About Us  |  Analysis  |  Stats  |  Stuff  |  News
Baseball Primer Sporting News Fantasy Baseball - Get in the Game Today!
Baseball Primer Sections

  You are here > Baseball Primer > Articles  > Joe Dimino and Robert Dudek  > Something Better

Something Better
by Joe Dimino and Robert Dudek

I absolutely love baseball history. Always have, as far back as I can remember, it’s been my favorite part of the game. The Hall of Fame has been central to this for me, since the first time I visited the shrine (during the 1983 World Series, my 11th birthday present was the trip, a 5-hour drive from Long Island). As I walked through the Plaque Room back then, I just assumed these were the greatest players of all time, because we were told they were, and writers must know more than an 11-year old.

Two years later, I read my first “Baseball Abstract”. At that point, I started questioning the conventional wisdom, and along with this, I started to wonder about the players in the Hall of Fame. I received my first “McMillan Encyclopedia” for my 16th birthday, and when I saw the numbers of some these guys I started thinking, huh?

I finally was able to read the “Historical Baseball Abstract” during the summer of 1991, and my respect for the Hall of Fame selections dipped some more. Then came the “Politics of Glory” (renamed “Whatever Happened to the Hall of Fame”). I was especially disturbed by the Veterans Committee’s dark periods, 1945 and the early 70’s. I’ve hated Frankie Frisch (the dominant force on the early 70’s Committee) ever since. These were to the two periods where the majority of mistakes and arbitrary decisions came from. Lee Allen on the other hand, who was a large contributor to the Committee in the 1960’s (a period where many of the mistakes of the mid-1940s were corrected) became one of my all-time baseball heroes. I started thinking this guy had the job I was born for. At this point, I was convinced that I needed to come up with something better, even if it was just for my own benefit.

In reading the historical book, I was very impressed with the article about “Honors”. In this article, James talks about how well constructed the MVP ballot is (the voters aren’t great, but the system is pretty solid), and how poorly constructed the Hall of Fame voting system is. The major drawbacks are that there is no way to express degrees in voting, and the election cutoff is arbitrary (75%), not absolute (say 2 per year).

So the Hall of Something came to my mind. I couldn’t think of a good name, and other than some 80’s baseball cards and a Bobby Ramos autographed Tommy John replica glove, I didn’t have much memorabilia to draw visitors to my basement. But I figured, “you’ve gotta start somewhere” and started working on it anyway. I re-worked my way through “Politics” and came up with a list of mistake players. I also came up with a list of unjustly shunned players.

Then I started thinking of going back to 1935 and trying see who should have been elected in each election, only I would take the top players each year based on actual voting, whether or not they garnered 75% of the vote. I started by allowing 5 in for 1935, 4 for 1936-37, 3 in 1938-39, then when the elections were down to every 3 years I’d take the top 9. By the late 1940s, satisfied that the 65 years of baseball history before elections had been made up for, I cut it to two per season (or a multiple of this for years where the elections alternated). I also tweaked my annual “elections” by allowing the Bill James top 100 lists to override the BBWAA vote when appropriate. This list came out better that what is in the Hall presently, but still shunned the 19th Century players and Negro Leaguers.

At that point (October 2000), I stumbled onto “Baseball Reference”, while looking for info about the 1996 Marlins. Then a few months later, I noticed the “Outside the Box” weblog and discovered that I wasn’t the only baseball lunatic out there. I started talking with Robert Dudek, and over several months, we refined these ideas. He came up with a name that made a lot of sense. “Fame” shouldn’t the criteria for selection, “merit” should be. We should label our “shrine” for what it takes to get in, not the reward for getting in. We had a name - the Hall of Merit.

Our basic premise is not that there are too many people in the Hall of Fame, but that there are too many mistakes. Around April 2001 I became aware of the r-s-bb Hall of Fame. I think their concept is excellent, with elections every year, etc. But I felt that it was too exclusive a club. Only 83 players have been enshrined, and it took greats like Willie McCovey over 15 years to get in. Again, I really don’t think there are too many people enshrined, just too many mistakes. Add Ron Santo and Stan Hack and remove George Kell and Freddy Lindstrom, for example, and the Hall looks better. Get rid of Tommy McCarthy and put Deacon White in and we take it up another notch. A few more of these “trades” and all of the sudden, the whole thing starts looking a lot better. Everybody knows that Ruth, Mantle and Mays are the greats. The key in my opinion is honoring the correct people that are just a notch below them as well. The r-s-bb Hall is great as an inner circle, but we feel that it is too exclusive.

By starting over, we can correct the mistakes of Cooperstown’s past.

Over the next several years, we will travel through time, selecting the greatest players in the history of baseball, and learning a good deal in the process. We want to correct the flaws of the Major League Baseball Hall of Fame voting system. We’ve decided to start with elections in 1915. This will put the great players of the 19th Century on the first few ballots, so we ensure that we are electing them as well. 19th Century baseball is largely forgotten, but w/the statistical advances of the last 30 years we can finally evaluate them fairly. The great players of that era deserve to be enshrined, even if they weren’t as great as today’s stars, they have a place in history, and real pennants were won and lost during that time.

