8.   The Harnett/Dickey/Cochrane Era

              Bob O’Farrell was the Cubs’ catcher in the early 1920s, with Gabby Hartnett as his backup.   O’Farrell had a serious injury in 1924—my memory is that it was a broken skull suffered when he had a foul tip off of an ill-fitting face mask—and Hartnett took over the catcher’s job, and then the Cubs realized that Hartnett was even better than O’Farrell.   O’Farrell had a nothing year in 1924, but still ranks as the number one catcher in baseball after the 1924 season because

              1)  Prior to 1924 he was way, way ahead,

              2)  The catchers who ranked 2^nd, 3^rd, 4^th and 5^th prior to 1924 all had poor seasons in 1924, and

              3)  Our system looks at performance over a period of years, and is to an extent forgiving of injuries.

              The years 1924-1926 are half in the old era and half in the new.  In 1924 Hartnett took the job away from O’Farrell and ranks as the #4 catcher in baseball, so the Cubs have the #1 and the #4 catchers in baseball, which looks odd but is actually true; the Cubs really DID have two of the best catchers in baseball, possibly #1 and #2.   After 1925 Hartnett ranked as the #1 catcher in baseball, but in 1926 O’Farrell had an MVP season with St. Louis, and took the #1 position back.

              A similar thing happened in the American League.  Cy Perkins was one of the best catchers in the American League, but Mickey Cochrane came along and took the job away because he was even better. 

              After 1926 we are in the Cochrane-Hartnett-Dickey era, although Dickey didn’t come up until a little later.   But Cochrane, Hartnett and Dickey were not just the latest in the line of catching greats; they were better than the players before. 

              If you take these lists of catchers and look them up, you’ll decide that some of them aren’t that impressive and maybe shouldn’t be on the list.   Some of the catchers on these lists played 80 or 90 games and hit just .250 or .260 with just a couple of home runs.   They don’t stand out.   You only realize that they actually DO belong on the lists when you try to find somebody better, and realize that there isn’t anybody.   If you played 115 games and hit .280 with 12 homers and good defense in that era, that might make you the best catcher in baseball.

              Hartnett, Cochrane and Dickey aren’t like that; they are a step forward from that, and not just because the hitting numbers are bigger in the 1920s than in the era before then.   They are larger figures relative to the game.   Cochrane, more than anyone else, ended the tradition of catchers batting eighth, because he was just too good a hitter to bat eighth.  So was Hartnett. 

              Our rating standard adjusts for the batting stats of the era.   There are no more Win Shares in 1926 than in 1906, although the hitting numbers are different, but that doesn’t matter because everyone is compared relative to the era.   But there is both a rank here, which I have been giving you—Gabby Hartnett ranks 4^th in 1924, 1^st in 1925 and 2^nd in 1926—and a ranking number, which I have not given you, but for Hartnett it would be 17.45 in 1924, 20.49 in 1925, and 18.48 in 1926.  

              Hartnett, Cochrane and Dickey rose to a higher level than any previous catcher.   O’Farrell was the highest-ranking catcher of 1924 with a rating of 18.8, but in 1935 Cochrane was at 27.6, Hartnett at 26.4 and Dickey at 24.2.  They were all three over 20 every year for years.  The OVERALL numbers don’t go up; the numbers for second basemen or left fielders don’t go up, but the numbers for catchers DO go up because these three were just better than anybody before them.   That’s the point I was trying to get to.  Hartnett and Cochrane took the jobs away from catchers who were among the best in baseball, not because it was their time but because they were just better. 

              These are the top catchers of the years 1927 to 1943:

              There are very few times in history where three players dominate a position together for a long, long time, much more than a decade, the way that Cochrane, Dickey and Hartnett dominate the catching position. Rudy York in 1937-1938 is very, very much like Gary Sanchez in 2016-2017.   York, on the bench until mid-July 1937, had a phenomenal hot streak and emerged as a sudden star; he hit 26 homers and drove in 77 runs in July and August of 1937, and he wasn’t even playing every day the first half of July.   In 1938 he was in the lineup every day and was impressive with the bat, hitting 33 homers and driving in 126 runs. 

