I believe that I wrote this article in September or October, 2002, although it has not been published until now. It is, I now realize, an article with many flaws, but I decided to publish it now as originally written.
The article introduces primitive forms of two articles which were later used to better effect in the article “Hanrahan’s List”, which you can also find here. The research here is more primitive, but sabermetrics is an evolving process, and we shouldn’t be ashamed to acknowledge that methods exist in multiple forms.
There are other problems with the research. The article outlines ways that the method could be improved—but five years later, this work has not been done. It hasn’t been done, in part, because I was hired by the Red Sox a month after I wrote this article, and I’ve been busy. It hasn’t been done, in part, because I’ve lost confidence in a basic assumption underlying the method—namely, that a method which identifies Hall of Famers in mid-career from the past will continue to work in the future. I am now more inclined to think that it won’t continue to work in the future. But I guess we’ll see.
The article refers to Bobby Abreu as a “young player”, which he was at the time, and also contains a comparison of Roberto Alomar and Craig Biggio which turned out to be, in retrospect, entirley incorrect. But I decided to leave it.
I. The Basic System
Take a young player. . .let’s say Bobby Abreu. A-Rod is too obvious. What is the chance that Bobby Abreu will eventually be in the Hall of Fame?
I have a method, which I developed many years ago, which monitors a player’s progress on the road to the Hall of Fame. That method has its flaws, and it has its benefits, but I am asking here a slightly different question. That method asks “What are Bobby Abreu’s Hall of Fame credentials at this point in time?” That method answers that question in such a way that one can give an answer to the question “If Bobby Abreu retired tomorrow, what would be his chance of going into the Hall of Fame?” But I am asking here a somewhat different question, not “What is his current location?”, but rather “Where is he headed?” Is he on target toward the Hall of Fame?
This research is based on the player’s Win Shares. As a general rule, a player has a better-than-50% chance to go into the Hall of Fame if he retires with 325 or more career Win Shares, and a less-than-50% if he retires with less than that. Pitchers are a little different; we’ll deal with them in a moment, but of 22 non-pitchers in history who are Hall of Fame eligible and who have 325 to 349 Win Shares, 12 have been elected to the Hall of Fame. Of 32 non-pitchers with 300 to 324 Win Shares, only 14 have been elected to the Hall of Fame. Above 350, the odds swing sharply in the player’s favor; below 300, a player is a long-odds Hall of Fame candidate:
350-374 Win Shares 15 of 18 83%
325-349 Win Shares 12 of 22 55%
300-324 Win Shares 14 of 32 44%
275-299 Win Shares 15 of 41 37%
So at 325 Win Shares, a hitter becomes a likely Hall of Famer.
My first thought, then, was that we could assess a player’s chance of making the Hall of Fame by assessing his chance of getting to 325 Win Shares. My best method to tackle that problem is another old invention known as the favorite toy. The favorite toy begins by asking three questions:
a) How far away from the goal is the player?
b) How fast is he moving?
c) How much time does he have?
Given answers a through c, we can estimate the chance that the player will reach the goal. If a player is 30 years old, sixty Win Shares away from the goal, and moving at a rate of 35.0 Win Shares per season, it is fairly obvious that he will probably pass the marker. If, on the other hand, the player is 35 years old, 150 Win Shares away from the goal, and moving at a rate of 15.0 Win Shares per season, it is fairly obvious that he is not going to reach the goal.
Let’s take a tougher case. . .suppose that a player is 29 years old, has 186 career Win Shares already, and is moving at a rate of 29 Win Shares per season. What is the chance that he will make 325?
Well, let’s see. . .if he’s 29 years old, he probably has about 6.5 seasons left, as a guess. If he earns 29 Win Shares per season for 6.5 seasons, that would be 188.5 more Win Shares. He needs 139 to make 325. . .thus, he appears to be on target to pass the 325 mark easily, thus he appears to be a likely Hall of Famer.
This is the actual data for Chipper Jones after the 2001 season—he was 29 years old, had an established Win Share level of 29.0, and had 186 career Win Shares at the end of the season. Let me explain a few details of the math . .
a) How far away from the goal is the player?
This one is easy. Jones had 186 career Win Shares; he is 139 away from the goal of 325.
b) How fast is he moving?
The formula I used here is this:
The player’s Win Shares four years ago, plus
His Win Shares three years ago, times two, plus
His Win Shares two years ago, times three, plus
His Win Shares last season, times four,
All divided by ten.
Chipper Jones earned 29 Win Shares in 1998, 32 in 1999, 27 in 2000 and 29 in 2001. That makes an established Win Share level of 29.0—29, plus 64, plus 81, plus 116 makes 290, divided by ten is 29.0. Chipper Jones’ established Win Share level was 29.0.
There are a couple of other formulas that I use sometimes to figure established performance levels. . .it doesn’t matter a lot which one you use. A player’s established Win Share level cannot be less than .75 times his Win Shares in the most recent season. That is, suppose that a player’s Win Shares in the last four years are 0, 0, 0, and 20. That would make an Established Win Share level of 8.0—except that there is an “override” which moves it to 15.0. If a rookie has 200 hits in a season, it’s silly to say that his establish hit level is 80. We may not want to move him to 200 right away, but we have to get him in a reasonable range.
c) How long does he have?
Also known as Years Remaining. The formula I used here is (42 minus age, divided by two). If a player is 20 years old, this gives him a career expectation of 11 years; if he is 30, this gives him an expectation of six more seasons. Again, I have used slightly different formulas at times. It’s just a guess; we don’t really know how long the player has. Chipper Jones was 29 years old in 2002, so we will guess that he has 6.5 seasons left.
OK, two more questions:
d) How do you put that together to estimate his chances of reaching the goal? and
e) He hasn’t been hanging around with strippers again, has he?
Sorry, Chip. Answering question (d). . .We put (b) and (c) together to make an estimate of how much career the player has left. The player’s established Win Share level (29.0) multiplied by his years remaining (6.5) indicates that Jones probably had about 188.5 Win Shares left.
We established in (a) that Jones needed 139 more Win Shares to be a likely Hall of Famer. Since (d) is greater than (a)—that is, since Jones’ expected Win Shares are more than his needed Win Shares—his chance of reaching the goal must be greater than 50%.
I put them together in this way:
Divide the Expected Remaining Win Shares
By the Win Shares needed
And subtract .50.
If the player’s expected remaining Win Shares are the same as his Win Shares needed, then the player has a 50% chance of reaching the goal. For Jones, this would make .86. Jones would have an 86% chance of reaching 325 career Win Shares.
Which is not our actual estimate for him; it’s too high. Jones was likely to make the Hall of Fame from that point, but not 86% likely. But I’ll explain a couple more details of the system before moving on. . .
Detail A) If a player shows less than a zero % chance of reaching the goal (which, in fact, the vast majority of players do, since the vast majority of players are not any threat to the Hall of Fame), we count that at zero.
Detail B) If a player would calculate as having a greater-than-100% chance of reaching the goal by the method above, we cap that at .97 per season. That is, if the player shows a 128% chance of reaching the goal, but is four years away from the goal, then his chance of reaching it is entered as .97 to the fourth power, which is .885. If he is 5.62 seasons from reaching the goal, his chance of reaching it cannot exceed .97 to the power 5.62, which is .843.
OK, I had to clear those details off the table, but we have bigger fish to fry. At this point, frankly, our system doesn’t actually work all that well. At this point, we are assuming that the player’s chance of reaching 325 Win Shares and his chance of reaching the Hall of Fame are the same. Of course, they are not. By this silly system, if a player has earned 326 Win Shares in his career, he has a 100% chance of making the Hall of Fame. On the other hand, if a player has earned 295 Win Shares, but is old and not moving very fast, we would give him no chance of being elected to the Hall of Fame, even though a depressingly large number of people have been elected to the Hall of Fame with less than 295 Win Shares, or, for that matter, less than 250. We have to spread the marker out to reflect the real situation, which is that Hall of Fame selection bears a somewhat more imprecise relationship to 325 Win Shares.
How do we do that? Well, in this system, what we are really asking is not “What are the chances that Johnny will make the Hall of Fame?”, but “What are the chances that Johnny will have a career of Hall of Fame stature?” We disengage from 325 Win Shares as a golden snitch by establishing a long series of markers. What we do is, we go through the exact process outlined above, except that we go through it ten times, rather than once. The ten questions we ask are:
1) What are the chances that this player will earn 235 Win Shares in his career?
2) What are the chances that this player will earn 255 Win Shares in his career?
3) What are the chances that this player will earn 275 Win Shares in his career?
4) What are the chances that this player will earn 295 Win Shares in his career?
5) What are the chances that this yahoo will earn 315 Win Shares in his career?
6) What are the chances that this player will earn 335 Win Shares in his career?
7) What are the chances that this player will earn 355 Win Shares in his career?
8) What are the chances that this player will earn 375 Win Shares in his career?
9) What are the chances that this player will earn 395 Win Shares in his career?
10) What is the likelihood that this player will earn 415 Win Shares in his career?
We address each of those ten questions with the method outlined above. Then, to convert that into a Hall of Fame likelihood, we do four things:
1) We take the average of all ten answers.
2) We take the average of the middle six, questions 3-8 on the chart above.
3) We add those two averages together, and
4) We divide by two.
Returning to our old friend Chipper Jones, we estimate that he had, at the end of the 2001 season:
1) a 94.98% chance to earn 235 Win Shares in his career,
2) a 93.01% chance to earn 255 Win Shares,
3) a 91.08% chance to earn 275,
4) an 89.18% chance to earn 295,
5) an 87.33% chance to earn 315,
6) a 76.51% chance to earn 335,
7) a 61.54% chance to earn 355,
8) a 49.74% chance to earn 375,
9) a 40.19% chance to earn 395,
10) a 32.31% chance to earn 415.
The average of these ten estimates is 71.587. However, the average of the central six is 75.895. Adding those together and dividing by two, we thus estimate that Chipper Jones at the conclusion of the 2001 season had a 74% chance of reaching the Hall of Fame—actually, 73.7411, except that it patently absurd to say that you “estimate” that somebody has a 73.7411% chance of doing something.
In essence, what we are doing is assuming that a player can reach the Hall of Fame with as few as 235 Win Shares or even less, but that he is not certain to reach the Hall of Fame unless he has 415 Win Shares, that he becomes more and more likely to be elected to the Hall of Fame as he moves upward within that range, and that his Hall of Fame chances accelerate most rapidly as he moves through the center of that range.
This leaves two large questions:
1) What about pitchers, and
2) Does this thing actually work?
II. So How About Pitchers?
For pitchers we use the same system, except that the standards are lower for pitchers post-1920. Up to 1920 the standards were about the same (or even higher) for pitchers, since pitchers pre-1920 had very large Win Share totals in individual seasons. Post-1920 the peaks for pitchers are comparable to those for hitters until about 1975. Since 1975, the peaks for pitchers have dropped off, so that pitchers rarely are serious MVP candidates, plus, since pitchers are more at risk of injury, there are fewer pitchers than position players who have long, healthy careers. The consequence is that the effective Hall of Fame standard, for pitchers post-1920, is about 50 Win Shares lower than for non-pitchers.
I suppose that I should document that, shouldn’t I? 70% of you don’t care, and 25 of the other 30 are willing to take my word for it, but just in case. . .There are nine pitchers in history who:
a) were born after 1885,
b) earned 300 to 324 Win Shares in their careers, and
c) are eligible for the Hall of Fame.
All nine are Hall of Famers—100%. This is not to say that 300 Career Win Shares is an absolute guarantee; Bert Blyleven, with 339 Win Shares, has not been elected, but he’s the exception. Otherwise, 300 gets you in. Charting the data:
300-324 Win Shares 9 of 9 100%
275-299 Win Shares 3 of 4 75%
250-274 Win Shares 7 of 12 58%
225-249 Win Shares 4 of 19 21%
The data is too thin to rest comfortably upon, but it would appear generally that the standard of the Hall of Fame, for pitchers, is about 50 Win Shares lower than it is for hitters.
For pitchers post-1920, then, I used the same system, except that I lowered the counts by 50. I figured:
1) The pitcher’s chance to earn 185 Win Shares in his career.
2) The pitcher’s chance to earn 205 Win Shares in his career.
3) The pitcher’s chance to earn 225 Win Shares in his career.
4) The pitcher’s chance to earn 245 Win Shares in his career.
5) The pitcher’s chance to earn 265 Win Shares in his career.
6) The pitcher’s chance to earn 285 Win Shares in his career.
7) The pitcher’s chance to earn 305 Win Shares in his career.
8) The pitcher’s chance to earn 325 Win Shares in his career.
9) The pitcher’s chance to earn 345 Win Shares in his career.
10) The pitcher’s chance to earn 365 Win Shares in his career
I then combined these in the same way—average all ten estimates, average the central six, combine those into one figure.
III. Does This System Actually Work
Of course it works; why would I be telling you about it if it didn’t? Believe me, I’ve got a filing cabinet full of research that didn’t work, and I’m not going to tell you about any of it.
The system works, in this sense: that if you take a group of players who are reported by the system to have a 74% chance of making the Hall of Fame and check in on them years later, 74% of them are going to be in the Hall of Fame. Well, maybe 72% would be a better guess, but somewhere in that range.
To test the system, we can go to back to some point in the past—let’s say 1955—and make assessments of the Hall of Fame chances of every player active at that time, and then check to see how those assessments worked out.
According to our method, the top 25 Hall of Fame candidates active in 1955 were as follows:
  Hall of Fame
  Chance
1. Stan Musial 100%
Ted Williams 100%
3. Robin Roberts 93%
4. Duke Snider 91%
5. Mickey Mantle 86%
6. Yogi Berra 84%
7. Eddie Mathews 81%
8. Bob Feller 77%
9. Warren Spahn 76%
10. Richie Ashburn 75%
11. Pee Wee Reese 73%
12. Larry Doby 73%
13. Willie Mays 68%
14. Johnny Mize 63%
15. Bob Lemon 61%
16. Ralph Kiner 53%
17. Early Wynn 53%
18. Gil Hodges 53%
19. Hal Newhouser 52%
20. Ted Kluszewski 51%
21. Mickey Vernon 49%
22. Enos Slaughter 49%
23. Al Rosen 48%
24. Billy Pierce 46%
25. Nellie Fox 46%
Johnny Mize actually retired after the 1953 season, but he shows up on the 1955 list because the system looks at performance over the last four seasons. . .anybody who contributed at least one Win Share to his team from 1952 through 1955 can show up on the 1955 list. The only certain Hall of Famers of 1955 were Musial and Williams. It is interesting to note that Mickey Mantle in 1955—23 years old, and a year before his first MVP season—is already shown by our method as being 86% certain to eventually be in the Hall of Fame. That’s fine; I’m trying to develop a system that looks forward, not one that looks back. If the system could look at minor leaguers and tell me which one of them would eventually make the Hall of Fame (which of course it can’t), that would be great.