We’ve kept the 5-year waiting period in tact, so the first elections will encompass careers ending in 1910 or earlier. The basic structure of the ballot will be an MVP-type vote, where electors will vote for the top 10, in order (ties are allowed on the ballot). We still aren’t sure about the weighting, and we’d like to open that up to the mathematicians out there for debate. We were leaning towards the 14-9-8-7-6-5-4-3-2-1 for the BBWAA, but recently we’ve heard that 14 might be too much for first place, we’re open to ideas there.

As for the number of people to be elected, we’ve run a spreadsheet that takes the “team seasons” into account (we’ve adjusted downward for the early years, it’s not a straight X teams equals X electees), and we want to allow for some “make up” selections, since the first election will encompass 40 years of careers (1871-1910). We are going to start with 5 for the 1915 and 1916 elections, 4 from 1917-19, 3 from 1920-25 and then 2 per year through 1977. At that point, we’ll start upping the number elected, to account for expansion and growth in the population. From 1978-83 we'll alternate 3 per year in even years, 2 in odd years. From 1984-94 we'll elect 3 players per season. Starting in 1995 we'll elect 4 players every 4th year, 3 in the other years. In 2007 we'll start alternating between 3 (even years) and 4 (odd years) players per season. Players will never lose eligibility. Both Robert and I feel that this is crucial: it means that if new information about a player comes to light that player can benefit (as an example, I offer Bill James’ reassessment of Phil Rizzutto based on new evidence of his defensive prowess). It also means that if a voter thinks there were more great players from a certain era than in others, he can vote for a player that might have been squeezed out by his contemporaries in his previous tries. Since the inherent structure of the vote forces the best players to the top of the ballot, there is no reason to remove players from the process artificially. Following this procedure, we'll have 218 honorees after the 2002 ceremony. The current Hall of Fame has 215 members (as players), with a few more coming in 2002.

We’ve thought of putting a positional quota in, with one player at each position (plus 4 pitchers) required in each decade. If a decade’s ballots come up short at a position, in the last year of the decade we’d have two elections, one for the position that is short, and one for everyone else. This is open to debate though.

As far as criteria, numbers aren’t everything: there are things we cannot account for in the numbers. But since we have them, we are going to make them available to help you with your ballots. Players’ contributions on the field are to be the main criteria for selection; off-field actions should only be taken into account for the effect they had on the players’ teams on the field of play. The language may be tweaked so as many people are comfortable with the criteria as possible. We want to make the criteria reasonably broad so that each voter is able to interpret them according to his own tastes.

Robert has done some excellent work on figuring the relative strengths of the leagues each year during the 19th Century, so this will help us to distinguish the greats as well. He’s also figured positional replacement levels from 1871-1919. Jim Furtado is going to have Offensive XWins going back to 1900 pretty soon.

We will present numbers to show players’ contributions in the proper context. We hope these tools will help you with your decisions. A page will be set up for each player, and we’ll post things like adjusted offensive wins and losses, Win Shares Gold Gloves, TPR (although I think the defensive part of TPR is useless, we’ll put the data there), etc. The goal is to make as much information as possible available. If any of you have your own stats that you think would be good to add (Hoyts, let’s say), let us know, and we’ll post those too. We’ll adjust things like Win Shares, TPR, etc. for the shorter seasons in the 1870’s and 80’s, strike years, etc., so everything is on a 162 game scale. There will be a Hall of Merit Weblog set up as well, where we’ll be able to discuss the ballot, the process, lobby for players, etc. We want to spend the next few weeks discussing these aspects of the project with the Primer readership.

As far as the first election is concerned, we want to wait for the Win Shares Book to come out in April. There are several reasons for this. I personally think the defensive analysis we get from Win Shares will be light years ahead of anything we’ve had previously. Although Win Shares pegs the replacement level too low (zero) this can be adjusted for. Also, since this information will be available, it’d be terrible to make mistakes that could have been avoided. Waiting a few months is a little frustrating, but it will also mean a more solid foundation for the first election, so why not wait? The goal should be to have as much information available as possible. This is especially important for the earlier elections, where we have the least amount of information.

Another reason for waiting is that the first election will encompass 40 years and a huge number of candidates. After the first election, a much smaller number will be added to the pool for subsequent elections, which will be much easier to handle.

We’ll set up a Plaque Room as well, where people can go and we’ll have all sorts of things there: pictures (depending on copyrights), stats, links to purchase books about the player, etc.

We look forward to your comments, suggestions, and ballots. The Hall of Merit weblog should be up shortly. In the meantime, feel free to comment here or drop Robert and I an email. We want to emphasize that this will be a Primer community project.

This is going to be fun.


Reader Comments and Retorts
Post a comment | E-mail us directly | Report an inappropriate comment

December 11, 2001 - Robert Dudek

I just want to clarify one point.

The by-line says Joe Dimino and Robert Dudek. The truth is that Joe wrote the piece - I threw in a few suggestions and helped edit it. That's why the article starts with "I" instead of "We".

December 11, 2001 - Ryan Wilkins (www)

I can't wait, fellas. However, does this mean a death to MostlyBaseball? Will any new content be present on the site from now on, or will everything be on 'Primer'? Just curious.