              He was a right-handed power hitter with BIG power, but he wasn’t a defensive asset as a catcher, so he’s similar to Sanchez on a lot of different levels, also similar to Earl Williams 1971-1972 on several levels.   After struggling through the 1939 season with York behind the plate the Tigers finally gave up on him as a catcher, moved York to first base and moved Greenberg to the outfield so that they could get both of their bats in the lineup, although Greenberg wasn’t really an outfielder, but he had to play the outfield anyway and drove in 150 runs and won the MVP Award.  So Sanchez and Judge are in a sense an echo of York and Greenberg.

9.  Third Stage Explanation of the Process

              I’d better explain how I’m deriving the rankings; this will be a little long-winded, but you don’t have to read it if you don’t care about the details.   We start with Win Shares, which give each player credit for playing time and for batting, but also for playing a key defensive position and for fielding. 

              I have a spreadsheet encyclopedia of my own creation that I use to study baseball, actually several of them, but one of the main ones has not had Win Shares integrated into it, which has always been problematic for me.   I haven’t addressed the problem because I have in my head a design for the step beyond Win Shares, which is Win Shares and Loss Shares, and I thought I shouldn’t invest the time to type 50,000 Win Share entries into the spreadsheet because I was moving on, but then I never find the three or four months that it would take to write out a full explanation of Win Shares and Loss Shares, so then I am trapped without either the old system or the new system. 

              Once the season ends I spend a couple of weeks of solid work updating the old Encyclopedias for the new season’s data.  While I was doing that this year, starting in October, I had the thought that while I couldn’t type ALL of the Win Shares data into the spreadsheet, I could put ENOUGH of the data into the spreadsheet that I could estimate the rest of it based on what I have.   The "estimated" data would not have full access to fielding data, to team wins and losses or to Park Effects, but it would have full knowledge of hitting stats.   I use Season Scores a lot, you know; Season Scores are like Win Shares but without adjustments for Parks, League norms or Fielding.   I realized that I could develop an algorithm to translate Season Scores into Estimated Win Shares.   I thought that if I typed 3 to 5% of the Win Shares data into the spreadsheet—2,000 to 3,000 data points—then I could estimate the rest of the data, and this would be useful to me in doing many different kinds of studies. 

              So I typed the data in, and the predictive algorithm works beautifully sometimes and doesn’t work too well sometimes.   The spreadsheet doesn’t have data for which league a player is in; it’s one of those poor decisions I made 20 years and am still living with now because after the fact it takes forever to fix it, you know.   I don’t have the LEAGUE in there, but I do have the years, of course, so I could easily enough put in the major league runs scored per game for each year, then put in all of the data from 1930 and 1968 and a few other years and study how much the estimates of Win Shares are too high for 1930 and too low for 1968, etc., and adjust the estimated Win Shares for the run environment of the season, so I put in those Win Shares and did that. 

              Then I studied the extent to which the estimates were too high for first baseman and too low for shortstops and catchers, etc., and adjusted the estimated Win Shares for that.   Then I studied the estimates by age; once I had enough REAL Win Shares in the spreadsheet, I could study it and see that young players tended to have 5% more actual Win Shares than estimated Win Shares, whereas very old players (37 and up) tended to have 8% fewer actual than estimated Win Shares, so I adjusted the estimates upward for young players and downward for old players.  

              I wound up putting a month into this project, and I wound up typing a little more than 12,000 seasons of actual data into the spreadsheet—about 22%-23% of the data--but by doing this and repeatedly improving the estimates, I wound up with Estimated Win Shares that are good enough to be used as stand-ins for actual Win Shares, mixed in of course with some actual Win Shares.   Also, I put in the ACTUAL win shares for anyone who was in the Hall of Fame or anyone who ever won an MVP Award and anyone who had a notably high estimated Win Shares without winning an award, so the players we will talk about most in these studies—the Cochranes and Dickeys and Hartnetts—those are all rated based on actual rather than estimated Win Shares. 

              OK, so then I had a Win Shares number for each player in each season.   The next thing I did to rate the players, trying to figure out who was the fourth-best first baseman of 1952, was to figure ESTABLISHED Win Shares for each player after each season.  (The fourth-best first baseman of 1952 was Luke Easter.)  An Established value. . .this is an old concept of mine that goes back to the 1970s, although I don’t know that anyone else ever adopted it or ever uses it. . .an Established Win Share level is a multi-year performance level.   There are different formulas that I use at different times, although they all amount to the same thing; about 50% of where the player rates now is based on what he did in the most recent season, and about 50% is based on where he was before the season.  

