I don't know where Hughie will end up on my ballot. But I do start my analysis with a look at peak value, and Hughie had a peak that is a real eye-opener.

At the end of the 1896 season, Jennings tied Billy Hamilton for the best active 3-year peak with 110 adjWS.

1897--Hughie was #1 on 3-year peak with 115. This is the highest 3-year peak between Herman Long's 120 in 1893 (which had the benefit of the final 2 years of fielding bonuses which Jennings did not have the opportunity to receive) and Honus Wagner's 122 at the end of the 1902 season.

1898--Jennings has the highest 3 year peak at 113 and is tied with Hamilton for the best 5 year peak at 179.

For the 1890s overall, Jennings, then, had 4 "mentions" among the highest peaks. That is, he was within 10 adjWS of the top 3 or 5 year peak 4 times (a player can get 2 "mentions" each year, counting the top 3 and top 5 year active peaks). That compares as follows:

Hamilton 12 mentions

If I were going strictly by peak, Jennings might be number one on my ballot. But I don't, so he will, if he's lucky, wind up at the bottom of my ballot.

This is maybe not our most pressing issue, but since I spent some time thinking about Dummy Hoy, I might as well explain what I found.

The website contains impassioned advocacy. It doesn't hurt to read that kind of stuff, but keep your common-sense filters activated.

I can see a sort of weak parallel between Hoy and Minnie Minoso - major league career didn't start until age 25/26, some confusion about age (in opposite directions) while active, aged well and lasted a long time. In both cases, the late start may have something to do with discrimination. In Hoy's case, that's not so much overt discrimination by baseball teams (although there probably was some) as much as it is the fact that society had written a script for the life of a deaf mute and Hoy had to break away from that script to become a full-time ball player. But you can't claim that Hoy was a great player before he was in the majors, and asking what would have happened had the script been different is a "what-if". We can't go there.

It's been suggested that players who mature late and age slowly as ball players tend to live long lives. Hoy lived to the age of 99.

Hoy played in 1796 games, compared to Mike Griffin's 1511. But with 285 more games played than Griffin, Hoy only had about 450 more plate appearances. Hoy's career raw stats were .287/.386/.373; Griffin's were .296/.388/.407. They were both fast, OBP-first outfielders, so we ask how they were at scoring runs, which leads to my gimmick, R*.

With his longer career, Hoy only has about 70 more R* than Griffin. Hoy rose above 90 R* in a season only twice - in the 1891 AA and the 1901 AL. Griffin reached that threshold four times - twice in the AA and in the 1895 and 1897 NL. The difference isn't huge, but I'd give Griffin a small advantage as a run scorer.

Prospectus likes Griffin's defense much more than Hoy's - in fact, it considers Hoy to be a below average CF. For that reason (among others), Griffin is well above Hoy in WARP3, peak and career. Win Shares are more ambiguous, and rates the two players' peaks and careers as a close match.

In the NBJHBA, James rates Griffin 45th and Hoy 47th among CF.

Hoy was squeezed out of the NL becuase he didn't go to Pittsburgh with the rest of Louisville's good players. He played for the pennant-winning White Sox of the 1900 AL, but that wasn't a major league. Whenever there was an unsettled league, Hoy's walk rate went up; without seeing the numbers, I'll assume he drew many walks in 1900.

In summary: Hoy makes a good story (including the hints that he might have been behind umpire's hand signals) and he was a good player. If you give him every break you can and interpret every ambiguity in his favor, you could put him barely ahead of Griffin. I suspect that the people that have Griffin in the lower middle of their ballots will have trouble seeing it that way. Myself, I would probably place Hoy just below Griffin - and Griffin wasn't in my top 15 last year.

PS, having had some further discussions with Paul Wendt about the 1860s, I am inclined to swap Dickey Pearce and Harry Wright. I had Wright 5th on my ballot and Pearce 10th (not to overlook Start, whom I had #1). But the fact is that not much is known about Wright's actual playing career, only that he was always in leadership positions and always having a very strong influence on other very elite, very prominent players. Whether he led by example or by, well, leading, is not clear. So I would revise my "career" rating for achievements in the 1860s from Start, Wright and Pearce to Start, Pearce and Wright. Start and Pearce will absolutely continue to have a place on my ballot and probably very high, like #1 and #5. Wright I am no longer quite sure about.

Comments on the leading candidates' voting patterns:

The Reds' Hall of Fame inducted W.E. "Dummy" Hoy two weeks ago. Read John Erardi, "Hoy enters Reds Hall today", in the Cincinnati Enquirer. http://reds.enquirer.com/2003/08/03/redsbaseballinsider03.html

This is NOT a 1908 preliminary ballot. This is the top PEAKS among eligible players, taking about equal account of 1) the height of the 3 year peak, 2) the height of the 5 year peak, and 3) how long the player maintained that peak (how long he was "among" [within 10 WS] the best players based on 3 and/or 5 year peak.

1. Tommy Bond--the highest 3 and 5 year peaks in adjWS

<i>AL 1900

Another observation:

Pitchers and where they ranked in last year's vote:

counting nichols and c young as 'hybrids' of course...

I attributed 407 adjWS to Joe Start last "year" and someone asked where the number came from. It is Chris Cobb's number, including NA WS. It ALSO includes 50 WS (10 per year 1865-1870) for Start's play pre-1870, so I was incorrect in saying that 407 was Start's 1871ff number.

On the other hand, 50 WS pre-'71 is incredibly conservative. I see two options, really. Ignore pre-'71 altogether and some will. Or, if you want to incorporate some kind of numerical score for pre-'71 I use 120, which is 10 per year for 12 years 1859-70.

Paul Wendt has confirmed that it is documented that Start played elite ball as early as 1859. There is some disagreement as to when he joined New York's leading club, the Atlantics--it may have been 1861 or '62. Either way, he played with one of New York's and therefore America's best clubs--the best, cumulatively, for the whole decade of the 1860s--for 8 or 9 years. He played with Dickey Pearce and seems from the histories that I have read to have been regarded as the better of the two, the team's real star.

If you normalize to 162 games, then 10 is way low. If you normalize to 70, as some seem to, 10 is merely low. On the other hand, Start was not presumably a star right from 1859. So I would attribute 120 WS to Start pre'71, which is 70 more than Chris' 50, and which therefore would increase Start's total adjWS to 477. I think this is a fair figure unless you want to discount pre-'71 altogether, which is your privilege.

Marc, I'm sure you've mentioned elsewhere, but are your three-year or five-year peaks consecutive? If they are consecutive, how are you handling the blacklist years for Charlie Jones (which interrupt his peak seasons) when you calculate three and five-year peaks?

I ask because I see Jones as having one of the better peaks around, but he doesn't make your top 25.