-RW

December 11, 2001 - Robert Dudek

Ryan...

In answer to your question, that has yet to be determined but we will most likely be posting new material at www.mostlybaseball.com very soon and, with Sean and Jim's permission, linking to it from here.

Joe and I would like to do a review of the players who retired during and after the 2001 season. We'll probably start looking at some of the less heralded ones soon, like Stan Javier, Eric Davis (remember when he was a baseball god?) and Tony Fernandez.

December 12, 2001 - Craig Burley

Yes, this _is_ going to be fun.

By the way, a note about Win Shares : having gone through the book more carefully, I really do believe that, based on some of the comments I've seen, the WS system does use a replacement level of the "marginal" level, and not a "zero" level. I do think James has some sort of normalization that brings negative figures up to zero... it's hard to tell. I have little to go on (and the comment that a zero-level WS is someone who "can't play" gives me pause) but a general feeling, though.

And yes, I agree about the Win Shares treatment of defensive numbers being advanced (I actually think that the way James is analyzing numbers is similar to how Diamond Mind do their defensive analysis.. does anyone know how DMB arrive at their ratings?) But we should certainly begin discussing the candidates before the book comes out... there are a LOT of candidates to sort through who retired before 1910.

This will be a great project, great for the website, and a huge time-sink for a lot of us!

December 12, 2001 - Jason Koral

Craig,

Judging from the New Historical Abstract, basically anyone who plays at all seems to get at least one win share. For example, in the 1962 Mets breakdown, Marv Throneberry gets about five win shares; and even guys with only a handful of at bats get one. So it appears that if you play in the major leagues for a team that wins a positive amount of games you will receive positive win shares, regardless whether you are any good or not.

December 12, 2001 - Jason Koral

Craig,

Judging from the New Historical Abstract, basically anyone who plays at all seems to get at least one win share. For example, in the 1962 Mets breakdown, Marv Throneberry gets about five win shares; and even guys with only a handful of at bats get one. So it appears that if you play in the major leagues for a team that wins a positive amount of games you will receive positive win shares, regardless of whether you are any good or not.

December 12, 2001 - Cameron

Gents, a query-

What about managers? Will they be enshrined? If not, will the managerial contributions of good but not great players like Torre and Fred Clarke (though admittdly Clarke was a P-M and most of his comps are HOFers) count in their favour? Looking forward to it fellas.

December 12, 2001 - Cameron

Gents, a query-

What about managers? Will they be enshrined? If not, will the managerial contributions of good but not great players like Torre and Fred Clarke (though admittdly Clarke was a P-M and most of his comps are HOFers) count in their favour? Looking forward to it fellas.

December 12, 2001 - Carl Goetz

Craig, If you go to DiamondMind's site, they have a complete article about their defensive ratings. www.diamondmind.com

December 12, 2001 - Carl Goetz

Where will the ballots and player analysis be posted? On this site? I want to make sure I'm in on this from the beginning.

December 12, 2001 - Robert Dudek

Our idea is to distribute the ballots via e-mail. They will be made available to anyone who wishes to vote. We already have a list of interested voters which numbers in the several dozens.

The elections will probably not take place for awhile, but info and articles will start appearing shortly.

We would like to make the ballots public and we will ask every voter to state their reasoning behind their selections.

December 12, 2001 - Steve Rohde

Joe and Robert,

This is an exciting project. I will look forward to participating in the discussion and voting.

Craig,

Since James has set a win share equal to 1/3 of a win, and since he has indicated that in every case the number of wins attributed to each player totals the number of wins that the team actually won, it seems to me that win shares has to be a raw number, rather than a number in excess of a replacement level, or as Joe has put it, the replacement level in effect is pegged at zero. While conceptually I have no issue with the concept of win shares as Jamese has defined it, and there is a certain elegance to it, in my opinion it is quite important, in evaluating players and comparing them with each other to take the extra step of figuring win shares in excess of replacement level. We should probably have some discussion of just what level to set the replacement level at, and precisely how to do it in the context of the win shares system.

By the way, does anybody know why James decided to set a win share as equal to 1/3 of a win. Was it simply that he wanted to report win shares without using decimals, and he needed more differentiation in the numbers allocated to different players? I would have preferred setting each win share as equal to one win, perhaps using a different term such as "wins created", and simply report the results to one decimal place, as TPR does. Or am I missing something basic to the concept of win shares?

December 12, 2001 - Carl Goetz

How can I get on the emailing list?

December 12, 2001 - Joe Dimino (e-mail)

To get on the list, send an email to the address above this post it is different than the email address I normally use when posting on this site. I should have put that in the article, but I wasn't sure how it would come out when published.

Robert has done some great work at pegging the replacement level for 19th century players and that will be published soon. He's also worked on figuring the comparitive strength of the leagues (such as 1876 NL v. 1882 NL) through 1900 at least, maybe 1910, I'm not sure how far he's come.

