who all happened to play at a time when the game was quite different and poor records were kept. All 1860s players are equally affected by the problems of the era; Grant is unequally affected among the best players of his era due to his skin color. That makes me more willing to be agreessive in extrapolating his shadowy numbers that I am for say Dickey Pearce.

If you feel Frank Grant is legitimately as good as McPhee or Childs, I have no quarrel with you. The "aggressive in extrapolating his shadowy numbers" part of your post bothers me. All of my 1860s guys are all on the conservative side of my analysis. Is Grant with you? I hope he's not getting a boost because of the "we need a Negro League player from the 19th century" argument.

I hope he's not getting a boost because of the "we need a Negro League player from the 19th century" argument.

Right on, John, but I suspect it is a substantial boost. I am certainly sympathetic to the fact that we are unable to test Grant against his white counterparts because of the racial issues. I agree that it isn't Grant's fault that he had to play in the IL and a poorly-organized Negro League system. But social issues and heart-wrenching analysis aside, how do we know how great he was? Being the finest known Negro League player from the 19th century (if that's what he was) and in a "high" minor league doesn't prove enough to me that he was as good as the crop we've elected to the HOM or those ahead of him on the ballot.

I would characterize myself as aggressive by putting Grant #12 on my ballot. You have to do a lot of extrapolating to get him to the top 5, and it seems to me that some are doing this extra extrapolation because the group --including me-- is anxious to elect a Negro League player. Nevertheless, just because Grant's the best Negro League player we've seen so far doesn't necessarily mean he ought to be in the HOM.

Neither I nor anyone else in this group can say with any certainty that Grant was as good as the white superstars because we don't have any reliable measuring sticks. So, if you think he is a slam-dunk HOMer, your argument is really that he should be elected because he was a good player, he was black and we don't have any other top black players on the ballot right now. That just doesn't seem like a reason to elect him, IMO.

I would characterize myself as aggressive by putting Grant #12 on my ballot. You have to do a lot of extrapolating to get him to the top 5, and it seems to me that some are doing this extra extrapolation because the group --including me-- is anxious to elect a Negro League player. Nevertheless, just because Grant's the best Negro League player we've seen so far doesn't necessarily mean he ought to be in the HOM.

Grant might be a top 5 guy. I'm still trying to find as much information I can about his career because, like you Jeff, I want to see a darker-hued player in the HoM (especially to annoy Anson!:-). However, at this time, I don't see him high on a ballot. Hopefully, one of the electors here can find something to add to his resume that it is lacking now.

Joe,

Do you think you can send me the Pennants Added spreadsheet with the WARP and adjWS information? I've used Wi n Shares very sparingly in my rankings because I didn't have full information for all of the players I was considering. My e-mail address is linked above.

(I sort of missed the discussion on adjWS. I believe that the adjustment gives more defensive credit to the fielders and adjusts for season length. Is this correct?)

If he were Denny Lyons or a slightly weaker hitting Cupid Childs with an 18 year career where would he rate? Somewhere between 10-15.

Maybe this is why I don't have Grant higher. I don't have Childs on my ballot and Lyons can't even see the ballot from where I have him ranked. So I suppose your rationale makes sense, Andrew. It's just that the midpoint for me would be the bottom of the ballot and I've bumped him up 3 places to #12.

Just found this description of WS written by TangoTiger when the WS system was first revealed. His thesis was that WS is the same as linear weights, except expressed differently. He also argues for "Loss Shares".

TangoTiger: "Take a hitter's offensive Linear Weights runs and divide by 10. That gives you batting wins. Take a hitter's outs (AB-H) and divide by 100. Add these two numbers. That gives you wins created. Multiply by 3, and you get Win Shares."

Not clear to me whether you ought to be adding CS to the outs number. I'm just reproducing what I found.

Anyway, Chris, you might want to look at this in connection with computing WS for the NA from WARP3 and see if it produces similar numbers. I have now finished the actual WS calcs for 1871-1872 -- slow going as you might imagine. There are a some horrible teams in 1872 that played very few games, so I've had to make some manual adjustments to avoid negative WS.

I digress.

I'm not contending, and most FOFG are not contending -- that he is an obvious HoMer.

Are you sure? Voting someone in the top 5 is pretty darn close to contending he is an obvious HoMer. Most people have an in-out line somewhere between #7 and #10.

>(4) If he were Denny Lyons or a slightly weaker hitting Cupid Childs with an 18 year career where would he rate? Somewhere between 10-15.

(5) How about we split the difference? OK, that puts him about 6th or 7th.

(6) Is splitting the difference too aggressive given uncertainty principles?

Andrew, it isn't splitting the difference that's too aggressive. It's the idea that his downside is Cupid Childs or Denny Lyons. We just don't know this. We don't even know if his big years in minor leagues were age 18-21 or 21-23.

Then there's us small hall guys, whose in-out line is between 2 and 3.

>There was real baseball being played then, but even the top teams

Our order of finish regarding the top 6 current contenders in the past several years:

Jason, no I don't think the answer is obvious. But just for the record, I didn't list Spalding as one of the top 5 in the '60s. He had notoriety for only 3 years in the '60s, and his overall place in history is based to a vastly greater extent on his documented record in the '70s. Yes, like George Wright, he was clearly one of the better players by decade's end, but if we had WS, he would not be one of the top 5 in WS for the decade as a whole.