With all the various adjustments applied, I have him at a three-year, non-consecutive peak of 114 WS, and a five-year, consecutive peak (across the blacklist years, and dropping the season he was not able to finish) of 175 WS. Does that match what you have? Where would that place him in your list?

So, I'm not very happy with my last ballot, and I think it has something to do with my pitchers. With Spalding in last time, a pitcher dropped off the top of my ballot, and with three new hitters coming on, Welch dropped off the bottom.

That left me with only Galvin, Whitney, and Caruthers of pitchers in my Top 15. While I'm pretty happy with 5.5 pitchers out of 20 (I think about 1/3 pitchers in the right ballpark), my ballot was stuck with only 2.5 out of 15, which seemed very low.

So, I went back and looked at pitchers again, especially the "Top 6" that Howie mentioned above, and looked at them compared to each other (rather than compared only to Radbourn as I had been doing previously). This what I saw:

All 6 had hit on ball in play rates (H%) better than their teammates. The best was Tony Mullane (139 hits fewer on balls in play) and the worst was Mickey Welch (43 fewer). Galvin was second best (106 fewer hits), but looked at as a rate Jim McCormick and Bob Carthters were better (78 fewer and 57 fewer, in 2/3 and 1/2 as many innings, respectively as Galvin).

Looking at strikeout rates, Pud Galvin was dead last among the six (2.71 per 9 IP). This struck me as odd, since I had been claiming that he was a victim of his defense. If you look at his numbers, though, Pud Galvin actually was a horrible strikeout pitcher. In fact, he only had 4 years (1882 through 1885) where he was above his career average. For the first 3 and last 8 years of his career, he was even worse than 2.71, and often considerably worse. Whitney, McCormick, and Mullane were 1-2-3 in K rate, although Whitney had the shortest career.

Whitney also jumps out as an odd-ball. He had the best K rate (4.04 per 9 IP), the lowest BB rate (1.06 per 9 IP) and was good at keeping balls in play from becoming hits (59 fewer than teammates). He was also a good hitter, better than anyone except Caruthers on OPS+ and BRARP. And yet, despite the ability to strike out hitters, not walk hitters, keep balls in play from becoming hits, and hit for himself, he didn't crack 200 wins, had a losing record, and his career was essentially over at age 30 (he died at age 32 of tuberculosis).

Also, I looked at ERA compared to RA to get a feel for defense. I was suprised that Galvin, Whitney, McCormick, and Welch all had ERA/RA of between 55% and 57%. At least in terms of being hurt by errors, they quartet were identical (Caruthers rate was 64% and Mullane's was 61%. Those two had the most AA time though, so something might not be comparable.)

Whitney, on the other hand, was the least "clutch" pitcher of the six. Looking at runs allowed based upon his stats, Whitney gave up 95 MORE runs than his component stats would suggest. Bad luck? Well, Galvin had , Welch had 154 fewer, Caruthers had 101 fewer, Galvin had 129 fewer, and Mullane had 115 fewer. It seems like being able to "bear down" was something good pitchers could do, but that Whitney couldn't.

Galvin, on the other hand, was an absolutely horrible hitter, even for a pitcher. BRARP puts him at negative -- he was below a replacement level hitter even for a pitcher! Caruthers BRARP of 302 laps the field -- Whitney is second at 177.

Tony Mullane does not lead in any category except H% and had the worst walk rate of the six (1.9 per 9 IP), but he is second or third in every other category. He ranked 2nd in ERA+ and WARP1, and third in Wins, IP, K/9, and the batting numbers (OPS+ of 87 and BRARP of 155). He may the best all around pitcher, but without being the best (or even the second best) at much. It's depend on the AA discount.

Welch gets longevity points, and has the second best W% to Caruthers, but doesn't do much of anything else.

I don't have any conclusions yet. All six pitchers have big strengths and gaping weaknesses, and I see why so many of them are having trouble gathering support. I think right now, I am leaning towards dropping Galvin down from first place on my ballot to maybe third or fourth, and them bringing up some of the others, since I just don't see the gap between them that I did before. Caruthers, Mullane, and Welch may move up. Whitney may move down for poor clutch pitching.

I'm just not sure.

Wright I am no longer quite sure about.

From what I have read about him, he was considered the best hitter during the 1850s and was a fine fielding centerfielder. I originally thought he was getting leadership points, too, but he appears to have been a terrific player (thought I think Pearce was the better of the two).

Here's my ballot from 1907 (without Hamilton, who I'd had in 2nd place).

1. Ezra Sutton

Does anyone expect to never vote Jennings into the HOM yet put Koufax in on the first ballot? Is that inconsistent??

Good discussion. 1) I think I am already done with 19th century pitchers.

2) Good question, Chris. I use 3 and 5 yr BOTH consecutive peaks. 3 AND 5 because I think a 3 year peak coincides with the casual use of the term "great." If a player is an MVP candidate for 3 straight years, that is a "great" player. On the other hand, many players have 3 great years, and so the 5 year peak begins to separate the men from the, er, men a little bit. But I think consecutive seasons are still what catch the attention of the average fan and therefore reflect the common sense, reflexive use of the term "great." James uses any 5 and I wouldn't argue the point, I just do it differently.

Therefore, Jones doesn't do well. Right now his best score is 108 for 3 yrs as of 1880 when Hines was the leader with 136. If I just ignore the blacklist years like I ignore WWII (Ted Williams '42 and '46 years are consecutive, e.g., unlike his various body parts, I guess), Charley only gets to 170 in '84 and '85 while the leader is Gore at 200 and 198, respectively, so Charley is still only "close" to "close," but not "close" if you see what I mean. And if I go to non-consecutive seasons then everybody else benefits just as much as Charley. (So, also, our numbers--yours and mine--are close enough for horseshoes, so I assume your Gore and Hines numbers are comp to mine also.)

His problem is the seasons surrounding the blacklist years--'80 and '83--aren't that great. His great years are spread far apart. So like I said, if I used non-consecutive, his number would jump, but so would many other players.

So Charley is "close" to "close" in the final analysis. I think we have underrated him here, but it's 1908 and the backlog is starting to really pile up. I think his "time" has passed.

3) As to Harry Wright, I still think he was a great player and not an Earl Weaver or Sparky Anderson, he was a player AND a pioneer, but I now agree with John. Pearce was better. That's all that's changed.

4) And Joe, if you've got Bennett 11th and Jennings below that (and Duffy below that), then you are no peak-man! As somebody said, what are your intentions re. Koufax or, let's say, Kirby Puckett?

I don't get it. What's the difference between 5 consecutive years of greatness or 5 non-consecutive years of greatness. Both have equal value. To say that consecutive years is what catches the attention of an average baseball fan doesn't sound very persuasive to me!

>What's the difference between 5 consecutive years of greatness or 5 non-consecutive years of greatness.

Choose your method, whatever works for you.

The point was that in fact it doesn't make much difference.