Steve, James uses 1/3 WS because, "it works". He thinks using single WS would not distinguish enough. That is an 8 WS would encompass too wide a range. Allowing him to use 23, 24 or 25 to describe "8" WS gives the system more flexibility. But on the same token, he thinks decimals are too precise, they imply a level or precision that does not exist. He's confident a 25 is slightly better than a 24, but not confident that 8.4 is better than 8.3, so a "25" makes them even. He also said that he could have used 1/2 or 1/4 as the scale and those would work, but whole numbers and decimals don't.

There is a big advantage to 1/3's, and I don't even know if he realized it when he chose 1/3's, but the pitcher WS tend to come out equal to wins. That's because one of the inherent assumptions is that on a normal team that does everything equally well, the offense gets 50% of the WS, the defense gets 16.25% and the pitchers 37.5%. Since a WS is 33% of a win, the pitcher WS are instantly meaningful, from an intuitive sense.

December 12, 2001 - Craig Burley

Jason: You don't have to be any good to be above the 'marginal' level... half the league average in runs scored, or 1.5 times the league average in runs allowed.

Carl: When you read the "Evaluating Defense" article, it goes into tantalizingly few details about how their rating system actually works. Tippett does talk about the need for individual defense ratings to "add up" in a team context, which leads me to believe that they use a system that may look similar to WS, but perhaps without James' adjustments for false normalization.

Steve: I assume that the WS calibration was set at 1/3 of wins, in order to make the number as equivalent as possible (over a career) to wins by a starting pitcher. Since a pitcher's contribution is ballparked by James at 32.25% of wins, (67.5% times the 50% of winning that is preventing runs) a pitcher's "win" is equivalent to about 1/3 of a win. Setting up the equivalence this way allows at a glance to judge a season or career by an already-established metric... we know what a 10-win season by a starter means, we know what a 25-win season by a starter would mean, and we know what 150 and 300 career wins "mean", roughly speaking.

Second point: simply because WS add to the raw total doesn't mean there isn't a replacement level calculation done; the calculation could take place internally in the figuring of the statistic rather than externally in discounting the "raw" statistic to replacement level. More simply: you can use a player's contribution _in excess_ of a base value (in James's case, I think, it's the "marginal" value) and use that to apportion team wins amongst the players. So instead of figuring each player's raw runs created, and apportioning the offensive win shares according to each player's percentage of runs created, you use runs created above marginal and use that to apportion the win shares.

An example from the book will show what I am saying. In the '69 Phillies example, Dick Allen has 22 WS and Cookie Rojas, playing next to him in the infield, has 3. So Dick Allen has seven times the win shares of Rojas.

All things being equal, we know how unlikely it is that the "raw" defensive contribution of a 2B playing 100 games can't possibly be much less than the "raw" defensive contribution of a 1B playing 118 games, no matter how you slice it. So if James is using raw figures, then Dick Allen must have created seven times as many runs as Cookie Rojas (he has about 496 PA and made 329 outs, Cookie had about 426 PA and made 330 outs. I don't see any adjustment necessary for the opportunity factor, as they made practically the same number of outs.)

Using basic runs created, Allen created 97 runs. Rojas created 31 runs.

That's only 3 times as many, which doesn't get Allen to seven times the contribution. There's a further problem, as well, which is that the defensive contribution needs to be subtracted from the 22 and 3 WS respectively. (Even if both had just one WS of defensive contribution, Allen has then TEN times more contribution with the bat). And Cookie's raw numbers aren't bad defensively; it would take a _big_ adjustment drop to take his defensive value down to zero.

But if we adjust those raw totals, and use a "replacement" level of marginal runs (as James seems to advocate) then the numbers can come out correctly.

Anyway, I hope that's clear. These numbers don't make any sense unless they are calculated using some sort of replacement level... the variations are much too large.

By the way, the NL average RS was 4.05 in 1969, making the marginal offensive rate about 2 runs per 27 outs. Rojas created about 2.5 runs per 27, making his contribution above marginal 0.5. Allen created about 7.9 runs per 27 outs, making his contribution above marginal 5.9, or twelve times as many runs above marginal than The Cookster, leaving lots of room for a figure for defensive contributions which makes sense.

December 12, 2001 - Joe Dimino

That should say pitchers get 37.75%, oops.

December 12, 2001 - Joe Dimino

That should say pitchers get 37.75%, oops.

December 12, 2001 - Joe Dimino (e-mail)

I'm pretty certain Diamond Mind uses play-by-play data for their ratings. From what I understand, Tom Ruane works with them or has worked with them, if he's reading, maybe he can expand on this.

Sorry for the double post earlier.

December 12, 2001 - Aaron Schatz

How will Negro Leaguers be handled? Will we able to elect great minor league players (such as Buzz Arlett) who played in the pre-farm system, independent minors?

Aaron

December 12, 2001 - Devin McCullen

As far as managers go, here's my suggestion, based on your format. Have seperate elections for managers, maybe every 5 "years" or so, and only a person's managerial record would be considered (this works in parallel with your decision to limit criteria for players for what they did on the field.) I don't think that there's any reason NOT to include them. I think most voters would feel comfortable with their ability to judge managerial skill.

I think that this sort of thing COULD be extended to other areas, and it might be worth considering an election category for exceutives/GMs/comissioners/labor leaders/whatnot, because the Hall of Fame has made mistakes in these areas as well (Morgan Bulkley, anyone?)

