The Expansion Era Ballot

November 18, 2013

I.  The Harmonies

                Let me take one step closer toward fully embracing the Hall of Fame standard toward which I have been working in recent years, and let me begin with this argument, which is rather a trite argument, but necessary to us as a point of departure.   Most of you recognize, I suspect, the inherent flaw in the concept of IQ, which is that it attempts to measure in one dimension something which in reality has many different dimensions.    Suppose, for example, that one were to try to summarize all persons’ bodies on a scale in which an average body was 100, the best bodies were at 150 or above—five standard deviations above the norm—and the worst bodies were at 50.    It is not absolutely impossible to do this, but at some point it presents us with unsolvable problems.    What exactly is a "good" body?   A person could be extremely attractive, but also terminally ill.   Is that a good body, in that it is attractive, or a bad body, in that it is near death?    One would have to measure strength, agility, quickness, balance; one would have to score 80-year-olds and 15-year-olds on a common scale; one would have to score men and women on a common scale.   Even to measure "strength"—a small part of the whole—is impossibly complex, in that we have hundreds of different muscles.   A person can be immensely strong in some muscles, weak in others.   One can be strong but fat; one can be lean but weak.   "Health" is more complex than "strength", and "health", again, is merely a small part of the whole.  

                And yet, this is not to say that we cannot do it or that we do not do it.   If you say that your grandmother is in great shape for 85, everybody knows what that means.    If you say that your blind date was a 7, everybody knows what that means.   Last spring, in spring training, I asked our farm director, about one of our minor leaguers, "is he out of shape, or is he just a bad-body guy?"   Everybody knows what that means, in general terms.

                Most of you recognize, I suspect, that the concept of "IQ" has the same issues:  What, exactly, is a "good" mind?   The mind, like the body, has many, many dimensions.   A person can be brilliant but crazy.   A person can be both brilliant and sane, but cruel.  A person who seems dull can nonetheless embody a positive philosophy that will contribute immensely to their mental health and their success in life, and that makes them a joy to be around.    A person can be brilliant at math but unable to carry on a normal conversation.   People who have fantastic interpersonal skills are, many times, irrational people who make bad life decisions as a matter of course.     Some elements of intelligence can be refined through education; others cannot.   Do we treat these the same?

                The concept of "IQ" is, on a certain level, silly.   We accept it to the extent that we do simply because we are unable to visualize its failings.   If, in the third grade, we were all assigned a "BQ"—a "Body Quotient"—society would reject that, because we would immediately SEE the failings and limitations of the concept.     We accept the one-dimensional measurement of IQ to the extent that we do only because that which it attempts to measure is elusive if not entirely invisible; therefore, we are unable to see immediately the failures and limitations of the measurement.      This is not to say that there is no value in the effort.    We understand in general terms what it means to be intelligent, what it means to be stupid.

                Such it is with Win Shares, WAR, and all other Total Player Ratings.    It is not that there is no value in the effort, but we should never forget that what we are trying to do here is, in the end, impossible.   We are trying to state all contributions to a team in one dimension, but in reality they exist in many different dimensions.   What we are trying to do is not merely impossibly difficult, but theoretically impossible, in the same way that it is theoretically impossible to state both height and weight in one scale.   You certainly could combine height and weight into one number, but by doing so we would end up saying that a person who is 6-foot-2 and 165 pounds is the same size as a person who is 5-foot-7 and 210 pounds.   You can say he is the same size, but he is not the same size.     We should never lose sight of the problems that this causes.

                Because this is true, I think that it is best, in reviewing Hall of Fame candidates, not to try to go all the way to the finish line with one metric.  Let us say, for example, that 350 Win Shares represents a Hall of Famer, or that 60 WAR represents a Hall of Famer.  You can say that Vladimir Guerrero and Dazzy Vance are the same, in that they are both at 59.9 WAR, but the reality is that they are NOT the same.   They are very different.    In saying that they are the same, we are merely pretending that something is true that we know very well is not actually true.

                The scale on which I evaluate players, then, is Win Shares and Loss Shares, and my criterion for the Hall of Fame is this:

                1)  There are two relevant standards, which are 300 Career Win Shares and 100 more Win Shares than Loss Shares,

                2)  If a player meets both of these standards, he is a Hall of Famer, unless he has some other disqualification,

                3)  If a player meets neither of these standards, he is not a Hall of Famer unless he has some other qualification, and

                4)  If a player meets one of these standards but not the other, then it is a matter best left to judgment as to whether he should or should not be selected. 

                These standards do not work for relievers, for reasons that I accept but do not quite understand, and they do not work, of course, for managers, general managers, scouts or pioneers.   There may or may not be an issue with catchers.  Other than that, they appear to work satisfactorily in every case—accepting certain issues, which we will discuss later.   I have been using these standards for several years, and I have yet to find a player, other than relievers, for whom this process does not seem to deliver an answer I can live with.  Such a case may exist; I just haven’t found it yet. 