I look at 3-year peak (not consecutive), 5-year peak (consecutive), WS per 162, and career. When running a composite ranking, Jennings? peak numbers are strong enough overcome his short career and top the list. He will be somewhere in the top 3 on my ballot.

I think Jennings clearly rates in the top 10 of this group, and can be placed much higher if one likes. He kills anyone who doesn't have a lot more stellar seasons, and it does seem that no one touches his best 5 seasons..

>Yes, the peak man may have pushed his team over the top

I've toyed with the idea (OK I actually developed a method) for pegging "replacement value" (that's not what it is at all, but by analogy) at approximately all-star level (i.e. top 3 at a given position). After that, you're helping your team finish 2nd, IMHO.

I don't actively use this method here, I might add, but it was fun. The top players by this method will not surprise anyone, which I actually take to be something of an endorsement:

1. Babe Ruth 84

Does anyone expect to never vote Jennings into the HOM yet put Koufax in on the first ballot? Is that inconsistent??

I don't expect to have Koufax that high either (though I think his career was longer for a pitcher who started his career during the high-offense fifties than Jennings as a shortstop during the Gay Nineties). If I'm going to be consistent with my ranking method, I can't give Koufax a pass because he retired at thirty.

Andrew, do you use 5-year consecutive peak or non-consecutive peak?

My current prelim ballot is fairly similar to my 1907 ballot.

1. Ezra Sutton (1)

>As for Jennings, at the moment for him to make my list I would have to extend to a Top 25. I don't see

In BJHBA2 (1988), Bill James listed Hugh Jennings on the all-star team of the 1990s, one player at each fielding position.

In the HOF book (1994), Bill James wrote of the 1945 election:

Koufax is unlikely to get a high ranking from me, but I will look at it when the time comes (as I will continue to look at Jennings and anyone else.)

Marc, your point is well taken, but illustrates where we differ. I choose to look at the players without emotion. I don't care who looked like they were a better player. Often looks can be deceiving. I try to stick to actual evaluation based on fact. I recognize that numbers don't tell you the whole story, but I would rather go by those numbers than a stilted account from a sportswriter who might have had preferences for a player for reasons other than his on-field performance.

Perfect example: Jennings met the Pope twice. And I am supposed to care about this... why? It has nothing to do with his skills on the diamond, and that is all I am evaluating.

(As an aside, if I were to bring in "extra points", they would go to Hoy. Playing under the hardships he had to endure is much more impressive than meeting some religious figure. Even if he did so twice.)

Can't anyone meet the Pope if they are willing to donate some $$? :-)

I wouldn't want to stand next to you when the lightning bolt hits you. :-)

This is incredible to me--Dummy Hoy ahead of Hughie Jennings?

I have Jennings over Hoy by a considerable amount because of my career attrition bonus for Hughie.

I don't see the big deal with Hoy. If he hadn't been deaf, I doubt anybody would be trying to organize a HOF effort for him. Not that he wasn't a fine player.

This is in response to Matt B's post on pitchers (#18 above).

The way you are looking at rate stats, Matt, is selling Pud Galvin short in a couple of ways.

First, it's usual that long-career pitchers will have lower rate stats -- they were able to keep pitching during a gradual decline phase, while the short career pitchers had to quit. To measure Galvin fairly against Whitney, Caruthers, etc., you need to compare peak to peak, and then, I would argue, give Galvin credit for career length. He is truly unique among 1870s & 1880s pitchers in the length of his career. Considering he has another 1000+ innings from 1876-1878 for which we don't have good records, his career is _much_ longer than anybody else's, and he was an above average pitcher for most of that time, though he was only great for 3 or 4 years.

Second, the fact that the conditions of pitching were changing so frequently and drastically during this period makes career rate stats an unreliable guide to a pitcher's talent, as the conditions during his peak years will heavily influence his overall rates. Check the season-by-season leaderboards, and you'll see that the rates of walks and strikeouts shifts by more than 100% during the 1879-1889 decade.

Galvin was not a great strikeout pitcher, but for the first half of his career he was a decent one: he shows up in the top 10 on K rates and near the top of the K list occasionally, since he threw so many innings. His K rate jumps in 1883 and 1884, but so does everyone else's. Since he pitched more innings, at a higher strikeout rate, in these seasons, than at any other time of his career, out of context he looks like an excellent strikeout pitcher for two years and a terrible one elsewhere, when he was a decent strikeout pitcher 1879-1884. In 84 he blew out his arm, and he was _not_ a strikeout pitcher after that. After a bad year in 1885, he was an effective "control" pitcher from 86-89.

In fact, Galvin was always a great control pitcher, the best of the era. He is first or second in BB/9 every year from 1879-1889 (I didn't look at seasons after that). His career rate isn't as good as Whitney's, however, because the overall walk rate is much higher in the later 1880s. Galvin threw a lot of innings then, still leading the league in BB/9, but his career rate was going up. Whitney didn't throw many innings under these conditions, so he has a rate that is more reflective of early 80s conditions. And he did have a better walk rate than Galvin, 83-87.

Folks on the board seem to have been cooling on Galvin a bit in the last couple of years, but he is far ahead of all the other pitchers currently eligible. At his peak, he was as good as the eligible pitchers with shorter careers, and his career value is _way_ beyond theirs. If Mullane didn't have AA issues, he would be in the ballpark of Galvin's career value, but Galvin's value is still higher.

Can't anyone meet the Pope if they are willing to donate some $$? :-)

How many WS do you have for Tom York? My notes have him far from 6th. (38th out of eligible players)

I'm curious to see Favre's numbers, but Tom York is indeed at #6 on my list among eligible position players in documented, fielding-adjusted, season-adjusted, league-adjusted, pythagorean-adjusted career win shares, including win shares for NA years translated from WARP, and pitching WS translated from WARP, where appropriate. My top 15 on this list:

1) Ezra Sutton, 458

Ah, I thought he was referring to straight WS, not adjusted.

So, it's generally been my view that when comparing players' careers against "replacement", you really have to consider a "replacement career" and not just what replacement level is for that day/ year. If Player X is injured, his replacement that day will likely be at replacement level, but the team is not now relegated to 15 years of the guy not quite good enough to make the 2003 Tigers.

So, anyway, I would consider a "replacement career" the level at which a team could expect to perform over 15 years without the star player -- a level that I would consider a little below average -- near the median level.

For a full season, I would consider that to be about 18 Win Shares. Here are my Top 15 players (offense only) and then my Top 10 non-HoMers. (adjusted for season length)

Top 15 WSoMP (Win Shares over Median Player)

212.16 Connor

184.57 Anson

182.52 Brothers

166.65 York

165.11 Hamilton

164.96 O'Rourke

158.67 Kelly

144.03 Sutton

138.64 Gore

136.83 Hines

132.88 McVey

131.42 Pike

124.73 Browning

118.65 Ewing

116.26 Stovey

As for Bill James and peak vs. career, I think of his Drysdale-Pappas discussion as pretty much the final word. When comparing two players of similar career totals, the one with the higher peak was probably more valuable as it relates to real pennant races.