              For this study, I used this process:

              1)  If a player had played less than 750 major league games by the end of the season in question, his established Win Share level was based on 3 times the most recent season, plus 2 times the previous season, plus 1 times the season before that, divided by six.  If a player’s Win Shares over the last three seasons were 10-20-30, this would yield an established Win Share Level of 23.33 [(10 + 40 +90)/6], whereas if his Win Shares over the last three years are 30-20-10, then his established level would be 16.67 [(30 + 40 + 30)/6)].   The most recent season counts most heavily, but the previous seasons still count. 

              2)  If a player had played 750 or more major league games, then I would figure the established value by the same process, but with four seasons rated 1-2-3-4, rather than three seasons rated 1-2-3.

              3)  However, no player’s Established Win Share level can be less than 75% of his most recent season’s Win Share level.    So if your Win Shares over the last three years are 0-0-30, that would be an Established Win Share level of 22.50, whereas by the previous formula it would be 15.00.   That’s basically a "rookie protection rule".   For a new player we have SOME doubt about whether he is really as good as he seems to be, but not 50% doubt, if that makes sense.   If a rookie hits 30 homers and drives in 100 runs, he might have a real performance level of 22 homers and 75 RBI, but he’s not that likely to be a guy who hits 15 homers and drives in 50. 

              OK, that’s the process I have used for years to rate players.  I added a wrinkle this year, which is what I will call the Giancarlo Stanton Rule.   In rating Top Tens for the MLB-TV show last year, I left Giancarlo off my list of the top ten right fielders; I think I did, anyway.  If I had him rated, it was pretty low.   (In my defense, I think I was the only one of the experts on the show who listed Jay Bruce among the top 10 right fielders.)  Anyway, obviously excluding Giancarlo was a bad decision, so why did I make that bad decision?

              Injuries.  Giancarlo played only 74 games in 2015 and only 119 in 2016, so I had him down as a guy whose value was limited because he didn’t play that much. . . .obviously a poor decision on my part, in retrospect.   What we are asking is "How good a player is this guy, really?"  That’s not asking how good was he last year or how good was he ten years ago, but how good is he, right now? 

              Injuries are relevant and you can’t ignore them, but you don’t have to adjust for them 1-for-1, either.  So what I did is, when I figured each player’s Established Win Share level, I also figured his established Games Played level, by the same method.   Then I adjusted each player’s numbers upward by giving him credit for:

              1)  One-half of his games missed, or,

              2)  50% of his games played, or

              3)  20 games.

              Whichever is the smallest number.   In other words, if a player had an established Games Played level of 140—missing 22 games per season—I would project his Established Value Level up to 151 games (missing 11 per season), giving him some compensation for value lost to an injury.   But this compensation is limited.  If a guy has an Established Games Played Level of 70, missing 92 games per season, you can’t give him credit for 46 additional games, because that’s a speculative projection; 70 games just isn’t the same as 116.   You can’t be projected into more than 20 games per year above what you have really been playing.   And if a player has been playing 10 games a year, you might project his value into 15 games a year, but that’s it; we’re not creating phantom value by supposing that small sample sizes can be projected onto a large screen.  

              So, for illustration, Mickey Cochrane in 1932.   Cochrane’s Win Shares in the four relevant seasons (1929-30-31-32) are 27, 31, 28 and 30.   Those are sensational numbers; 30 Win Shares is an MVP-Candidate level of performance.   Anyway, that makes an established Win Share Level for Cochrane of 29.30.   His games played in those four seasons were 135, 130, 122 and 139, so that makes an established Games Played level of 131.70.

              They played 154 games a year then, so Cochrane was missing 22.3 games per season.   We’ll project him upward by one-half of that, or 11.15 games per season.   We’ll project him upward from 131.7 games per season to 142.85.   That increases his Established Win Shares from 29.30 to 31.78, so his ranking number for the 1931 season is 31.78.   This is the highest number posted by any catcher in the 44-year period 1900 to 1943—thus, were I treating that as his "value", Cochrane would be the #1 rated catcher of the 1900 to 1943 era.