                 These standards work because of two coincidences, or maybe they are not coincidences; maybe we should describe them as harmonies.   First, it happens that the size of a player’s career is essentially the same as the size of a team.  Again, the problem of measuring things in one dimension which exist in fact in multiple dimensions, but. . .how many players make up a team?    There are 25 players on a roster, but I am not asking how many players are on a roster; what I am asking is, how many players does it take to win a pennant?     About 17, 18, something like that.   You need 8 regular position players, 9 with a DH; you need 5 starting pitchers, and you need at least 3 relievers.   That’s 16, 17; you need a bench player or two.   Some of the roster is filler material.    A team really consists of about 17 or 18 contributing players.

                And how long does a player’s career last?   17, 18 years in most cases is a full career.     Think of it in terms of outs.    If a team plays 162 games and makes 27 outs, that is 4,384 outs, or 8,768 outs if you include pitching, defense, hitting and baserunning.   A full team is about 8,768 outs.

                And how many outs is a player responsible for in his career?    It’s about the same number.   Let’s take. . .Willie Stargell, or Willie McCovey.    McCovey made 6,259 batting and baserunning outs in his career; Stargell accounted for 5.938.    McCovey, however, also played 18,818 innings in the field, and Stargell also played 17,956 innings in the field.     Let’s start with the number 18,000 innings, which is 54,000 outs.    About 40,000 of those are probably best assigned to pitchers; that leaves 14,000.    When those 14,000 outs are split among eight fielders, that’s 1,750 per fielder.    McCovey and Stargell were not defensive stars or defensive specialists; more credit must be given to shortstops, catchers and center fielders for their defensive work than to first basemen and left fielders, so. .. .let’s say 1,200 each.   Add those 1,200 outs to the batting outs, and. . .a player’s career, in terms of outs, is roughly the same size as a team/season.  

                The second coincidence or second harmony is that the standard deviation of success for players in a career is similar to that of players on a team.    I never realized this until the last year or two.   I had always assumed that the spread of talent among players on a team was larger than the spread of talent among seasons of a career, but actually, it isn’t, or at least I now believe that it isn’t.    Willie Stargell had some seasons in which he was absolutely great (1966, 1971, 1973, 1978) and some seasons in which he was less than average.     I don’t think that anyone actually knows exactly what the normal standard deviation of winning percentage is for players on a pennant-winning team or for seasons of a long career, but I think that when sabermetrics finally arrives at consensus numbers for those two, they will be about the same.

                Because of these two harmonies, a player’s career may be represented as if it were a team/season—not in terms of all of the elements, having a shortstop, having a closer, etc., but in terms of size and quality.     Well, think about it:  What is 300 Win Shares, and 100 more Win Shares than Loss Shares?   It is a championship team.   300 Win Shares is 100 wins.    100 more Win Shares than Loss Shares is 33 games above .500.    If you win about 100 games and/or finish 33 games over .500, you win the pennant, or you win your division.   You’re a championship team.

                What we are really saying, then, is that a player is a Hall of Famer if, when his career is looked at as a team/season, that is a championship team.   A player who has a good long career is assigned responsibility for about one season’s worth of wins and losses.   If that’s a championship season, he’s a Hall of Famer.

                There is a third harmony here, which is that 300 Wins is a recognized standard of a Hall of Fame pitcher.   A pitcher’s Win Shares and Loss Shares are, in many cases, about the same as his wins and losses.   300 Win Shares by a position player. . .300 wins by a pitcher.   It’s about the same thing.   That’s another way of explaining why it works.  

But I could give you a thousand reasons why such an approach should work, and, if it doesn’t work, it doesn’t work.   This not only should work, but actually does.   I have yet to find any case in which it doesn’t—other than closers.


This article is the first of a four-part series about the Expansion era ballot; the other three will be published over the next three days.   As a writer, I very much dislike publishing a "series" of articles like this, because I dislike reading articles that way.    When somebody publishes a series of articles on a topic, I normally read the first one and maybe the second one and then lose interest.   However, the four articles total about 12,000 words, and in the modern world it is just not practical to publish a 12,000-word article on the web.   

What I am saying is, I understand if you drift off in the middle of this series, but the most significant of the series is Part IV, which I think will be published on Friday.    That part of the article deals with a couple of sabermetric issues, and suggests new approaches.   If you don’t get time to read the whole article, I understand, but if you could check back in on Friday, I’d appreciate that.   Thanks.    Bill


COMMENTS (15 Comments, most recent shown first)

Or, put Win Shares on the x-axis, Loss Shares on the y-axis, and each person can draw their own line, giving whatever weight they want to those two, based on the slope of the line they draw.

Here's an example, based on something a Straight Arrow reader did:
2:47 PM Nov 20th
I look at Win Shares like the plain white canvas and paints for an artists painting. For a HOF career, it takes a pretty big canvas, 300 win shares. It also takes a lot of bold colors to draw the eye, 100 more win shares than loss shares.

You might also draw the eye to an odd painting, less than 300 win shares but a big difference in win shares and loss shares. Pedro Martinez come to mind. Mike Mussina maybe, although 270 wins is a pretty big canvas.

You cant see a whole career in Win Shares, but you get a pretty good idea of the size and depth of the painting.
11:39 AM Nov 20th
Thanks studes.
9:24 AM Nov 20th
" If, in the third grade, we were all assigned a "BQ"—a "Body Quotient"—society would reject that, because we would immediately SEE the failings and limitations of the concept. "

But we don't. We're using the pseudo-scientific BMI on children as young as 2. Society loves its single-dimensional evaluations of quality, common sense be damned. The advantage of WARP, Win Shares or whatever is that they provide a great place to *begin* the conversation.
9:01 AM Nov 20th
Bill has published many articles here detailing the exact calculations for win shares and loss shares by position. I don't mean to put words in his mouth, but I think his system keeps evolving and he doesn't want to put it in a book because the system is liable to change a month after a book is published. Also, a book is too much work.
8:34 AM Nov 20th

I would think that a typical baseball season would have 486 game shares (3 per game) to be divided between win shares and loss shares. If I understand it correctly, you can calculate a hitters shares by dividing the outs he made by 27 and multiplying the total by 1.5, then subtract his win shares from the total which will leave his loss shares, which in the case of great hitters will be zero. For pitchers, take their innings, divide by 9, multiply by .354, then multiply by 3 to get game shares, from which you subtract win shares to get loss shares. For fielding shares, take the innings played, divide by nine, then divide by 12 for catchers, 12.5 for shortstops, 14 for 2B, 19 for 3B, 20 for CF, 26.67 for LF and RF and 38 for 1B (I got this from Bill's short form method for calculating win shares for fielding). Round off to match the team's won loss record for the season. I may be wrong, but that is the best I can do unless Bill has published his formula somewhere.​
7:11 AM Nov 20th
My father published an article in a book called The Fallacy of IQ in 1974 that makes the same point that IQ is a one-dimensional number trying to capture a multi-dimensional quality. He used the one-on-one football games that my brother and I played in the backyard to help illustrate one point. The skills to be good at one-on-one football are correlated with the skills in real football, but only very weakly. The skills measured by an IQ test are correlated with the skills to succeed in school, but only somewhat, and are even more weakly linked to many dimensions of intelligence. Moreover, the differences between groups may be more due to test-taking skill than intelligence. He was trying to counter Herrnstein who was arguing that there was deep significance to the disparity between racial groups on IQ test performance.
7:35 PM Nov 19th
Thanks for this article,and as always for making me think beyond baseball.

As to baseball, I have an unscientific sense that there will be more deserving players on this list than most people think. At least by the standards of the existing hall. This era has just the standard process - no oldtimers committees, etc. And for some of this period hitting stats were depressed, and then They normalized just before a huge run up these players missed, which causes hitters of this era to be under valued in general.

I am very interested to see if my gut feel checks out, or is distressingly wrong again.
5:02 PM Nov 19th
Bill - don't worry about doing part series. All of us loved the Brooks Robinson Tournament, which ended up being a long series of posts. The Man From the Train saga is one of your most popular semi-regular posts. I probably speak for several in wishing that the Omars & Ozzies series popped back up. So... we like series. They're good reads.
3:40 PM Nov 19th
Are Win and Loss Shares published anywhere yet? Is there a plan to detail the methodology at some point? I realize it's probably a book level effort, and I've seen the assorted articles here over the years, just wondering if it's all in one place somewhere that I might not be aware of?

I realize Bill has a day job :-) Just curious if there is a plan or a timeline for something like that, that's all. Maybe better asked as a Hey Bill?

Every time I see the references to them, it makes me want to see the whole thing, dive into the details of the system, etc. But this is also coming from someone who built a spreadsheet capable of calculating the original Win Shares for a 16-team league, assuming you don't mind doing the data input.
2:13 PM Nov 19th
This could be one of your most thought provoking articles ever-I'll be reading every word.
11:28 AM Nov 19th
Sorry, Bill, I'm totally bored already. :-)
8:56 AM Nov 19th
I don't know why we insist on trying to reduce complexity to a single number--perhaps it's just mental laziness--but there has surely been that tendency within sabermetrics, whether it's been called WAR or Total Runs or whatever.

Bill, as usual, you're ahead of the curve. While the rest of us were waving the banner of Win Shares, you saw how far it was from capturing the complexity of the game and came up with Loss Shares to--I won't say complete the picture but to add a healthy portion of nuance.

You can be damn sure I'll read all 12000 words.
8:37 AM Nov 19th
They seem to be coincidences. If they weren't the relationship would work in other sports. Obviously for a sport like basketball, you need about 8/9 contributors; a great career would have far more seasons than that. I don't know football as well baseball, but there are clearly more than 18 contributors. There are 22 starters, plus, I would guess another 5 or so. No football career lasts close to 27 years...
7:47 AM Nov 19th
A person can be brilliant at math but unable to carry on a normal conversation.

How sweet! You remembered me!
7:07 AM Nov 19th
©2020 Be Jolly, Inc. All Rights Reserved.|Web site design and development by|Terms & Conditions|Privacy Policy