My point about Jennings meeting the pope was only partly in jest. He was a guy who made things happen, who achieved his goals. He did that on the field, too. He was the straw that stirred the drink.

By "fares very well" above, I meant that Jennings makes sense "on ballot", not "on top of ballot".

Can somebody help me figure out how Tom York has so many adjusted WS?

Andrew, you're forgetting that he was very durable. OPS+ is not going to show that.

However, I do think adjusted WS does overrate him a little (as it does with all other players from that time). I think he's in Tiernan/ Thompson/Stovey/etc. territory.

"When is John McGraw eligible?"

Part of the killer class of 1909:

***1909 (September 14)-- elect 2

People asked about my numbers regarding York. I used the adjusted WS and WARP-1 figures listed in the "Pennants Added" thread (my rankings are from the adjusted totals, not the rankings from the pennants added).

One reason that York has so many adj. WS is that he shows up as an excellent defensive outfielder in both WS and WARP, esp. during the NA years, when all position players have more defensive value. Translated WS sees almost 40% of York's value in those years coming from his defense. In part that's because his offense is nothing special during those years; in part that's because defensive value is generally higher in the NA.

Taking that into account, two adjustments might be order in placing York by means of WS.

1) If York's fielding WS seem high, it might make sense not to apply the usual WS fielding adjustment to York's NA numbers, which would drop his overall adjusted WS by 12, to 329 in my calculations.

2) I think it's appropriate to set average or slightly below average value for position players higher for the NA years to take account of the fact that fielding WS are going to be higher across the board. Matt B uses 18 as slightly below average career replacement level -- I'd probably set it at 21 for 71-72 and 20 for 73-75. I calculate peak as value above average: I set average as 25 for 71-72 and 24 for 73-75. That brings York's value above average down a chunk.

With these two adjustments, I see his career, as John Murphy does, as pretty similar to Thompson, Tiernan, Griffin, et al., except that he had a higher peak than Griffin and a couple more average years than T & T. The differences between 12 and 25 in the rankings right now quite small.

I differ from John, of course, in seeing Stovey a good notch ahead of this pack of outfielders.

Going by bat alone, I see no reason to put Hughie Jennings ahead of Cupid Childs. Childs played 1456 G at .306/.416/.389 (career OPS+ 119) while Jennings played 1285 G at .311/.390/.406 (career OPS+ 117). That looks pretty similar except for the 170 or so extra games. Childs's best hitting year was 1892. His .317/.443/.398 was 3rd in the league in BA, 1st in OBP, and about 50 points off the leaderboard in Slg. Jennings' best hitting year was 1899. His .401/.472/.488 was 2nd in BA, 2nd in OBP, and about 50 points off the leaderboard in Slg. There's a huge context difference between 1892 and 1899, so both years come out as OPS+ right about 150 - pretty much a wash.

So Childs was as good a hitter as Jennings in rate terms and had a slightly longer career (which is still on the short side). But Childs was a second baseman, and Jennings a shortstop, which matters. Maybe that brings Jennings level with or maybe slightly ahead of Childs. But to put Jennings well ahead of Childs, I would need to convinced that Jennings was, for a brief time, not just a shorstop or even a good shortstop, but a great defensive shortstop. Can anyone convince me of that?

One thing to note: at age 29, Jennings was putting in some time at 2B, and from age 30 on, he was a first baseman. I don't know what to make of it, but that career path isn't normal for great defensive SS. Did he have an arm injury?

John McGraw had better offensive rate stats than Jennings (.334/.466/.410) but even less playing time at 1099 G. McGraw had trouble staying in the lineup.

But to put Jennings well ahead of Childs, I would need to convinced that Jennings was, for a brief time, not just a shorstop or even a good shortstop, but a great defensive shortstop. Can anyone convince me of that?

Dunno what sort of evidence you like, but WS sees Jennings as the best defensive shortstop in the game in 1894, 1896, and 1897 and as second best in 1895. So, yeah, he was a great defensive shortstop for a few years, where Childs was never a great defensive second baseman.

Joe wrote:

"There's a definite flaw in the logic of replacement level approaching average, although I still haven't been able to pin it down. I think it's along the lines of if you remove an average player from the league, the overall quality of play goes down, because he gets replaced with a player that is far below what was average previously, dragging the new average down."

Assume that the best player in baseball is Barry Bonds, and that the 751st best player is Clay Bellinger, currently at AAA Fresno. Assume that if Bonds gets injured, Bellinger will be called up. The quality of the Giants, the National League, and all of baseball will definitely drop by the difference between the level of play of Bonds and Bellinger.

If Bonds is injured for all of 2004, though, and decides to retire, Clay Bellinger will still be the 750th best player (let us assume) and will still deserve a roster spot. But the Giants will have $20 million freed up and will find a better outfielder than Clay to replace Bonds.

Clay will find a job with the Tigers or D-Rays or someone. The quality of major league baseball has still decreased by the difference between Bonds and Bellinger, but the cost of that decrease has now been spread over a number of teams, with Tigers marginally worse for replacing their outfielder with Clay, and some other team with Detroit's outfielder also marginally worse, and the Giants marginally worse by the difference between Bonds and whoever they replace him with.

The important thing to note here is that, even though the league has declined by the difference between Barry and Clay, the Giants no longer have. If you are rating Barry based on "Value to the Game of Baseball", then his value is value over replacement. However, if you are rating Barry based on "Value to the Team on Which He Plays", then the value is signficantly less.

Either seems to be an valid method of determining value. Great players are often praised for their "Contribution to the Game." But if you want to determine how valuable a player is TO HIS TEAM, then you have to consider what actually happens when a player retires (rather than, say, sprains his ankle and is out for two weeks.) What happens is that the cost between replacement and median quickly disapates among the teams in the league, and the cost to the individual team drops to only the difference between the player and the median.

When I consider the costs of a short career player, I value to costs of that career TO THE TEAM HE WOULD HAVE OTHERWISE PLAYED FOR, not to MLB as a whole. I think that, given that definition of cost and value, value over median is appropriate for considering time periods greater than one season in length.

Let's look at this year's inductee -- Billy Hamilton. He played his first two years for the Kansas City AA team, and then they folded, so no need to replace him.

Then, he went to PHiladelphia for 7 years. In the 7th year, 1895, he was in the outfield playing center between Sam Thompson and Ed Delahanty. Pretty good outfield. Hamilton has a 1.084 OPS (8.3 WARP).