You're damn right that it is an educated guess as to who the top 5 actually were, a very educated guess. Start and Pearce are givens, Harry Wright almost a certainty. Al Reach a good possibility, Jack Chapman a possibility. For peak, there's Creighton in the beginning, Start in the middle, and G. Wright and Spalding at the end. Yes, all of those represent extremely educated guesses. There is LOTS of qualitative data from the '60s.

Now, who were the top 5 minor league players of the '80s? And who were their opponents?

By the way, looking at Denny Lyons again in the context of the comparison to Frank Grant makes me appreciate how great a player he was.

Terrific hitter for his position, but his fielding holds him back so he's behind a comparable hitter such as McGraw (who had an abridged career like Lyons).

Galvin was the record-holder for wins from 1888 (when he passed Bobby Mathews) until 1903 (when Cy Young passed him). He was a compiler, perhaps comparable to Jim Kaat or Tommy John, or Don Sutton.

Comparing Pud to Kaat, John or Sutton really doesn't help his cause since the latter three weren't remotely the best pitchers of their era. Galvin stood out to a greater degree.

Our HOM voting by position so far (though that's a hard thing to define):

CATCHER (1): BUCK EWING (C-1B/0F), see also White, Kelly

Yikes, sorry about the all-CAPS in the HOMER list at the top....

Joe#86

I'm in the process of calculating the actual WS for the NA (rather than translations from WARP). There are some glitches, because I don't have reliable park factors, passed ball data or home/road W-L records. I'm using half of the park factors from baseballreference.com (i.e, if baseballreference says 103, I'm using 1.015 as a park factor).

Why half? You may be repeating part of the derivation of 103. In an eight-team league, the principal basis for Batter or Pitcher Park Factor 103 is a home ballpark with run factor ~107 and seven road ballparks with average run factor ~99.

Yes, baseball-reference park factors are Total Baseball BPF and PPF

>Joe#86

Marc#54 counts six factors in the Bill James rating.

re: positional differences

Question: Who hurt their teams more because of missed games in 1894 (a 132 game schedule) - Sam Thompson (99 games) or Pop Schriver (96 games)? Obviously, it's Thompson. Huh? How can this be?

The average amount of total games for a right fielder during that season was 113 games, while for catchers it was 89 games. Big Sam was well below average for his position, while Pop was above average for his.

If the average team for that year was missing 43 games from their starting catcher, then it ceases to be a minus because all teams are affected by it. The Phillies were hurt greatly by Thompson's absence from the lineup, while the Cubs had a little benefit from having their best catcher playing a few more games than the average catcher for that season.

BTW, I'm not saying Schriver was better than Thompson for that season.

Charlie Bennett. I've explained my position before. You've got to play to help your team win. So, you've got to be a heck of a player to get in as a part-time player. He wasn't quite there.

Here is an earlier post of mine that refutes your argument, Clint:

re: positional differences

Question: Who hurt their teams more because of missed games in 1894 (a 132 game schedule) - Sam Thompson (99 games) or Pop Schriver (96 games)? Obviously, it's Thompson. Huh? How can this be?

The average amount of total games for a right fielder during that season was 113 games, while for catchers it was 89 games. Big Sam was well below average for his position, while Pop was above average for his.

If the average team for that year was missing 43 games from their starting catcher, then it ceases to be a minus because all teams are affected by it. The Phillies were hurt greatly by Thompson's absence from the lineup, while the Cubs had a little benefit from having their best catcher playing a few more games than the average catcher for that season.

BTW, I'm not saying Schriver was better than Thompson for that season.

I have posted variations of this subject before and usually get silence. Will someone show where my logic fails? If it can be shown that I'm wrong, I'll concede the point and won't bother you guys about it anymore (until I feel like it again). :-)

I tend to agree with your logic John, though I must admit that hadn't occurred to me as a problem until now. It looks like I'll have to re-evaluate my catcher rankings a bit for the next election.

Thanks for responding, Carl. The best way to find out if my argument is half-baked or not is if I get some feedback about this point.

I concur that John's reasoning is valid. Besides catchers, I use the same type of reasoning to evaluate a pitcher's workload. Modern-day starting pitchers are asked to throw many fewer innings per season than pitchers of long-ago. It is appropriate to compare a player's workload to that of his contemporaries at the same position.

Modern-day starting pitchers are asked to throw many fewer innings per season than pitchers of long-ago. It is appropriate to compare a player's workload to that of his contemporaries at the same position.

Agreed. I do the same thing. Every position is evaluated on its own terms. That's why I'm not as crazy with Caruthers as some others are (though his value has gone down appreciatively with our voters since '98). If he was the only one playing multiple positions, he would stand out greatly. However, we know there were others accomplishing the same feat (though he was the best peak-wise as a pitcher/position player for the 19th century).

I think the logic is dead-on.

I think the logic is dead-on.

Not surprising from a person that makes me look like an EOCB. :-)