              But that’s not exactly what we are doing.  What we are doing here is counting the number of years that each player is the top player or a top player at his position.    For that, we need another set of formulas. 

10.  The Other Set of Formulas

              One can say in general that in 1900 Mike Grady was the best catcher in baseball, and then Johnny Kling, then Bresnahan, then Chief Meyers, then Schalk or Schang, then Bob O’Farrell, then Mickey Cochrane, then Bill Dickey.  This is in general a valid statement, but what do we do with it?  Johnny Kling is not the equal of Mickey Cochrane, because they each could be described as the best catcher of their era. 

              I will note that by the method I outlined above a rookie is "double-protected", so rookies actually can rank near the top of their position after just one great year.   Not many rookies rank first at their position after one year, but quite a few rank second; Aaron Judge ranks second, I think.  A rookie is protected first of all by the 75% rule—a player’s established value cannot be less than 75% of his most recent season—and second by the games adjustment.   If a rookie plays 160 games that’s an established Games Level of 120, but we project it up to 140,so the rookie winds up ALMOST even with the veteran player; not quite, but close.  

              The other set of formulas summarizes where a player ranks over a period of years into one number, which is his YOPDI, or Years of Position Domination Index.  

              As I said before, a YOPDI is, I think, a reasonable thing to look at in a Hall of Fame debate, but we don’t want to over-value it, either.  It is reasonable and important to ask "Was this player the best player in his generation at his position?", but Roberto Clemente, for example, has a very low YOPDI.   Roberto Clemente hit .351 with 23 homers in 1961, which is a hell of a year, but comparing him to other right fielders in 1961, he is always going to rank below Henry Aaron.  The two MVPs in 1961 were both right fielders, Roger Maris hitting 61 homers, driving in 142 runs, and Frank Robinson hitting .323 with 37 homers, 124 RBI.   I think actually the top-ranked right fielder after the 1961 season may have been Rocky Colavito, who hit .290 with 45 homers, 140 RBI, so Clemente, despite his tremendous season, is way down the list.   In other years he is behind Al Kaline or Jackie Jensen or Tony Oliva; Kaline, Clemente, Aaron and Maris were all born in the same year, Colavito the year before that and Frank Robinson the year after.  

              Clemente, in his career, ranks very poorly in terms of how much he dominates his contemporaries at the same position, but that doesn’t prove he wasn’t a great player; it’s just an unusual situation in which there are several truly great players at the same position at the same time.   Bert Campaneris, on the other hand, does extremely well in this area, because he happens to fall into a hole in which there just aren’t any great shortstops.   Bert Campaneris will have a far higher Position Domination Index than Clemente, and that reflects a real underlying truth, but it doesn’t prove that Bert Campaneris was a greater player than Clemente; it just happens. 

              So here’s what I did.   There are three principles that guide the YOPDI (Years of Position Domination Index).  Those are:

              1)  That the #1 player at the position always receives 10 points, whether it is 1901 or 2017,

              2)  That the number of players receiving points is always one-third of the number of teams, rounded down, and

              3)  That the ratio of points to players is constant over time. 

              In this era, when there are 16 teams, I give

              10 Points to the #1 player at the position each year,

              7 points to the #2 player,

              4 points to the #3 player,

              2 points to the #4 player, and

              1 point to the #5 player. 

              So, for example, Bill Dickey so far has been the #1 player at his position than 7 times, the #2 player 3 times, the #3 player 4 times, and the #4 player once.    That’s 109 points—7 times 10 (70), plus 3 times 7 (21), plus 4 times 4 (16), plus 1 times 2 (2).  Dickey through 1943 has 109 points on the Years of Position Dominance Index.   Through 1943, these are the leaders:

              Schalk having 43 points rather than 42 has to do with their being 24 teams in 1914 rather than 16; don’t worry about it, and I’ll explain it later.   Dickey’s total would grow to 111 before he retired.

11.  A Brief Note about Hall of Fame Standards

As a generalization we can say that

              a)  A Hall of Famer should be above 50 points on the YOPDI scale, and

              b)  If a player is above 65, he is usually a Hall of Famer. 