In one sense, this chart shows that our system works. . .all of the players listed here as most likely to go into the Hall of Fame did in fact do so, the top 17 players. The 1955 study also validates the method in another sense. The 1955 study predicts that 27 players active in 1955 would eventually be selected to the Hall of Fame. The actual number was 33, or 35 if you include two players (Monte Irvin and Satchel Paige) who were in the majors in that era, but selected for their play in the Negro Leagues. Twenty-seven, thirty-three. . .that’s a pretty good estimate.
But in another sense, this is an unsettling outcome, in that, in this test group, everyone who shows up at 60% or higher eventually winds up in the Hall of Fame. We’ve got eleven players who are shown with a Hall of Fame probability between 60 and 90%--but all eleven of them are actually in the Hall of Fame. Thus, what this test group suggests is that, if the system says you have a 60% chance of getting into the Hall of Fame, you actually have a 100% chance of getting into the Hall of Fame.
It’s too small a sample, of course; one season’s worth of players isn’t enough to actually determine whether the system works. I started out to test the system by checking the Hall of Fame chances of every player active in 1920, 1930, 1940, 1950 or 1960 (or in the three preceding seasons). That turned out to be an interesting exercise, plus I figured out how to replicate the process so as to do additional seasons pretty easily. . .I wound up checking the Hall of Fame chances for every player from the 1900 season, 1910—and every season between 1920 and 1976.
This makes a total of 39,816 player/seasons evaluated. Since each of those data lines is 72 columns wide, this makes a spreadsheet which has just short of three million cells. As an aside, it is astonishing to me to be working with computers that can actually handle spreadsheets of this size. When I first worked with spreadsheets twenty years ago, I thought they were wonderful, but at that time a personal computer spreadsheet could handle a few hundred cells worth of information; anything above that and they would start to weird out on you. This thing is about 10,000 times larger than the capacity of the old one, yet it sorts the data easily, allows me to insert and remove new information without complaint, and makes room for new formulas and new calculations as if it had hardly noticed.
Anyway, of these 39,816 seasons, 87 are by players like Sparky Anderson, Al Lopez and Leo Durocher, who were selected to the Hall of Fame, but not as players, or who were selected as players, but from the Negro Leagues (Satchel and Monte Irvin). Another 22 are players who are not eligible for the Hall of Fame, either because they’ve been banned from the game or because they haven’t been retired long enough yet.
Eliminating those 109 players from the data, we then have 39,707 assessments of a player’s chance of going into the Hall of Fame. Of course, the vast majority of those—30,426, to be exact—are “zero”; this player’s chance of going into the Hall of Fame is “Forget it, loser.”
There are 223 players in the study who show as having a 100% chance to make the Hall of Fame. This is almost as many players as are actually in the Hall of Fame, which you may think strange, but a player can be on our list as a 100% Hall of Famer for several years. Henry Aaron appears on our list as a 100% certain Hall of Famer in 1965, and again in 1966, ‘67. . .on until 1976; he’s actually 12 of the 223 certain Hall of Famers in our file. Anyway, of these 223 certain Hall of Famers, all 223 are in fact in the Hall of Fame, so we have no problems there:
Hall of Fame
Chance Candidates Hall of Famers Percentage
100% 223 223 100%
In our study we have 265 players who are shown as having a 90 to 99% chance of going into the Hall of Fame. 244 of those 267, or 91%, are now in the Hall of Fame:
Hall of Fame
Chance Candidates Hall of Famers Percentage
100% 223 223 100%
90 to 99% 267 244 91%
Extending the data to its natural conclusion:
Hall of Fame
Chance Candidates Hall of Famers Percentage
100% 223 223 100%
90 to 99% 265 244 91%
80 to 89% 211 177 84%
70 to 79% 264 190 72%
60 to 69% 303 174 57%
50 to 59% 366 192 52%
40 to 49% 399 166 42%
30 to 39% 522 177 34%
20 to 29% 754 165 22%
10 to 19% 1177 148 13%
Up to 9% 4797 245 5%
No Chance 30426 208 <1%
In general, the study shows that our estimates are very close—not perfect, but close. If you take a group of players who are shown by our system to have a 74% chance of getting into the Hall of Fame, it is very likely that about 74% of them will eventually be in the Hall of Fame.
In a moment, I’ll talk about ways that the system could be improved. First, though, I wanted to take up a couple of other subjects.
IV. Fun Stuff We Learn From Doing This
So who are the 208 players who had zero chance of going into the Hall of Fame, but did, anyway?
Well, that one doesn’t turn out to be so interesting. . .that’s more under the heading of “Things that could be done to fix the system.” A player when he reaches the major leagues will show as having a zero percent chance of making the Hall of Fame until he plays at a certain level. In the language of the system, a player will show a zero percent chance of making the Hall of Fame until he posts an established value of about 10.0—higher than that if he starts late. Virtually all of the “bad zero estimates” are simply the early-career estimates for players who haven’t yet made the lineup. Harmon Killebrew, for example, shows up with a zero percent chance of making the Hall of Fame in 1954 (when he had 13 at bats), 1955 (when he had 80, but didn’t hit), 1956 (when he had 99, but didn’t hit), 1957 (when he had 31 at bats) and 1958 (when he had 31 more). He finally made the lineup in 1959—and showed an 8% chance of making the Hall of Fame after the season.
The other end of this question, however, is very interesting: Who are the players who were almost certain to make the Hall of Fame, but didn’t? The players who showed up in our list as having an 80% chance or greater to make the Hall of Fame, but didn’t, are:
1. Dick Allen, 1974   97%
2. Bill Dahlen, 1910 95%
3. Wilbur Cooper, 1924 94%
4. Ron Santo, 1972 93%
5. Wes Ferrell, 1936 92%
6. GeorgeVan Haltren, 1900 92%
7. Joe Torre, 1971 90%
8. Jimmy Sheckard, 1910 89%
9. Frank Howard, 1971 89%
10. Jimmy Wynn, 1976 88%
11. Carl Mays, 1921 86%
12. Sherry Magee, 1910 86%
13. Stan Hack, 1945 85%
14. Jim Fregosi, 1970 85%
15. Willie Davis, 1974 84%
16. Rusty Staub, 1971 84%
17. Vern Stephens, 1950 81%
18. Charlie Keller, 1943 80%
There are 18 players on this list, whereas the chart before showed 55 players who were over 80%, but aren’t in the Hall of Fame. This happens, again, because of multiple appearances on the list. . .Dick Allen, for example, shows up not only at 97% in 1974, but also at 95% in 1972, 95% in 1973, 93% in 1975, 91% in 1976, 85% in 1971, 83% in 1968, 80% in 1967, and 80% in 1969.
Anyway, this is a very interesting list of players. It includes guys who seemed like certain Hall of Famers, but got hurt (Wes Ferrell, Charlie Keller), guys who had Hall of Fame ability but haven’t made it because too many people hated them (Dick Allen, Carl Mays), good players who jumped briefly onto the list after they had career years (Joe Torre, Jim Fregosi), players who were never really appreciated by the press (Jimmy Wynn, Wilbur Cooper, Frank Howard) and combinations of these attributes (Willie Davis and Sherry Magee.) These are the “interesting biography” type guys, the guys who were just almost up to Hall of Fame standard, or who were up to the Hall of Fame standard, except for that (fill in the blank. . .injury, feud with the press, drinking problem, odd combination of skills, etc.)
Another fun thing about this system is that enables us to trace a player’s battle for immortality over the course of his career. Henry Aaron, for example:
Year Player Age WS Career Chance
1954 Aaron, Henry 20 13 13 .00
1955 Aaron, Henry 21 29 42 .33
1956 Aaron, Henry 22 30 72 .45
1957 Aaron, Henry 23 35 107 .73
1958 Aaron, Henry 24 32 139 .81
1959 Aaron, Henry 25 38 177 .88
1960 Aaron, Henry 26 35 212 .91
1961 Aaron, Henry 27 35 247 .94
1962 Aaron, Henry 28 34 281 .96
1963 Aaron, Henry 29 41 322 .98
1964 Aaron, Henry 30 33 355 .99
1965 Aaron, Henry 31 31 386 1.00
1966 Aaron, Henry 32 27 413 1.00
1967 Aaron, Henry 33 34 447 1.00
1968 Aaron, Henry 34 32 479 1.00
1969 Aaron, Henry 35 38 517 1.00
1970 Aaron, Henry 30 25 542 1.00
1971 Aaron, Henry 37 33 575 1.00
1972 Aaron, Henry 38 21 596 1.00
1973 Aaron, Henry 39 20 616 1.00
1974 Aaron, Henry 40 13 629 1.00
1975 Aaron, Henry 41 9 638 1.00
1976 Aaron, Henry 42 5 643 1.00
Aaron established himself as a Hall of Fame candidate in 1955, his second year in the majors, when he hit .314 with 27 homers, 106 RBI. He crossed over to being a likely Hall of Famer two years later, when he won the National League MVP Award with his first 40-homer season (44, 132, 322). He was then 23, but, like Mantle at the same age, already a likely Hall of Famer. In fact, why don’t we set this aside for a second, and make up a list of all the players within the study who were likely Hall of Famers by the age of 23:
Year Player Age Career Chance
1910 Cobb, Ty 23 186 .91
1942 Williams, Ted 23 150 .88
1955 Mantle, Mickey 23 148 .86
1935 Vaughan, Arky 23 130 .84
1955 Mathews, Eddie 23 125 .81
1932 Ott, Mel 23 144 .80
1944 Musial, Stan 23 108 .78
1931 Foxx, Jimmie 23 122 .74
1957 Aaron, Hank 23 107 .73
1931 Ferrell, Wes 23 87 .72
1974 Cedeno, Cesar 23 119 .72
1974 Blyleven, Bert 23 101 .70
1910 Collins, Eddie 23 93 .70
1944 Newhouser, Hal 23 77 .70
1942 Feller, Bob 23 137 .69
1938 DiMaggio, Joe 23 94 .66
1962 Pinson, Vada 23 109 .66
1969 Jackson, Reggie 23 68 .63
1971 Bench, Johnny 23 107 .62
1960 Drysdale, Don 23 91 .61
1954 Mays, Willie 23 64 .60
1961 Cepeda, Orlando 23 98 .59
1964 Chance, Dean 23 61 .58
1963 Ellsworth, Dick 23 58 .57
1958 Kaline, Al 23 108 .55
1920 Youngs, Ross 23 84 .54
1935 Medwick, Joe 23 85 .54
1959 Robinson, Frank 23 98 .53
1944 Stephens, Vern 23 76 .50
Vern Stephens is a hair under 50% (49.9), so all of the players who were perceived as likely Hall of Famers at a very young age became actual Hall of Famers except Wes Ferrell, Cesar Cedeno, Bert Blyleven, Vada Pinson, Dean Chance and Dick Ellsworth. Pinson is listed here on the assumption that he was born in 1938, which is the date listed for him in the Encyclopedias, although I’m fairly sure that is a baseball age, and he was actually born in 1936. He might not make the list with corrected data.
Willie Mays is projected at 53% really based on one season, his MVP season in 1954; he was in the army in 1953 and almost all of 1952. Ellsworth and Chance failed to follow through on their superb seasons in ’63 and ’64, but the interesting thing there is that this list isn’t polluted by a large number of fluke-season pitchers, as I would have thought it might be.
Another thing we can do is list the top Hall of Fame candidates by age:
Age 18 Career WS HOF Chance
Bob Feller, 1937 19 .03
Age 19 Career WS HOF Chance
Bob Feller, 1938 41 .34
Wally Bunker, 1964 19 .17
Gary Nolan, 1967 19 .17
Age 20 Career WS HOF Chance
Bob Feller, 1939 73 .71
Mel Ott, 1929 57 .48
Mickey Mantle, 1952 45 .47
Age 21 Career WS HOF Chance
Bob Feller, 1940 107 .83
Eddie Mathews, 1953 58 .64
Mel Ott, 1930 85 .59
Age 22 Career WS HOF Chance
Bob Feller, 1941 137 .83
Ted Williams, 1941 104 .78
Mickey Mantle, 1953 107 .75
Age 23 Career WS HOF Chance
Ty Cobb, 1910 186 .91
Ted Williams, 1942 150 .88
Mickey Mantle, 1955 148 .86
Age 24 Career WS HOF Chance
Mickey Mantle, 1956 197 .91
Arky Vaughan, 1936 165 .87
Jimmie Foxx, 1932 162 .86
Age 25 Career WS HOF Chance
Mickey Mantle, 1957 248 .95
Babe Ruth, 1920 231 .94
Mel Ott, 1934 213 .90
Age 26 Career WS HOF Chance
Babe Ruth, 1921 284 .97
Mickey Mantle, 1958 287 .97
Rogers Hornsby, 1922 236 .94
Age 27 Career WS HOF Chance
Mickey Mantle, 1959 317 .98
Babe Ruth, 1922 313 .98
Mel Ott, 1937 284 .96
Age 28 Career WS HOF Chance
Babe Ruth, 1923 368 1.00
Mickey Mantle, 1960 353 .99
Mel Ott, 1937 316 .98
Age 29 Career WS HOF Chance
Babe Ruth, 1924 413 1.00
Mickey Mantle, 1961 401 1.00
Mel Ott, 1938 352 .99
Age 30 Career WS HOF Chance
Mickey Mantle, 1962 434 1.00
Babe Ruth, 1925 426 1.00
Kid Nichols, 1900 408 1.00
Mel Ott, 1939 380 1.00
After age 30 you just have a lengthening list of players who can be accurately described as “certain Hall of Famers.”
Bob Feller had a 71% chance of being a Hall of Famer by age 20, which was the year the Hall of Fame opened. Feller and Ted Williams, both born in late 1918, were by 1941 the two leading 22-year-olds in my study; no doubt that is why they call that the Greatest Generation. Ty Cobb only makes one appearance on the list (1910) because I haven’t run the data for the other years between 1901 and 1919. . same with Babe Ruth pre-1920, also Kid Nichols pre-1900 (although nobody much talks about Kid Nichols.) Walter Johnson, Pete Alexander. . .there are other guys who would show up on these lists if I had run the data for their years.
I’m not embarrassed that Wally Bunker and Gary Nolan show up on the list as 19-year-old pitchers; after all, the number is 17%. The system says that these guys had a one-in-six shot at developing into Hall of Famers. They just didn’t happen to do it.
OK, doubling back to where we were when I gave you the Henry Aaron year-by-year. . .Henry Aaron is very unusual, in that his Hall of Fame chances went up, and up, and up. . .they went from zero to 100 without ever dipping. Most players’ Hall of Fame battles are somewhat more circuitous . . . Early Wynn, for example . . . I’ll add his conventional stats to the chart:
Year Age W-L ERA WS Career HOF Chance
1939 19 0- 2 5.75 0 0 .00
1941 21 3- 1 1.58 4 4 .00
1942 22 10-16 5.12 3 7 .00
1943 23 18-12 2.91 19 26 .08
1944 24 8-17 3.38 8 34 .01
1945 25 (War) 34 .00
1946 26 8- 5 3.11 9 43 .00
1947 27 17-15 3.64 20 63 .09
1948 28 8-19 5.82 3 66 .00
1949 29 11- 7 4.15 8 74 .00
1950 30 18- 8 3.20 21 95 .10
1951 31 20-13 3.02 24 119 .21
1952 32 23-12 2.90 21 140 .35
1953 33 17-12 3.93 16 156 .34
1954 34 23-11 2.73 24 180 .47
1955 35 17-11 2.82 21 201 .53
1956 36 20- 9 2.72 28 229 .69
1957 37 14-17 4.31 10 239 .60
1958 38 14-16 4.13 11 250 .55
1959 39 22-10 3.17 23 273 .67
1960 40 13-12 3.49 16 289 .70
1961 41 8- 2 3.51 9 298 .70
1962 42 7-15 4.46 5 303 .63
1963 43 1- 2 2.28 6 309 .77
Wynn pitched in 1939, but never looked anything like a Hall of Famer until 1943, when he won 18 games. He lost 17 games the next season, however, then went off to war, and his Hall of Fame chances dropped from 8% to zero. He didn’t re-establish himself as a potential Hall of Famer until 1947, when he won 17 games for the Senators, putting him back on the charts at 9%.
Again, however, he failed to follow through, losing 19 games the next season, and returning to his status a Hall of Fame non-candidate. Traded to Cleveland, he began a string of outstanding seasons in 1950—eleven seasons after he had reached the majors—and built his Hall of Fame credentials in almost regular increments: 10%, 20, 30. . by 1956 he had had seven consecutive good seasons, and was up to 69%.
He followed that with a couple of losing years, however, and fell back to 55%. Winning the Cy Young Award in 1959 (he was 39 years old), he regained the ground he had lost in the previous two seasons, but it was not until 1963, when he won his 300th game, that he actually solidified his Hall of Fame credentials.
In some ways a similar career was that of Curt Simmons, Philadelphia and St. Louis left-hander of 1947-1967. Like Wynn, Simmons early in his career spotted excellent seasons around injury seasons and seasons donated to the United States Army. Like Wynn, he dropped off the map as a Hall of Fame candidate about age 30, but then rebuilt his career with a series of solid seasons. As late as 1964, when the 35-year-old Simmons went 18-9 for the Cardinals, he still had a 35% chance of reaching the Hall of Fame. He didn’t have the finishing kick that Wynn had, however, and thus ultimately failed to meet the Hall of Fame standard:
Year Age W-L ERA WS Career HOF Chance
1947 18 1- 0 1.00 2 2 .00
1948 19 7-13 4.97 3 5 .00
1949 20 4-10 4.59 5 10 .00
1950 21 17- 8 3.40 16 26 .06
1951 22 (Military Service) 26 .00
1952 23 14- 8 2.82 17 43 .07
1953 24 16-13 3.21 19 62 .15
1954 25 14-15 2.81 21 83 .32
1955 26 8- 8 4.92 4 87 .13
1956 27 15-10 3.36 13 100 .11
1957 28 12-11 3.44 15 115 .15
1958 29 7-14 4.38 6 121 .06
1959 30 0- 0 4.50 0 121 .00
1960 31 7- 4 3.06 15 136 .09
1961 32 9-10 3.13 17 153 .17
1962 33 10-10 3.51 9 162 .15
1963 34 15- 9 2.48 18 180 .30
1964 35 18- 9 3.43 16 196 .35
1965 36 9-15 4.08 7 203 .29
1966 37 5-8 4.23 4 207 .20
1967 38 5-8 4.24 3 210 .13
Another interesting comparison is Dick Bartell—Rowdy Dick—compared to Luke Appling, Old Aches and Pains. Bartell and Appling, both shortstops and both singles hitters, were both born in 1907. Bartell was seven months younger, but reached the majors three years earlier. Bartell came up at the end of the 1927 season, was a half time player in 1928, and a full-time regular in 1929, while Appling came up at the end of 1930, was a half-time player in 1931, and a more-or-less full-time regular in 1932:
Year Player Age WS Career Chance Year Player Age WS Career Chance
1927 Bartell 19 0 0 .00
1928 Bartell 20 7 7 .00
1929 Bartell 21 16 23 .01
1930 Bartell 22 18 41 .03 1930 Appling 23 0 0 .00
1931 Bartell 23 11 52 .03 1931 Appling 24 5 5 .00
1932 Bartell 24 21 73 .13 1932 Appling 25 12 17 .00
1933 Bartell 25 13 86 .09 1933 Appling 26 25 42 .06
1934 Bartell 26 18 104 .13 1934 Appling 27 14 56 .02
1935 Bartell 27 18 122 .18 1935 Appling 28 24 80 .10
1936 Bartell 28 24 146 .33 1936 Appling 29 29 109 .26
1937 Bartell 29 28 174 .55 1937 Appling 30 28 137 .38
1938 Bartell 30 17 191 .52 1938 Appling 31 9 146 .18
1939 Bartell 31 10 201 .35 1939 Appling 32 24 170 .25
1940 Bartell 32 11 212 .24 1940 Appling 33 28 198 .40
1941 Bartell 33 15 227 .27 1941 Appling 34 29 227 .57
1942 Bartell 34 12 239 .29 1942 Appling 35 20 247 .59
1943 Bartell 35 13 252 .33 1943 Appling 36 40 287 .84
1944 Bartell 36 (Military Service) .17 1944 Appling 37 (Military Service) .59
1945 Bartell 37 (Military Service) .10 1945 Appling 38 6 293 .46
1946 Bartell 38 0 252 .07 1946 Appling 39 26 319 .67
1947 Appling 40 22 341 .74
    1948 Appling 41 17 358 .78
    1949 Appling 42 19 377 .90
    1950 Appling 43 1 378 .90
Appling became as good a player as Bartell or a little bit better by 1933, but Bartell, given his quicker start, remained a stronger Hall of Fame candidate than Appling through 1939. It wasn’t until 1940, when Appling outhit Bartell .348 to .233, that Appling had a better shot at the Hall of Fame. Even then, Bartell was still ahead in career Win Shares (212 to 198), although Appling was the better Hall of Fame candidate because he was moving forward more rapidly.
Another fun list generated by this research is a list of the highest expected remaining value—what I have sometimes called “Trade Value”—within the study. Who are the most valuable players ever, in terms of being young players with very high established value—thus, unlimited futures. The top 25 within the study are:
Year Player HOF Age Es Val Ex RWS
  (Trade Value)
1910 Ty Cobb 1 23 42.5 404
1957 Mickey Mantle 1 25 46.9 399
1921 Babe Ruth 1 26 49.1 393
1942 Ted Williams 1 23 40.2 382
1920 Babe Ruth 1 27 44.9 382
1956 Mickey Mantle 1 24 41.7 375
1958 Mickey Mantle 1 26 44.8 358
1955 Mickey Mantle 1 23 35.6 338
1935 Arky Vaughan 1 23 35.3 335
1922 Rogers Hornsby 1 26 41.3 330
1936 Arky Vaughan 1 24 36.3 327
1923 Babe Ruth 1 28 46.4 325
1941 Ted Williams 1 22 32.2 322
1928 Lou Gehrig 1 25 37.5 319
1955 Eddie Mathews 1 23 33.2 315
1922 Babe Ruth 1 27 42.0 315
1944 Stan Musial 1 23 32.8 312
1933 Jimmie Foxx 1 25 36.6 311
1941 Bob Feller 1 22 30.8 308
1964 Dick Allen 0 22 30.8 308
1953 Eddie Mathews 1 21 29.3 307
1937 Joe Medwick 1 25 35.8 304
1940 Bob Feller 1 21 28.9 303
1932 Jimmie Foxx 1 24 33.4 301
1954 Mickey Mantle 1 22 29.9 299
The “HOF” column is “1” if this player is in the Hall of Fame—which all of these guys are except Dick Allen—and “0” if he is not (or “2” if he is not eligible, or “3” if he is in the Hall of Fame, but not as a player.) You will note that all of the players who had extremely high expected remaining value did go on to extremely fine subsequent careers. Below is the same chart again, except that I’ll take out the “HOF” column, eliminate redundant entries for the same player, and add the player’s ACTUAL remaining Win Shares, for the rest of his career:
Year Player Age Es Val ExRWS ActRWS
1910 Ty Cobb 23 42.5 404 536
1957 Mickey Mantle 25 46.9 399 317
1921 Babe Ruth 26 49.1 393 472
1942 Ted Williams 23 40.2 382 405
1935 Arky Vaughan 23 35.3 335 226
1922 Rogers Hornsby 26 41.3 330 266
1928 Lou Gehrig 25 37.5 319 355
1955 Eddie Mathews 23 33.2 315 325
1944 Stan Musial 23 32.8 312 496
1933 Jimmie Foxx 25 36.6 311 232
1941 Bob Feller 22 30.8 308 155
1964 Dick Allen 22 30.8 308 301
1937 Joe Medwick 25 35.8 304 151
1939 Joe DiMaggio 24 32.9 296 259
1959 Hank Aaron 25 34.8 296 466
1969 Reggie Jackson 23 30.8 292 376
1910 Eddie Collins 23 30.7 292 481
1934 Mel Ott 25 33.7 286 315
1954 Willie Mays 23 30.0 285 578
1946 Hal Newhouser 25 32.6 277 116
1972 Johnny Bench 24 30.1 271 212
1910 Walter Johnson 22 27.0 270 488
1962 Frank Robinson 26 33.7 270 323
1943 Charlie Keller 26 33.4 267 70
1974 Cesar Cedeno 23 28.1 267 177
All of these players went on to impressive subsequent careers except Charlie Keller, who was injured, and perhaps Newhouser, who had actually finished his best seasons at the time that his trade value peaked. The actual remaining Win Shares for these players seem, at least on an intuitive level, to match the expected remaining Win Shares fairly well. Some guys do better than expected; some players do less than expected. On balance, it looks about right.
These charts discriminate in favor of the young. We can remove the discrimination in favor of the young by focusing on the player’s Established Value—a weighted average of his last four seasons, weighted in favor of the most recent. There are 74 players within the confines of the study who had Established Value of 30.00 or higher, an accomplishment that requires either a series of MVP-candidate type seasons, or one historic season. This chart lists those 74 players, the year in which their established value peaked, and gives their trade value and actual remaining win shares at that point for the top 25:
Year Player Age Es Val Trade V Actual Value
1921 Babe Ruth 26 49.1 393 472
1957 Mickey Mantle 25 46.9 399 317
1910 Ty Cobb 23 42.5 404 536
1949 Ted Williams 30 41.4 248 233
1922 Rogers Hornsby 26 41.3 330 266
1910 Honus Wanger 36 40.8 122 147
1975 Joe Morgan 31 40.6 223 197
1965 Willie Mays 34 40.3 161 173
1949 Stan Musial 28 39.2 274 341
1934 Lou Gehrig 31 38.4 211 133
1936 Hank Aaron 29 37.1 241 321
1933 Jimmie Foxx 25 36.6 311 232
1910 Nap Lajoie 35 36.5 128 82
1936 Arky Vaughan 24 36.3 327 191
1931 Lefty Grove 31 36.2 199 199
1955 Duke Snider 28 36.0 252 134
1937 Joe Medwick 25 35.8 304 151
1936 Mel Ott 27 35.6 267 244
1941 Joe DiMaggio 26 35.5 284 187
1968 Carl Yastrzemski 28 34.5 242 283
1961 Eddie Mathews 29 34.4 224 131
1952 Jackie Robinson 33 34.4 155 74
1910 Ed Walsh 29 34.4 224 85
1967 Ron Santo 27 34.2 257 129
1970 Willie McCovey 32 34.1 171 126
Others over 30.0: Al Rosen, 1953 (34.0), Frank Robinson, 1962 (33.7), Al Simmons, 1931 (33.5), Charlie Keller, 1943 (33.4), Joe Cronin, 1933 (33.2), Bobby Murcer, 1972 (33.2), Charlie Gehringer, 1936 (33.1), Christy Mathewson, 1910 (32.9), Tris Speaker, 1920 (32.8), Dick Allen, 1967 (32.8), Frank Howard, 1969 (32.7), Hal Newhouser, 1946 (32.6), Carl Hubbell, 1936 (32.3), Hack Wilson, 1930 (32.3), Wally Berger, 1934 (32.3), Paul Waner, 1929 (32.2), Three Finger Brown, 1910 (32.1), Robin Roberts, 1954 (32.1), Johnny Mize, 1940 (32.1), Harmon Killebrew, 1967 (31.9), Sandy Koufax, 1966 (31.9), Dizzy Trout, 1944 (31.5), Norm Cash, 1961 (31.5), Ralph Kiner, 1951 (31.3), Dizzy Dean, 1936 (31.3), Mike Schmidt, 1976 (31.2), Willie Stargell, 1974 (31.1), Roberto Clemente, 1967 (31.1), Johnny Bench, 1975 (31.1), Jimmy Wynn, 1969 (31.0), Yogi Berra, 1954 (30.9), Bob Feller, 1941 (30.8), Joe Torre, 1971 (30.8), Reggie Jackson, 1969 (30.8), Larry Doby, 1954 (30.7), Eddie Collins, 1910 (30.7), Luke Appling, 1943 (30.6), Ernie Banks, 1960 (30.5), Edd Roush, 1920 (30.5), Harry Heilmann, 1925 (30.4), Goose Goslin, 1926 (30.4), Bill Terry, 1932 (30.3), Bobby Grich, 1976 (30.3), Rod Carew, 1976 (30.2), Fergie Jenkins, 1971 (30.1), Bobby Bonds, 1971 (30.1), Rusty Staub, 1971 (30.0), Earl Averill, 1934 (30.0), Steve Carlton, 1972 (30.0).
Since we have here a list of trade values (expected remaining future values) and a list of actual remaining values, another thing we can do is to pick out the players who had extremely high trade value, but whose actual futures were not so bright. This list is dominated by pitchers who got hurt:
Year Player Age Es Val Trade Actual V Shortfall
1936 Dizzy Dean 26 31.3 250 35 215
1976 Mark Fidrych 21 20.3 213 9 204
1969 Denny McLain 25 24.4 207 7 200
1943 Charlie Keller 26 33.4 267 70 197
1966 Sandy Koufax 30 31.9 191 0 191
1967 Jim Ray Hart 25 27.2 231 42 189
1970 Billy Grabarkewitz 24 21.8 196 13 183
1941 Cecil Travis 27 25.5 191 12 179
1900 John McGraw 27 26.8 201 24 177
1963 Dick Ellsworth 23 24.0 228 56 172
1941 Pete Reiser 22 25.5 255 84 171
1924 Ross Youngs 27 25.8 194 23 171
1965 Zoilo Versalles 25 24.0 204 34 170
1936 Wes Ferrell 28 26.7 187 22 165
1956 Herb Score 23 18.8 178 14 164
1926 Pete Donohue 25 21.9 186 23 163
1953 Al Rosen 29 34.0 221 58 163
1947 Whitey Kurowski 29 25.8 168 5 163
1935 Paul Dean 21 16.5 173 12 161
1946 Hal Newhouser 25 32.6 277 116 161
1940 Chet Ross 23 18.0 171 10 161
1943 Lou Klein 24 18.8 169 10 159
1930 Freddy Lindstrom 24 24.7 222 64 158
1920 Happy Felsch 28 22.5 158 0 158
1940 Hal Trosky 27 25.1 188 31 157
1942 Max West 25 19.0 162 7 155
1937 Joe Medwick 25 35.8 304 151 153
1941 Bob Feller 22 30.8 308 155 153
1947 Ralph Branca 21 19.5 205 52 153
1950 Art Houtteman 22 18.8 188 35 153
1921 Austin McHenry 25 18.8 159 7 152
1973 Nate Colbert 27 22.0 165 13 152
1975 Steve Busby 25 18.7 159 7 152
1943 Elbie Fletcher 27 24.8 186 35 151
1920 Jim Bagby 30 27.1 163 12 151
1970 Jim Fregosi 28 28.2 197 47 150
1970 Bobby Tolan 24 21.8 196 46 150
1972 Carlos May 24 21.8 196 47 149
1970 Sam McDowell 27 23.4 176 27 149
1951 Ralph Kiner 28 31.3 219 72 147
Other than pitchers who got hurt or, in some cases, sent to jail, this is another “interesting biography” list—players who seemed to have unlimited futures, but didn’t for one reason on another. There are 40 players on this list, 16 of them pitchers, almost all of whom had arm trouble, including Dizzy Dean and his brother Paul. Also on the list are a couple of players in died in mid-career (Youngs and McHenry), several players who lost time and ability to World War II, one player who was kicked out of baseball, several dramatic and famous injuries, and a handful of unexplained flameouts like Zoilo Versalles. Nine players on this list made the Hall of Fame despite falling 147 or more Win Shares short of their expected remaining value.
The opposite list, players who EXCEEDED expectations by a large amount, is of no interest, being only a repeat of the list of young players who had Hall of Fame careers starting with a little-value “season” in a late callup. A much more interesting subset of that group, however, is players who exceeded expectations by a large amount from age 30 or higher. . .that’s another list of good biographies. I’ll put the “HOF’ column back in to mark the Hall of Famers who were late-career surprises:
      Surprise
Year Player HOF Age Trade Actual V Value
1970 Phil Niekro 1 31 102 289 187
1947 Hank Sauer 0 30 5 168 163
1957 Hoyt Wilhelm 1 33 29 192 163
1922 Dazzy Vance 1 31 66 223 157
1900 Cy Young 1 33 130 281 151
1951 Warren Spahn 1 30 137 286 149
1920 Sam Rice 1 30 104 251 148
1946 Eddie Joost 0 30 8 155 147
1925 Babe Ruth 1 30 196 330 134
1969 Gaylord Perry 1 30 133 256 123
1938 Luke Appling 1 31 111 232 121
1961 Willie Mays 1 30 212 333 121
1953 Ted Williams 1 34 50 170 120
1950 Early Wynn 1 30 95 214 150
1947 Sal Maglie 0 30 11 130 119
1975 Tommy John 0 32 37 152 115
1921 Jack Quinn 0 37 35 149 115
1946 Ellie Kinder 0 31 25 139 114
1971 Luis Tiant 0 30 47 155 108
1976 Joe Niekro 0 31 25 128 103
1970 Willie Stargell 1 30 121 223 102
1927 Bill Terry 1 30 122 221 99
1947 Virgil Trucks 0 30 46 144 98
1948 Mickey Vernon 0 30 83 181 98
1920 Ken Williams 0 30 77 171 94
There are two players on this list (Dazzy Vance and Hoyt Wilhelm) who made the Hall of Fame after being listed as having zero percent probability at age 30 (or later), and seven others who made the Hall of Fame after having only a very slim chance at age 30. Again, we have mostly pitchers, proving again that pitchers are difficult to predict, and other than pitchers we have late-in-life surprises like Hank Sauer, Sam Rice, Ken Williams and Eddie Joost, along with a few superstars whose careers ran on longer than even superstar careers usually run on. We have another pair of brothers, the Niekros, which officially makes the Niekros the historical opposite of the Dean brothers. Jim Perry didn’t make the list, but can’t have missed it by very much.
V. Current Players
OK, we’ve farted around along enough here. Looking up the Hall of Fame candidates from 1933 is fun and instructive in its own little way, but the real purpose of the research is to tell us where people stand right now. What current players are Hall of Fame locks, which are on target, and should we abandon all hope for Miguel Batista?
Unfortunately, I can’t run the numbers yet; I’m waiting on the 2002 Win Shares. I’ll have to come back to this when I get the 2002 data.
VI. How the System Could Be Improved
Let me begin by pointing out something that probably isn’t a flaw in the system. Back in the early part of this article, when I was documenting that the system works, you probably noticed that most of our estimates were within reasonable ranges, but appeared a little high. Of players whom we believed had a 90 to 99% chance to go into the Hall of Fame, 91% have—but the percentage should be around 94 or 95%. All up and down the chart, our numbers look a little low.
But this probably isn’t a flaw in the system, and I wouldn’t try to “correct” it. The Hall of Fame is still electing people. While most of the players from the era 1920-1976 who are going to be in the Hall of Fame are probably there by now, not all of them are. Suppose the Hall of Fame comes to its senses and elects Ron Santo, or suppose it loses its senses, and elects Dick Allen? That 91% figure then would—eventually it will—become 92%, or 93%, or 94%.