1896 he's in Boston. Philadelphia's centerfielder? Mixture of Duff Cooley (.724 OPS, 1.7 WARP), Joe Sullivan (.687 OPS, 1.1 WARP), and Sam Mertes (.610 OPS, -0.1 WARP). They played a combined 149 games for a combined 2.7 WARP. In 1897, the Phillies settled on just Duff Cooley, and he gave them 7.5 Wins Above Replacement in center.

Last year's inductee was Al Spalding. In 1876, he jumped from Boston to Chicago, taking his 12.5 WARP with him. Boston replaced him with Borden (1.1 WARP), Bradley (2.2 WARP), and Manning (3.1 WARP) for a combined 6.7 WARP. In 1877 they signed Tommy Bond (8.7 WARP).

These are just two examples (although not quite taken at random, I didn't know what I'd see before I looked at them.) Of course, good teams like Philadelphia in the '90s and Boston in the '70s will do better jobs replacing stars than most. But I think it shows that teams are not SOL when they lost players in the 19th century -- especially between seasons. In both cases, HoMers were replaced with above-average players within two short years. The stories with Spalding's and Hamilton's replacements upon retiring from Chicago and Boston are similar, with solid performances just a year or two out (Hamilton was actually partially replaced with Duff Cooley again).

It is by far the expection rather than the rule in the 19th century that a team can't replace a star with at least mediocre to average talent within a year or two.

Assuming replacement level talent absent the player's existence is simply counter-factual in most cases that occur between seasons.

I agree that based on career totals (including Jennings great defense) that Hughie is only slightly ahead (but ahead!). But the peaks! Childs had a good peak--he was among the leading active players in 3 yr peak adjWS for 1 year. Jennings had a huge peak--among the top 3 and/or 5 year adjWS peaks 4 times in 3 years and the highest raw peak adjWS numbers between 1892 and 1899.

OCF, are you uninterested in the shape of those careers? Is the height of their respective peaks of no consequence at all?

These are just two examples ... Of course, good teams like Philadelphia in the '90s and Boston in the '70s will do better jobs replacing stars than most.

Here's another example in the same vein, which still doesn't prove much. In 1896, John McGraw played just 20-some games. The Orioles reached out and found Jim Donnelly, who hadn't been a major league regular since 1888, and used him at 3B. Donnelly had an OPS+ of 110 for the year. Maybe that it was just dumb luck to get that kind of performance (anyone remember Cesar Cede?o as the Cardinal first baseman in September, 1985?), but it also means that the Orioles weren't exactly helpless.

Joe is right on. To take it a step further, only if some Retrosheet guru can show us that Hughie (or somebody) carried his team through the critical stretch run and/or excelled against the team(s) they had to beat, can I really see giving him a lot of credit for contributing "something extra" to the team's pennant runs.

I know that Pennants Added would say otherwise, but...

Philosophically, I believe that a team made up entirely of John Oleruds would finish second. Hey, that's better than 5th. No shame in second place (LOSER!).

Whereas a team full of average players with one ARod or Sammy, a guy that can put a team on his shoulders and carry it along, might actually win it all. I saw Kirby Puckett do that.

So I think 5 years of MVP candidate (Hughie Jennings), in the real rather than statistical world, has more impact on real pennant races than 20 years of John Olerud.

Cannot justify this mathematically. It is a priori but it is how I feel.

Philosophically, I believe that a team made up entirely of John Oleruds would finish second. Hey, that's better than 5th. No shame in second place (LOSER!).

Whereas a team full of average players with one ARod or Sammy, a guy that can put a team on his shoulders and carry it along, might actually win it all. I saw Kirby Puckett do that.

So I think 5 years of MVP candidate (Hughie Jennings), in the real rather than statistical world, has more impact on real pennant races than 20 years of John Olerud.

Here's my problem. When a Jennings-type player leave the field in a hurry after his five year of greatness, his team could wind up with Frank Taveras/Tom Veryzer/Johnnie Lemaster type players (I know I'm showing my age :-) for the next fifteen years (think about the Mets at third base all those years). An average player at shortstop would be better than those clowns.

While Olerud is not Gehrig, he gives his team an above average player at first every year for a very long time. They don't have to worry about being stuck with Charlie Comiskey over there.

I'm all for giving credit for the peak guys, but let's not get too crazy about it.

"Philosophically, I believe that a team made up entirely of John Oleruds would finish second. Hey, that's better than 5th. No shame in second place (LOSER!).

Whereas a team full of average players with one ARod or Sammy, a guy that can put a team on his shoulders and carry it along, might actually win it all. I saw Kirby Puckett do that."

Try playing it that way on Diamond Mind and I think you will be proven wrong. I haven't tried it, but I would think so.

And anyway, how would you explain Bill Simmons' Ewing theory? :-)

John Murphy wrote:

"When a Jennings-type player leave the field in a hurry after his five year of greatness, his team could wind up with Frank Taveras/Tom Veryzer/Johnnie Lemaster type players."

To continue with my previous theme, though, we could look and see who ACTUALLY replaced Jennings. In 1898 (the last of his "Big 5" years), Jennings played 143 games for Baltimore and earned 12.8 Wins Above Replacement (WARP-1).

In 1899, he was replaced by a combination of Bill Keister (90 games at short) and George Magoon (62 games at short). Magoon was worth a mere 3.6 games over replacement for Baltimore and Bill Keister was worth 7.5 (pro-rated to 5.0 at shortstop replacing Jennings and 2.5 at second base).

So, while Jennings was worth 12.8 Wins over Replacement in 1898, he was only worth 4.2 Wins (12.8 - (3.6 + 5.0)) over the actual people who replaced him in 1899.

Of course, when Jennings or second baseman Gene Montreville sat out a game in 1898, the team was forced to plug in Art Ball (career 0.8 WARP). The fallacy is assuming that -- just because some mindless team today might end up giving Art Ball a four year contract in a panic -- over the course of a season or two, most teams will continue to get Art Ball production after a Hughie Jennings leaves.

In fact, even though the top three offensive stars Hughie Jennings (12.8 WARP in 1898), Willie Keeler (8.8 WARP), and Joe Kelley (8.4 WARP) and top three pitches Doc James (9.8 WARP), Al Maul (8.3 WARP) and Jay Hughes (5.4 WARP) -- all going from Baltimore to Brooklyn in 1899 for a gross loss of 53.5 Wins Above Replacement! -- the Orioles dropped from 96 wins to 86 wins in 1899.

If replacing talent was so difficult that "replacement level" was the proper point of comparison shouldn't it be considered absolutely shocking that the Orioles were able 43.5 out of 53.5 Wins in just one year? (Although, not surprisingly, the finished 1899 fifteen games behind Brooklyn, the team that took all of their stars.)

For a strong team like Baltimore, replacing stars with around-average players was simply not difficult.

On the Olerud thing, several issues:

1) Which John Olerud are we talking about? The one who has put up 22 & 21 WS in 2001 & 2002 or the one who has put up 33 and 37 WS seasons in the past? Or the one who has put up 26 and 27 WS seasons? Olerud has been a top-notch player.

2) For "Kirby Puckett" to put those "Twins" over the top, you need all those average players, too. Kirby can't "make the difference" unless they're all average, and that's not something you can just assume. They earn being average.

3) How do you tell who puts a team over the top? What kind of credit do you give the player who averages 16 WS a year but puts up 24 this year? Could those be the ones that make the difference?

All of this is to say that, as a pennant is a pennant is a pennant, a win is a win is a win. The players who help their teams win more games are better players, and the metrics reflect that.

Peak value is worth counting because it shows how much a player did to help his team win the pennant in a particular year.

But there are no "special wins" that are the wins that put a team over the top. The wins in April count just as much as the wins in September, and the wins that a 15-win share player earns are exactly 50% of the wins a 30-win share player earns. In most cases, a team needs every win it can get.

"Matt -- didn't Baltimore get all of their replacements from Brooklyn when the clubs were merged?"

Two of the major replacement were Candy LaChance and Jimmy Sheckard, did come from Brooklyn. Bill Keister came from Boston, Dave Fultz came for Philadelphia, and Joe McGinnity and Harry Howell were unproven rookies. You work with what you have. Without LaChance and Sheckard, there would likely have been less ground made up, but again, those positions would not have been played by replacement players.

You take them where you can get them. The point is that I think the burden should be on the "low replacement level" folks to show that -- generally, or in the majority of cases -- if a star is lost unexpectedly results will look more like Baltimore 2000-2001 (steep decline after losing Albert Belle) rather than Baltimore 1898-1899 (piecemeal replacements make up lots of the slack).

It is certainly difficult to replace Billy Hamilton -- but it is extra difficult to replace him if he is still there (injured) compared to if he is on another team or retired. For one thing, with Hamilton off the team, you don't have to devote any resources to Hamilton (such as paying him). Also, players will not want to play for you if they think they'll only play for a few months until Hamilton returns.

It is not "easy" to replace a star, but good teams do it, even if John McGraw did it better than most. Only the worst of the worst could not put together a better team with an off-season to work on it, compared to a single day. Using "single day replacement" to consider the value of your 15-year-career stars assumes that failure to compensate is the norm -- that along with failing to have Barry Bonds for 5 years, you also have to use a replacement player because you are still paying his salary and have no other option.

The case for Jennings is the claim that he has a huge peak. The evidence for the huge peak rests very heavily on the claim that he was a great defensive shortstop. He was a very good hitter, of course, but there were other very good hitters. In his Prospectus numbers, his FRAR is roughly equal to his BRARP, at 60 runs per year or so. Is that saying that about half of his value above replacement is his fielding?

One thing that makes me a little cautious: he was leading leagues in putouts every year but not in assists. The problem with that is that there are putouts at second base which can be handled by either the shortstop or the second baseman. Just checking the second baseman with the most playing time: in 96, 97, and 98, De Montreville and Reitz for Baltimore had 2.17, 2.20, and 2.20 PO per game played. Checking a couple of other good teams for PO per game by the 2B: Childs in Cleveland, 3.18, 2.80, 2.43, Lowe in Boston, 2.64, 2.20, 2.74.

I know WS is supposed to correct for this, and WS likes Jennings's defense. Is the correction sufficient?

Whether the Orioles might or might not have signed the 1890s version of Cesar Cedeno, if Jennings had gone down,

Let's leave Cede?o out of it - different roster rules, different case, and nobody has a right to expect an OPS+ > 200 from a replacement. How does that sentence read if it starts out "Whether the Orioles might or might not have signed a second player like Jim Donnelly ..."? Of course, there's a pretty good chance that Donnelly was playing way over his head for a year.

To continue with my previous theme, though, we could look and see who ACTUALLY replaced Jennings.

First of all, I was referring to a Jennings-like player, not Ee-Yah himself.

Secondly, of course a team can find above average players to replace a great player. My point was that it doesn't happen all the time.

Although it may be viewed as an accident of fate, Jennings peak is worth more to me because it helped (put over the top might be a bit strong) win actual pennants for the Old Orioles. If he had these seasons on some team going nowhere (like AROD in Texas) I would give them less value.

(sigh)

So you reward Jennings for having good teammates, but penalize him if he had the misfortune for playing with lousy ones. With all due respect, I don't understand that (unless Jennings was the de facto GM).

Life is one big game of chance. Baseball is life.

The "chance" that a given player will be in the right place at the right time to be "the one" who puts his team over the top is of course miniscule. That is, that he joins a team with just enough other talent to contend, but not enough to win without him. This happens so rarely in the life of a given player that it is virtually a hypothetical situation. But it is a hypothetical we frequently use to make a point, and of course one could argue that MVP voting each year is based on who in the voters' opinions comes closest to having lived out the hypothesis.

Since this is all imaginary anyway--the player who "puts his team over the top" (who is an MVP candidate and someday a HoF or HoM candidate) is the one who has the most value. Everybody else is in support.

My concept of "greatness," then, is NOT based on the literal reality of this scenario but only on it as a hypothetical example. A player may play for an 8th place team in reality. But the question of greatness, of HoM-worthiness, etc., is: Was this player one who could or would "put his team over the top" if he ever had the opportunity?

It is analogous in a way to some Keltner Test questions, such as: If he was the best player on his team, could the team win a pennant (or whatever James' precise language was)?

My concept then is that *the John Olerud playing 1B for the Seattle Mariners in 2003*, over the course of an entire career with a more or less normal performance curve (i.e. not always at the 2003 level), would *never* be the best player on a team that could win a pennant. He would never be the guy who "put his team over the top." He is not a "great" player. He is not a HoF or HoM candidate even if he plays for 40 years and finishes up with 600 WS. You can't "become great" by hanging around.

He would never be the best player on a team that could win a pennant.

I doubt seriously that I have clarified my earlier statement but I'm sticking to it.

I want to share something that I have picked up using a formula to find out what the correlation was (is) between actual wins and hitting, pitching and fielding WS.

I have only completed four seasons, but these are the results so far (in order of importance):

1876: fielding (47%), hitting (25%) and pitching (25%)

Said of Pud Galvin:

I've with John Murphy on this one. (First time we've ever agreed on something?)

I've agreed with you on a number of things, but I guess are differences are more pronounced. Since it's only about baseball, I don't think we would get into a fistfight over them, however. :-D

You can't "become great" by hanging around.

Except his durability would be, undeniably, great. I think that matters. If you don't, that's cool. It's subjective.

The quality of major league baseball has still decreased by the difference between Bonds and Bellinger,

I think that's the key point.

but the cost of that decrease has now been spread over a number of teams, with Tigers marginally worse for replacing their outfielder with Clay, and some other team with Detroit's outfielder also marginally worse, and the Giants marginally worse by the difference between Bonds and whoever they replace him with.

It doesn't matter whether it's spread over 1 or 10 teams, the total replacement cost is the same. I think that how that cost is spread around the league is a reflection of the relative acumen (and differences in priorities) of its organizations. That a good/rich organization can come up with good replacements by taking advantage of the not-so-good organizations says little about where the true replacement level is. (OTOH a very bad organization can distort the whole picture by giving too much playing time to sub-replacement players.)

I don't think that looking at how the Orioles (or Phillies or whoever) replaced a specific player is particularly revealing. I believe that a look at the talent turnover at the position league-wide is required to get a better picture.

2002 John Olerud almost All-Stars:

Joe's team was a single-year team, but most of the players on that team were having years in line with their careers.

I know it's too staggering a prospect to contemplate doing for real, but it's interesting to speculate about the looks of a Hall big enough to include, say, Tim Salmon and Ray Lankford. These are guys who could play ball. These are guys who created wins for their teams, lots of them. Every one of you who has been a fan attached to a particular team for a while could name players like this - but they're not eveywhere, and most of the players your team has used were far worse.

Has there been a list of new eligibles posted yet? Yes, I know Jennings is eligible, but is there anyone else worth mentioning?

Here are the next 5 years subject to usual caveats: Negro league players (Grant in 1909) and occassional other guys fall thru the cracks. My script aint perfect and I havent tried to fix it.

1909:

I havent seen anybody point this out, just saw it in an article on BP.

Marc wrote: "PS, having had some further discussions with Paul Wendt about the 1860s, I am inclined to swap Dickey Pearce and Harry Wright. I had Wright 5th on my ballot and Pearce 10th (not to overlook Start, whom I had #1)."

Marc, I understand why you would drop Wright on your ballot b/c his reputation on the field is unknown, but why does that information propel Pearce from #10 to #5? If Wright goes to #10, why doesn't Pearce go to #9?

As part of a Jennings comparison, I've recently seen a post that put Joe Kelley in the "shouldn't have been elected" group along with McCarthy. Maybe that was intentional, but I think it is George Kelly that belongs in that group (as stated in Brad Harris' post), not Joseph Kelley.

RobC: Yeah, I knew that, and with his short career, it means that Hughie just dominates the single-season HBP list. One assumes that it comes from of crowding the plate and intentionally or semi-intentionally putting some body part in the path of the ball as a tactic for getting on base. Of course, "intentionally" is illegal, but it takes a gutsy umpire to make that call. Any volunteers for the position of "gutsy umpire" in Baltimore in the 90's?

That would be most of it, but I wouldn't be surprised if Jennings were such an ornery cuss that pitchers also threw at him sometimes.

At first glance from BP, Joe Kelley looks like a clone of Sam Thompson playing LF. Comparable peak, similar career value over somewhat longer career, more championships. OTOH, Win Shares sees Kelley as clearly the superior player.

>Marc, I understand why you would drop Wright on your ballot b/c his reputation on the field is unknown,

Joe Kelley can be a subject of conversation now, since he's part of the Baltimore story along with Hughie Jennings, Steve Brodie, Wilbert Robinson, John McGraw, Willie Keeler, and some others from post #101. But Kelley won't be eligible for election until 1914, the next to the last of the old Orioles to hit the ballot. (Only Keeler comes later.) The main purpose of either agreeing or disagreeing with jimd in #107 would be to comment on Sam Thompson as a candidate.

Sorry I jumped the gun. It was more to point out that Joe was a legit candidate (like Thompson, and unlike George) than to start a premature debate on his merits.

Although I do not award bonuses for being the best at a position, I think it is a reasonable reality check. For example, after Pud Galvin is inducted, I can imagine a person looking at the remaining pitchers on the ballot (McCormick, Mullane, Welch . . .) and see that they're all about the same, and that therefore even the one who is marginally better is not really HoM-worthy.

On the other hand (and here is my point) if you look at catchers, you get a different picture altogether. This year, Wilbert Robinson is eligible, which gives the first real competition to Charlie Bennett since Jack Clements and -- like Clements -- he is really no competition at all. This is a good opportunity to re-examine catchers:

Here are some (adjusted) Win Shares to mull:

Charlie Bennett: 234

Joe D. - Thanks for post #102, quite informative. I'm sure you understood that I wasn't making any kind of serious suggestion.

BTW, Lankford was much better than Landreaux.

" think having the line where we do is about right, throw in 25 Negro Leaguers and you're at 198 players (some of those aren't eligible yet), so a little below 300 WS is about the cutoff, assuming WS were the perfect metric (it isn't)"

I agree with OCF -- really interesting numbers. Except it looks to me that there are 247 players with over 300 WS, which would put the cutoff somewhere above 300 -- not below.

On the other hand, I have seen any number of posts "awarding" "bonuses" for things like being the top 3B on the ballot, or the best eligible player from the '70s, or the best AA player and so on.

Let me be clear again just in case anybody suspects me of doing this, I have never done this. Never have, never will.

At first glance from BP, Joe Kelley looks like a clone of Sam Thompson playing LF. Comparable peak, similar career value over somewhat longer career, more championships. OTOH, Win Shares sees Kelley as clearly the superior player.

I have those two at almost exactly the same value. They are both "close, but no cigar" ballot choices.

Marc,

Either your example of John Olerud is poorly chosen or your theory doesn't hold. Olerud was clearly the best player of the '93 Jays but he didn't put them over the top?

I don't believe he did because I don't believe in this phenomenon but if it wasn't Olerud, who was?

And how do you explain Bonds losing so often? He doesn't have the magic to put is team over the top? I don't get it.

1908 prelim

Paul,

Thanks for the info on Pud Galvin's 1878 season! Pretty impressive, he was. . .

Was the number of innings that he threw that year a real outlier in IP, or were some of the top pitchers of the 1870s in the NA/NL throwing an additional 200-300 innings in non-league games at that time? I had been assuming that 700 innings pitched was more or less a physical maximum during this decade, but Galvin's 1878 season shows that my assumption was mistaken in at least one case. So now I'm wondering how typical this sort of workload was?

"That was a running total. . ."

Thanks for the clarification. I understand now.

John Murphy wrote: "Let me be clear again just in case anybody suspects me of doing this, I have never done this. Never have, never will."

Same for me. In some cases it's a fine line between (1) moving a player from low on the ballot to high because another player at the same position (or from a pre-documented era) was elected or has dropped lower on the ballot and (2) strategic voting.

I'm certainly not accusing anyone. Just want us to be mindful.

Seconding Joes's post #123, having played fast pitch softball (a long time ago), I once watched a guy pitch 36 innings on the 2nd day of a 2-day double elimination tournament after pitching 2 CG the day before. His team won the tournament.

Joe,

I see that line of reasoning, but I have difficulty reconciling it with the fact that we don't see any pitchers from the 1870s who were successful for more than a few years -- Spalding is the most durable, and he ended his career as a first baseman. Maybe the injury that forced him to give up pitching was not related to pitching per se (those who know the history better than I do probably know whether it was or was not), but if pitching didn't wear out the pitcher's arm prior to the overhand rules, why aren't there more long-career pitchers? The only pitcher who played 1871-1879 (as far as I know) was Bobby Mathews, and he was out of the majors (I guess) for most of 1877 and all of 1878.

>Posted 7:37 a.m., August 21, 2003 (#116) - Philip

"By the same reasoning, Derek Jeter is a worthless player."

Not worthless. Just worth less than one would assume if you assumed "replacement level" was the proper benchmark.

Let's try a thought experiment.

You are a GM. Player X, just out of college, comes to you and says, "I am a third baseman. I can guarantee you that, if you sign me to a 15 year contract, I will play for your team every day at third, and perform at exactly one Win Below Median (but well above replacement level) for each one of those 15 years. You third base problems will be solved, and you can start working on the rest of your team!" Assume you can bank on his claims. Do you sign him and plug him in at third until 2018?

Well, you think to yourself, I'm a GM of about average intelligence, and my team has been finishing around the middle of the pack. If I'm trying to build a contender, having third baseman who will be just a tad below median could be a significant advantage. But wait! You think. Assume I don't sign him. Assume I check out free agents or independent leagues or wherever -- what can I expect to get at third base if I don't sign him? Let me go to my stat books and see. You check through your records and realize that -- for the past 15 years together, on average -- you have received exactly median performance from the series of men who played third base for you.

Jason says, "The problem with the hypothesis that median value is the appropriate benchmark for assessing value is that it ignores opportunity cost."

The problem is that there is an opportunity cost for signing everyone. The opportunity cost for signing the guy in my hypo is that you don't get the series of third basemen who work out to median value. Why must that opportunity cost be dismissed out of hand when considering the value or Player X?

TPR is a total player rating. They've just set it up such that zero has value.

Except TPR will give a zero rating to a player who had a twenty year career, which equals the rating for myself, yourself, JoeDimino, Jason Korval, Marc, Matt B., etc. :-)

Sure, John. Because like a player with a zero rating, I have done absolutely nothing to push my team above or below average.

Take two players who had exactly the same numbers for ten years. One retires, while the other plays at well below average. TPR is going to paint the second player as well below the player who had retired in their rating, which is ridiculous. At the very least, their value should be the same. A player should not be penalized because a GM is an idiot.

That doesn't mean he should be rewarded with more counting stat numbers either.

Mark,

You would be right about TPR if its creators had never made ranking

The two last comments are right on. There is danger in the rankings. Similar to a list that would be generated by straight Win Shares, with no regards to playing time.

Agreed. Two players can have the same number of WS, but one created his in 100 games, while the other created his in 162 games. That's why I look at WS per 162 games, too.

All that said, I think it's all moot. Does anyone actually use TPR?

Not too many anymore, but years ago they did. I know, because I did then, too. All your arguments were mine in the mid-eighties.

Maybe different people here are using different definitions of "replacement level".

In a perfect world, the 30 best SS's would be starting in the majors today. Because it's not a perfect world, this doesn't always happen because a few of those top 30 guys may wind up as backups to others. But it's probably pretty close, meaning there isn't much difference between the worst guys playing and the best guys sitting.

If the #3 SS goes down, and his team acquires the #7 from another team - that isn't the "replacement level". The team that had had the #7 may then acquire the #22 from another team who may then acquire the #33, who wasn't playing in the majors - now we're getting somewhere. The whole chain must be examined to see the (league-wide) replacement level.

Mark,

Bill James' criticism is 100% right, not 100% wrong. Yes, the zero point is arbitrary. But, as soon as you start doing math with it, or comparing 2 numbers, it starts invalidating results. If the creators of TPR had never used it to sum numbers together over a career, or rank players into lists using it then you would be correct. Of course, why bother at that point.

The bad thing is that James then made the same mistake (of course, he doesnt rank by career Win Shares, so he seems to realize this). WARP, despite its many faults, at least tries to stick the zero point in a place that would allow aggregating the numbers together.

At risk of entering deceased equine abuse territory:

Andrew wrote:

"In a world of completely rational GMs, perfect information, no transaction costs, no breaks on player movement based on salary constraints, no longterm contracts, etc., a team who loses a player would replace him with a replacement level player."

Exactly, for tomorrow's game. If any of the top 30 shortstops are injured, they would be replaced by shortstop #31 -- who would be replacement level. Assuming Andrew's world, though, in the off-season every player will be available (no long term contracts), and the team with the injured shortstop will be at no disadvantage, and by April 1 of the next year, the replacement shortstop will, on average, be #15 or #16, not #30 or #31.

Anyone who subscribes to Prospectus should definitely read the Homepage link above from June of this year: "Lies, Damned Lies: Redefining Replacement Level." For those without, this paragraph summarizes the argument:

" Intuitively speaking, we've reflected the notion that the longer the period of time that a player is in the lineup for, the higher standard we ought to apply in evaluating his performance. A player that performs somewhat below average for a couple of weeks? It's easy to envision a scenario--say as a quick-fix injury replacement--under which that player has value. But a player that performs somewhat below average for a number of seasons consecutively? It's hard to imagine that such a player will accumulate much of that playing time with a championship-caliber team. Over the long run, the number of opportunities to improve upon such a player are many, and the opportunity cost of settling for such mediocrity is high."

For those so inclined, their replacement level formula is:

"PRAR (Progressive Runs Above Replacement) = BR - PA * (.0154 ln(PA) - .1324)"

If you understand what that means, you are a better man than I (my mind went numb right around logarithms.)

I posted my top 25 eligible peaks earlier. Here are the top 25 careers. For my prelim. ballot, just average the two ;-) Don't forget to tell me how it comes out. BTW, the list represents subjectively balancing about 7 different metrics. * means the player did NOT make the top 25 peaks.

1. McPhee* (but "close")

Prelim. Ballot

1. Ezra Sutton (2) - Jumps over Start and McPhee and Galvin.

>That?s ok for some purposes, but for the Hall I prefer subtracting 90, which generally moves the high peak, short career pitchers higher. I don?t mind saying that value is one thing, making progress toward the Hall of Merit another. So in the end I agree with raising the ?replacement level? when evaluating careers for the Hall.

>I kind of agree with Jim. 90 might not be the exactly correct number but it's close enough. A reasonably high replacement level, but nothing like average.

Thank you, thank you. This is what I have been trying to say. When it comes to separating out a HoMer from a guy who is just close, a nice high "replacement value" does wonders for your clarity. Heck, I would use average or even above average. I know you don't agree but you're headed my way!