December 12, 2001 - Joe Dimino (e-mail)

Negro Leaguers will be considered with all of the other players, there won't be a special committee or anything like that. If we see people excluding obvious Negro Leaguers from their ballots then we might have to adjust the process, but I don't see that happening, we have a pretty knowledgable group here :-)

As for managers, that's tougher. We could maybe set up a Manager's Committee, or better yet a seperate ballot, that elects 1 every 5 years? I'm open to suggestions there. I think a few managers from each era would be a great idea.

Minor Leaguers are much tougher. There is little data, and even most hard core people don't know a lot about these guys, other than what Bill James has written, unless they've taken a particular interest and followed through on it. Maybe we could set up a seperate "wing" for them. I could definitely be underestimating the knowledge level.

I wouldn't be comfortable with running that operation due to my lack of knowledge, but I'd love it if someone would volunteer to run that "wing", provide relevant, strong information so we can vote knowledgably, etc. I think that would be great. If someone wants to volunteer to run that end, leave a post here or send me an email.

December 12, 2001 - Joe Dimino (e-mail)

Devin, I hadn't seen your post, when I posted.

That's pretty funny, that we both came up with that idea for manager elections at the same time. Great minds must think alike!

December 12, 2001 - Steve Rohde

Craig,

Maybe you are on to something. I don't know. But if replacement level is somehow figured in, I don't understand the logic of the system. It seems that every person who has any kind of meaningful playing time in the data presented begins to accumulate win shares, even if he is performing at a level well below what we would normally consider replacement level. For example for the 1973 Texas Rangers, Pete Broberg picked up 2 win shares by pitching 119 innings with an ERA of 5.61, in a league when the average ERA was 3.82. For the 1962 New York Mets, Gus Bell picked up a win share with a batting average of .149 in 101 at bats. Bell had an OBP of .225 and a slugging average of .188. Maybe his win share was for defense, but clearly Bell was well below a replacement level performer. In fact, for the 1962 Mets, the average performer was probably below what we have traditionally considered replacement level performance.

December 12, 2001 - Jordan

>> But if replacement level is somehow figured in, I don't understand the logic of the system. It seems that every person who has any kind of meaningful playing time in the data presented begins to accumulate win shares, even if he is performing at a level well below what we would normally consider replacement level. >>

I think that makes a lot of sense. A replacement level player isn't worth zero. In fact, no player is worth zero, unless they never do anything positive. A player with a 10.00 ERA is better than a player who doesn't ever get anyone out, just like a .100/.150/.150 hitter is better than a .000/.000/.000 hitter.

Even the worst player on the worst team contributes something to winning.

December 12, 2001 - Joe Dimino (e-mail)

Jordan,

What we (I say we meaning those of us that peg replacement as zero) are saying is that a replacement level player has zero value because you can pluck a guy from AAA or Japan or Mexico or the Devil Rays that can do the exact same thing. So what a player does up to that level is essentially not worth anything, because you can find "anyone" to do it. You can pull an even trade and not give up anything of value for a replacement that can accomplish the same thing. The player performing at replacement level does not bring any added value to the team, so he is easily replacable.

I agree that everyone on the team contributes, but when you set replacement level at zero, you are only giving credit for the extra value a player brings. A player above the replacement level can't just be replaced by someone from AAA or Baltimore without the team losing something, so that is where his true value begins.

December 12, 2001 - Joe Dimino (e-mail)

One more thing Jordan . . . a player below the replacement level really does hurt the team, because he could be replaced by someone off the proverbial scrap heap that is better.

December 13, 2001 - Curious

Will Joe Jackson and Pete Rose be eligible?

December 13, 2001 - Joe Dimino (e-mail)

Good point curious. I can't believe it took a day for the Rose topic to surface.

I felt like Rose and Jackson would be eligible. There is a rough (subject to discussion/change) layout of the rules on www.MostlyBaseball.com. In there, we state that a player's performance on the field is what should be considered, and his off field performance should only be considered if it affected his team(s) on the field.

I realize I'm in the minority on the Rose issue. I think even if he did bet (which we don't know for sure), he's been punished long enough. I also think the facts are cloudy on Jackson at best.

I imagine there will be a strong enough block against to keep them out even if they are eligible, but it was my intent that they would be eligible. I'm not sure what Robert thinks, I really can't remember discussing it with him.

I hate opening this topic up, because all discussion inevitably shifts there, like your eyes when you drive by a car wreck. Once the blog is set up (I think Dec. 19 is the goal), we'll set up a thread to discuss this topic. Can this wait until then?

December 13, 2001 - Carl Goetz

If we're basing this entirely on on-field performance, I think they both should be eligible. I'll probably give Jackson a 1st place vote the 1st election he is eligible.

December 13, 2001 - Eric Enders

I agree that both Jackson and Rose should be eligible.

However, it could be argued that their off-the field activities did influence their team's performace on the field. Indeed, that possibility is the very reason that both were banned in the first place.

Each voter will have to weigh this issue carefully on their ballot. How great can a player really be if he tries to lose the most important games?

Anyway, I'm not losing much sleep over it either way. Sometimes I just wish Rose and Jackson had never existed so I wouldn't have to listen to all the inane drivel that people spew forth about them.

December 13, 2001 - Robert Dudek

The following are my own opinions only, not Hall of Merit policy statements:

1) Managers should have their own wing, but I think we should wait until the Players wing is up and running before we start on that project.

2) In principle, every player should be eligible. This should include Shoeless Joe and Pete Rose, however I think a voter ought to have the right to leave a "disgraced" player off his/her ballot if they so choose and if they are ready to justify their decision.

3) Negro leaguers and minor leaguers are eligible. The candidacy of every player depends a lot of the quality of objective information available about their performance (especially the context). I would not feel comfortable putting a player I knew almost nothing about on my ballot. It would be a great help if anyone who had access to info, especially numbers, to post it on the site (i.e. the eventual Hall of Merit section of Baseball Primer). But I also respect the right of voters to use more subjective criteria if "objective" information is lacking.

4) In this regard, any studies which estimate the "playing strength" of minor leagues and negro leagues would be an invaluable aid in assessing the merits of players who were not able to play in the "majors" for a significant period of time or indeed at all.

December 14, 2001 - Jordan

>> One more thing Jordan . . . a player below the replacement level really does hurt the team, because he could be replaced by someone off the proverbial scrap heap that is better. >>

I certainly understand that, but I think it's non-trivial to determine what exactly a replacement player is at a given time, and more importantly, how a replacement player from say, the 1920s, compares to a replacement player from the 1990s.

As an aside, I think the availability of these replacement players is exaggerated. It's not as if you could always find players in your own organization that would be as good as replacement level, or get them from other organizations without giving something up.

I don't think Win Shares are particularly good for predictions, but I do think they are excellent for showing who really contributed the most towards winning on a specific team in a specific time period. They're good marking points for stuff like team MVP awards.

December 14, 2001 - jdw (e-mail)

I see a number of people ask if it's possible to get zero win shares while playing a reasonably significant amount of playing time. My recollection is that Tom Seaver's horrific 1982 season (111.1 IP of 5.50 ERA in a 3.72) was listed as 0 win shares. Someone with the book handy can confirm this. I suspect there are some others that are listed, but that's one that caught my eye.

You have to play quite awful to pull a 0 (or a 1 or a 2) with 111.1 IP. I don't know how many of the other pitchers list pulled 68 ERA+ with 100+ IP in a season. Matty pulled a 71 ERA+ in 1915 with nearly 200 IP. Spahn pulled a 67 in 173.2 IP in 1964. One can work they way down the list of 300 winners and see how their truly wretched seasons do in WS. Maybe even do a little chart on IP, WS and ERA+ to see if there's some linkage and threshold.

John

December 14, 2001 - Joe Dimino (e-mail)

jdw-

Seaver was a zero in 1982. Spahn's 1964 was a 1 and Matty's 1915 was actually a 4.

Ed Brandt has 3 WS to show for his efforts in 1935, which included 5-19, 76 ERA+ over 174.667 IP, the Braves 50-103 mark didn't hurt.

Lefty Grove in 1934 was 8-8, but with a 74 ERA+ in 109.3 IP and he picked up 2 WS.

Dennis Eckersley's 1983 was classic, 176.3 IP, 9-13, ERA+ 78, 3 WS.

Greg Maddux 1987, 155.7 IP, ERA+ 77, 6-14, 2 WS.

Steve Carlton 1986 and 87 - 9-14, 176.3 IP, ERA+ 78, 4 WS; 6-14, 152 IP, ERA+ 80, 4 WS.

I wish I had Mike Parrott's 1980 season 1-16, 7.28 ERA (57 ERA+), only 94 IP though. That HAS to be a zero. Might be the worst year any pitcher has had in the last two decades. 7.28.

Could anyone forget Don August's 1989? 12-12 and just 2 WS, in 142.3 IP. 73 ERA+. What was his Run Support?! 1.67 WHIP in County Stadium in a pitcher's year (DH league, 3.88 ERA) and he goes 12-12.

Still couldn't find anyone to rival Seaver 1982 with 0 WS in more innings. I'm sure there's someone, but we'll have to wait for the book. Craig Anderson of the 1962 Mets had 1 WS in 131.3 IP. He was 3-17 with an ERA+ of 78. Craig Minetto had 1 WS for the 1979 A's, he was 1-5, with an ERA+ of 73 in 118.3 IP.

December 15, 2001 - Steve Rohde

I am still intrigued by Gus Bell picking up a win share for the 1962 Mets, with a .149 batting average in 101 at bats, a .225 OBP and, according to Total Baseball, a .198 slugging average. Maybe he picked up zero win shares for offense and one for defense, but certainly, as a total package, he was way, way below replacement level.

December 15, 2001 - Craig Burley

Steve, don't forget that James's "replacement level" is really really low... the "marginal rate" is a player who contributes at a level of one-half the league average on offense or 50% over the league average on defense. Gus Bell looks to me to have been performing at a bit under the marginal rate.

I also suspect that negative win shares on either offense, defense, or the mound are arbitraily "zeroed" and then redistributed amongst teammates.

By the way, since the zero-win-share level (the marginal rate) is 50% over the league average defensively, a raw, context-independent estimate of the zero-level for a pitcher is an ERA+ of 67.

December 15, 2001 - Steve Rohde

Craig,

What is your opinion of my other example? Pete Broberg pitched 119 innings for the Texas Rangers in 1973 with an ERA+ of 66, and still picked up 2 win shares.

December 16, 2001 - Craig Burley

Steve,

The WS "zero-line" of an ERA+ of 67 is only a guideline. Broberg's 2 WS with 118 2/3 innings and an ERA+ of 67 could be explained in any of a few possible ways.

First, and most likely, the '73 Rangers were a poor defensive club. I would offer this as a conjecture, but I really don't think much conjecture is needed. Jim Mason and Toby Harrah (neither of whom was very good defensively) split the year at short, Jeff Burroughs played in right, a converted infielder (Vic Harris) took most of the centrefield duties. Rico Carty, in his last extended PT in the field, played 53 games in the outfield.

If the Rangers are poor defensively, they take "credit" for some of Broberg's poor ERA, therefore moving his own contribution off the zero-line. This is by far the most likely thing to have happened.

(Of course, whether the Rangers' defense should necessarily take much credit for Broberg getting bombed (given his perhipherals, which are pretty awful) is another story.)

There are other possibilities. James may be using different park factors, which could move the figures a little bit... BR.com has his ERA+ at 67, Total Baseball at 66. The other possibility is that Broberg might have given up an unusually low number of unearned runs. I don't know if that's true; I don't have RA data from '73 on hand. Finally, Broberg may get a boost from situational data.

December 17, 2001 - Steve Rohde

Craig,

Sounds plausible. Assuming you are right about James's methodology, the replacement level he set may still be too low to use raw win shares to compare and evaluate players with different playing times. What do you think?

December 17, 2001 - Craig

Steve, I agree absolutely... the true "replacement level", it seems, has always been higher than the marginal rate, at least in the 20th century (since there are few if any players who perform for any length of time at the marginal rate without being themselves immediately replaced).

But I'm wondering if James isn't doing this from a different tack. (Forgive me, everybody, I'm winding up for a long one here. I need a blog o' my own...) Naturally, we wouldn't expect a replacement-level team to win 0 games... as I remember saying in another thread, I'm pretty sure that I could take a bad AAA team like last year's Ottawa Lynx, and they would win at least 27 out of 162 games, one in 6. You could sign nearly every one of those guys for a minimum deal, and they're pretty much all replacement players, a few above that level, a few a bit below. According to a "replacement-value" calculation, they should have a value of 0.

That doesn't mean that the team would get 0 wins. Which, if you take it too literally, a team with 0 win shares would do. (Now, a word of caution: that won't happen. A team with 0 win shares could probably win as many as 12% of its games... going 19-143 and obliterating the Cleveland Spiders from the record book on the way). (That calculation uses the Pythagorean calculation and an exponent of 1.81, using the marginal offense and defense rates of 0.5*league average RS and 1.5*league average RA).

This doesn't mean that James' marginal runs value calculation is out of whack. I don't think it is out of whack at all for normal teams... in other words, for 99.9% of all teams in MLB history. At extreme levels, it does tend to create aberrational data... most statistical evaluations do.

What James is getting at, I think, and this is why it was such a horrible thing to do this book first instead of the WS book, is that he's actually trying to redefine value from the ways we've seen it done before. I don't actually think that he makes this clear, but I'm quite sure he doesn't want to use the traditional "replacement value" as a barmoeter for value. James wants to say that nearly everyone has value, in that nearly everyone contributes something to winning. (Clearly, even someone with 0 win shares is infintely better than no player at all... as Stengel said about catchers, if you don't have one the ball just rolls to the backstop after every pitch.)

This makes some sense to me. The problem I have with using the "replacement level" as the barometer of value is that it's a little high. As someone on another thread said, when you're a White Sox fan and have seen two years of Mike Caruso starting, you know that the true "replacment player" is a hell of a lot worse than we often think. James wants to peg the "replacement level" player as (in a phrase I recall from the book but cannot find) "a player who can't play" as opposed to a guy who can play a bit. It's a matter of taste.

Essentially, the difference is summed up in the following : traditional "RP" analysis answers the question of whether the player has economic use-value; WS analysis asks what the player contributed to winning games.

December 17, 2001 - Steve Rohde

Craig,

Interesting analysis. I think that this whole issue needs a lot more research, discussion and debate. I think different levels could be used in different types of analyses. I think there are even some uses where Palmer's approach of comparing to a League average player may be useful. For example, if a team in the off-season, has a chance to not re-sign a free agent but use the money saved to sign a league average player, comparing a player's performance to league average could be useful.

December 19, 2001 - Joe Dimino (e-mail)

Steve, in a rating system, I don't think comparing a player to the league average is really of any use, at least no more use than comparing a player to 1.2 or .86 or whatever of the league average. It's a just a number of no real significance.

You'd be better off comapring the player to the league average above the replacement level, or anything relative to the replacement level IMO. What you need to do in the situation you propose, is look at the available options, and compare each option in cost and performance above the replacement level. This is really all you need, as it covers every possible situation.

Player X 3.3 Player Y 4.5 Average 3.2 Replacement level 0

If you compare Player X and Player Y to the average, you'd say that Player Y is 13 times more valuable (+.1 vs. +1.3). This is inaccurate. If I could have Player X for $1 million or Player Y for $13 million, I'd take Player X, because it frees up more resources to help the team.

If you compare Player X and Player Y to the replacement level, you'd say that Player Y is 36% better. And you'd be exactly correct. Player Y will produce at a rate 36% better than player Y. If you are going to have to pay Player X $9.5 million and player Y $13 million, you've got a much tougher decision, and really, you could flip a coin, because both players are being paid the same, relative to their contribution.

December 19, 2001 - Steve Rohde

Joe,

I completely agree with you that in rating players, comparison to the replacement level is the most appropriate approach. My comment was perhaps oversimplistic, but what I was thinking was, for example that if you had an existing player and you were trying to decode whether to re-sign him, and you believe that the money could be used instead and that you could probably get an average player for it, just knowing whether the the player you weredeciding on was above or below average would be useful. And in thinking about the overall strengths and weaknesses of a team, knowing who is below or above average is useful in thinking about where it might be easiest to upgrade.

But I agree that the approach you suggested to determining real value is clearly better and more accurate.

December 21, 2001 - Dan Schmidt

Just thinking out loud here, but shouldn't it be possible to use a similar system to dole out Loss Shares to individual players from the team's loss total?

So maybe that sucky guy who still has 2 Win Shares actually has 8 Loss Shares as well, and now it doesn't seem so weird that he has a positive number of Win Shares.

December 26, 2001 - David J. Wheeler (e-mail)

First off, let me express my eager anticipation to finally see Pete Rose on a hall of fame list. And second, have you guys considered sending out a weekly or monthly newsletter about elections and the election process? If so, put me on it!!

December 27, 2001 - jimd

Definitely sounds like an interesting project.

I'm curious to see how the change in voting system works out. That aspect would be most interesting though if the balloting began in 1936, which would make it more directly comparable to the actual BBWAA results.

The proposed MVP-style ballot sets up a directly competitive process that should work well when there are obvious candidates, but may produce some surprises when there are no strong candidates on the ballot. (How profound, just like the MVP awards.) The inability of the electorate to say "There are NOT ENOUGH qualified candidates on this ballot" may result in a clunker or two (relatively speaking) being elected.

December 28, 2001 - Joe Dimino (e-mail)

Jim, I don't think that will be a problem (clunkers). I ran through a few "mock" elections while I was developing this whole idea, and there really was never a shortage of candidates.

I'm pretty confident there are definitely 200+ qualified players out there. The problem with the current system is all of the mistakes. If we could trade out the 20-30 worst Hall of Famers for the 20-30 most qualified non-Hall of Famers, we wouldn't be talking about how there are too many in there.

I think it's very important to go a notch below the Mantle's and Mays's, but to GET IT RIGHT when you do. For one, what would a Hall of Fame of Mantle's and Mays's tell us that we don't already know? We all know who those guys are, and in our heads they are in an inner circle already. The more inclusive the honor is (within reason, and I think ~200 is reasonable), the more careers we get to look back on remember when we visit the institution.

But by honoring the Bobby Grich's and Jimmy Wynn's, instead of the Kiki Cuyler's and Bill Mazeroski's, we'll end up with a much better appreciation of who the truly valuable players of a generation were, after the household names. So when the average fan peeks in, he'll say, "wow, I didn't realize (insert player) was that good." I can't overstate how important I think this is.

January 29, 2002 - Richard Salvato

While we consider "greatness" a requirement for candidates for the Hall of Fame let's not forget that the soul of baseball is the lifetime .273 hitter who also played his position with intelligence and physical competence. Which is another way of saying that the Hall of Fame is highly overrated. Robin Ventura, for instance, will not make it into the Hall but he is enshrined in the hearts of Mets' fans for his grand-slam single against the Braves in game five of the 1999 NLCS, which also goes into my Book of Great New York Baseball Joyous Moments. Memory is really the best Hall of Fame.

July 25, 2002 - Tester

Just a test.

 

You want baseball news?  We've got baseball news!
Archive

Joe Dimino Archive

Bi-Weekly Review: A.L. East
 The A.L. East through April 24th.

New York Yankees
 Will the Yanks win their first 21st century championship? Yes, Virginia, there was no Year 0.

Archive

Bi-Weekly Review: A.L. Central
 The A.L. Central through July 1st.

Bi-Weekly Review: A.L. East
 The A.L. East through July 1st.

Protection: Fact or Fiction?
 Dylan takes a crack at one of the most hotly contested stathead/non-stathead debates.

July 1, 2003
 Stand up for hot dogs!

Bi-Weekly Review: N.L. East
 The N.L. East through June 27th.

Bi-Weekly Review: N.L. West
 The N.L. West through June 26th.





About Baseball Primer | Write for Us | Copyright © 2001-2003 Baseball Think Factory
User Comments, Suggestions, or Complaints | Privacy Policy | The Fine Print | Advertising