If I re-calibrated the system now so that 95% of the players listed at 90-99% were in the Hall of Fame, the system would simply need to be re-calibrated again as soon as there were a handful of new selections. That would be a waste of time. There is little or no reason to believe that our 90-99% estimate is too high; it is much more likely simply that history is not finished yet. The Hall of Fame has not yet selected quite everybody that they are going to select.
However, even if we assume that our estimates are dead on target. . .that is, that of players that we say have a 74% chance of going into the Hall of Fame, 74% eventually do. ..even if that is true, our system can still be improved. Looking at them as individuals, each player either is a Hall of Famer, or isn’t. The only way a system could be 100% accurate would be to say about each player “yes” or “no”—0% or 100%--and be right every time.
Of course, we can’t ever achieve that level of accuracy, but we can move in that direction by pushing people toward the margins. If we can push the figures for the players who eventually will be in the Hall of Fame up and push the figures for the players who won’t be in the Hall of Fame down, then we have a better system. We can improve the system, while retaining the accuracy we have now, by pushing people toward the margins—creating more 80% estimates, more 90%, and fewer of those nebulous “tweener” numbers which just define groups of players, some of whom might be Hall of Famers.
How could that be accomplished? The three most obvious ways that the system could be improved would be:
1) Reduce or eliminate the “bad zero” estimates for young players,
2) Take into account the impact of big seasons, and
3) Recognize that catchers are different.
a) Reducing or eliminating the “bad zero” estimates for young players.
Henry Aaron, as a 20-year-old player in 1954, hit .280 with 13 home runs, 69 RBI. Back when I showed you the Henry Aaron year-by-year chart, it said that Aaron after that season had a zero percent (established) chance of becoming a Hall of Famer at that time.
Well, obviously, that doesn’t quite make sense. There are very limited number of players who play this well at the age of 20. From 1920 through 1976 there were 37 players (non-pitchers) who earned 10 Win Shares or more at age 20; Aaron, with 13, was near the middle of this group. In that group of 37 players are Aaron, Willie Mays, Mickey Mantle, Ted Williams, Al Kaline, Mel Ott, Frank Robinson, Johnny Bench, Jimmie Foxx, Arky Vaughan, Orlando Cepeda, Travis Jackson, Freddy Lindstrom, Eddie Mathews, Bobby Doerr, Bill Mazeroski and Robin Yount (not to mention a number of other players who had distinguished careers and might yet make the Hall of Fame, like Vada Pinson and Joe Torre.) Aaron’s chance of being a Hall of Famer, at age 20, far from being zero, was probably somewhere near 50%. It’s just that our system is structured in such a way—looking for performance over a period of seasons, and projecting it forward in a normal progression—that it can’t pick it up. The system needs some sort of sub-routine to focus on this type of players, and recognize their Hall of Fame potential as soon as it emerges.
While this particular example is unusual, many or most Hall of Famers start their career with a near-zero-value season at age 21 or 22. The system thus assumes that they have a zero percent chance of making the Hall of Fame. This is not the best assumption. Among all players who get late-season looks at ages 21 or 22, the percentage who go on to Hall of Fame careers is low—but it’s certainly not zero. The system should recognize the Hall of Fame potential of these players. We have 208 bad zero estimates in our study. Realistically, almost all of those could and should be eliminated. The only bad-zero estimates that we couldn’t get rid of would be those for players who have very unusual career paths, like Dazzy Vance and perhaps Hoyt Wilhelm.
b) Taking into account big-impact seasons.
Hank Greenberg was credited with 267 Win Shares in his career; Ron Fairly, with 269. Greenberg was in the Hall of Fame ten years after he retired; Fairly was dropped from the ballot after drawing only 1% of the vote in 1985.
Sandy Koufax, with 194 career Win Shares, was elected to the Hall of Fame the first time his face appeared on the ballot. Virgil Trucks, with 198 Win Shares, was dropped from the ballot after drawing only 2% in his first year. What’s the difference?
Obviously, Jewish sports writers. No, seriously, the difference is big-impact seasons. Greenberg and Koufax had short careers, but monster seasons.
One can track this in the system by, in essence, double-counting (or even triple-counting) Win Shares over 20. Koufax had seasons of 32, 24, 33 and 35 Win Shares—a total of 44 Win Shares above the “all-star” level of 20 Win Shares in a season. Trucks’ “big impact total” is 14. Greenberg’s total for “big impact seasons” is 78; Fairly’s is 8. If you add the big impact total to the raw Win Shares, Greenberg is at 345, Fairly at 277.
There is no reason the system couldn’t, or shouldn’t, track big-impact seasons in this way or some other. The only reason it hasn’t been done is just the work that it would require to re-calculate the values and re-calibrate the whole system.
c) Recognizing that Catchers are different.
Rick Ferrell is in the Hall of Fame with 206 Win Shares, Roy Campanella with 207, Ernie Lombardi with 218. Some other players are in with similar totals, but not very many. Catchers have short careers, and the Hall of Fame has (appropriatelly) adjusted for this. So should we.
If I was going to put another month into refining this system, those would be the first things I would look at. There are, however, many other things that could be done to perfect the system. Among those are:
1) Treat different skills differently.
Tim Raines had 390 Win Shares through 2001, which causes the system to view him as a near-certain Hall of Famer. In fact, Raines may not be all that certain to earn Hall of Fame selection, because he doesn’t do the things the Hall of Fame prefers. If Raines could trade half of his career walks for half as many singles, this would make him
a) less valuable (since a single is not worth twice as much as a walk), but
b) much more certain to earn quick selection to the Hall of Fame, since this would raise his career batting average to around .320, and give him comfortably more than 3000 hits.
The other system, the Hall of Fame monitor, looks essentially at whether the player does the things the Hall of Fame likes—whether he gets 200 hits in a season, whether he drives in 100 runs, whether he wins batting titles. This system ignores all of that, and studies only the issue of the quality of performance. I don’t want to turn this system into another version of that one, but there is some truth in each approach. A system which could look at a broader spectrum of information, as cumbersome as that might be, would be able to reach a greater level of accuracy.
2) Find some way to take notice of post-season play.
Post-season play helped put Rollie Fingers in the Hall of Fame, and it very possibly will do the same for Mariano Rivera. A perfect system would take this possibility into account.
3) Make some allowances for park effects and other important record distortions.
Craig Biggio has been as good a player as Roberto Alomar, but the fact is that the public still does not recognize this, and probably never will. A pitcher who pitches on good teams has a tremendous advantage in terms of earning recognition, relative to his actual value. A system which focuses only on actual value has no chance to deal with these things.
4) Find ways to recognize and eliminate players who show up as low-level Hall of Fame candidates based on an early start and a period of solid years, but who are clearly not Hall of Famers.
These guys are the opposite of the “bad zeroes”; these are the “false positives”. By 1955 Elmer Valo had been in the major leagues since 1940. He had not been a regular for six years—yet in our study, based on his 155 career Win Shares and steady forward progress, he still shows as having a 2% chance of having a Hall of Fame career. This is not necessary; there should be some way for the system to identify those kind of guys and shoot them down.
But nothing is perfect, and this is a good system. Whether I will get the time to work more on it, make some of these changes. . .I don’t know. As it is, I am happy to add it to the chest of tools with which I study the game, and I will plan to update you regularly on where current players stand with regard to Hall of Fame credentials.