              Not Hard-and-Fast rules; just a generalization.   65 points is "six years as the #1 player at the position, or equivalent performance in multiple seasons near the top of the list.   50 to 65 points is kind of a gray area.  

12.  A Word About the 19^th Century

              You know, I don’t recognize 19^th century baseball as major league baseball, because it has no characteristics consistent with what we call major league baseball, and I don’t really care about 19^th century players.   Also, if you are going to study 19^th century data with a process like this you have to constantly fiddle with the inputs because the number of teams goes up or down almost every year, so the number of points you would award every year has to go up or down, and, as I said, I don’t really care about it, so I didn’t figure all of that out.

              However, I do have Win Shares for 19^th century players, and we can sort of generally state how those players would rank if evaluated by these methods.   The top 10 catchers from 19^th century baseball are:

1.  Buck Ewing, Hall of Fame (1880-1897).  Ewing had six seasons as the number one catcher in baseball,  nine seasons as one of the top five, and would have about 67 YOPDI points.   The only 19^th century catcher who is in the Hall of Fame, or deserves to be. 

2.  Mike Grady (1894-1906).   Two seasons ranking as the #1 catcher in baseball and four seasons as #2, eight seasons among the top five.   About 53 YOPDI points.   If I were to tell you that I have learned one thing so far from doing these rankings, that one thing might be that Mike Grady is a seriously overlooked star.   He appears to have been a real quality player that nobody writes about.   Here is a two-year gap in his career, 1902-1903; I would guess he might have been in the Western League in those years, maybe.

3.  John Clapp (1876-1883).   Three seasons as the #1 catcher in baseball, about 51 points.

4.  Charlie Bennett (1878-1893).  Three seasons as the #1 catcher in baseball, about 49 points.   Bennett was a popular player whose legs were cut off in a train accident which occurred near where I live.   Tiger Stadium was originally named Bennett Park in his honor.

5.  Duke Farrell (1888-1905). One season as the #1 catcher in baseball, but an impressive twelve seasons among the top five, about 47 points.

6.  Deacon McGuire (1884-1912).  Four seasons in the 1890s as the #1 catcher in baseball, about 44 points. 

7.  Doggie Miller (1884-1896).  One season at the top but eleven seasons among the top 5; about 42 points. 

8.  Fred Carroll (1884-1891).  No seasons as #1, but five seasons as #2, and about 37 points.

9.  Jack Clements (1884-1900).  The last left-handed catcher; best left-handed catcher of all time.  Six seasons among the top four catchers in baseball, about 34 points on the YOPDI scale.

10.  Ed McFarland (1893-1908).   Two seasons as #1, about 31 points.  

              In addition to those ten, I will mention two other Hall of Famers, King Kelly and Wilbert Robinson.   Kelly was a fantastic player and a colorful player, but not really a catcher, and never even a semi-regular catcher until after his best seasons.   Robinson, Uncle Robby, is in the Hall of Fame as a manger and was a decent player and around for a long time, but his YOPDI is zero.  He was never one of the best catchers of his era.  

13.  Ferrell and Lombardi

              Rick Ferrell was in the majors from 1929 to 1947; Ernie Lombardi was in the majors from 1931 to 1947, and both are in the Hall of Fame.    Lombardi ranked as the #1 catcher in the majors one year, second once, third once, fourth three times, fifth twice, a position dominance score of 29.   That’s about half of what a Hall of Famer should have.   Ferrell never ranked higher than fourth (twice), and ranked fifth five times, a position dominance score (YOPDI score) of 9.  

              The other way to look at it is the player’s peak level of Established Win Shares, game-adjusted as explained before.   Ferrell’s peak was 18.83, after the 1934 season, which ranks 21^st among catchers, 1900 to 1947.   The Hall of Famers on this chart are highlighted in blue:

14.  The Cooper-Masi Era

              From 1944 to 1948 the best catchers in baseball were Walker Cooper and Phil Masi.   Cooper was more famous; his brother was the best pitcher in the National League, which created a story line, and he played in the World Series in 1942-1943-1944, hitting .300 in 16 World Series games.   Cooper was a little better player than Masi in terms of peak value, although neither was a Hall of Famer.  I’ll expand the chart to include the Rank Value, since I’ve created a frame of reference for that now: