Beyond rating players, there is a potential analytical value in creating stable measurements.   Assessing a player’s value with a moving average, only 10% in the focus year, creates values that cannot jump up and down.  Stabilizing an element has many standard uses.  We stabilize explosives by adding inert ingredients.  Same with medicines, chemicals of all kinds.  Sometimes stabilizing them in the only way you can work with them.

            So I had the idea that, since this method I have been writing about all week stabilizes a player’s value, perhaps there would be other uses for that.  One such notion was that perhaps we could use this to address the problem of . . .

            How do we describe this?  Relative dominance to the era? 

            Some eras in baseball history are more dominated by stars than are other eras.  The reason for this, of course, is that the competition gets stronger over time, thus harder to dominate.  Lower levels of competition are dominated by the best players in the league.  As the competition gets stiffer, the ability of any one player to dominate the league becomes less.   But in comparing players across time, how do you adjust for that?

            Speaking in specifics, the highest-rated player since 1900, by this method I have been using, is Babe Ruth, 1923, at 40.25 Win Shares per season.  He is followed by Honus Wagner, 1906 (38.72), Ty Cobb, 1914 (36.41), then Barry Bonds, 1997 (35.68).  Bonds is "modern", sort of, 25 years ago now, but he is followed by Mays, Mantle, Tris Speaker, Stan Musial, Aaron, Cy Young, Lou Gehrig, Walter Johnson, Eddlie Collins and Mel Ott (one listing for a player, of course.  Otherwise, about 7 of the highest scores ever would be different Babe Ruth years.)  In the top 16 players, only one—Bonds—is from the last half-century.   One should know intuitively that that can’t be right, that almost all of the greatest players in history cannot have retired a long time ago. 

            We know that that can’t be right, but how do you adjust for it?  Well, here’s an idea; I will warn you in advance that it doesn’t quite work, but it works well enough to suggest that it MIGHT work with more refinement, more effort, more study. 

(1)   Collective Win Shares are a relatively stable measure over time.  There are 3 Win Shares per team win all across baseball history.

(2)  But the highest Win Share totals of all time are concentrated in the 19^th century, the highest since 1900 concentrated in the 1900 to 1930 era.

This process creates stabilized Win Shares over the course of a career.  Thus—since there are NOT more Win Shares per player in 1915 than in 1985—if the highest numbers are higher, it must be because the game is more dominated by its stars.  We can measure that, perhaps, by isolating the averages for the dominant players in each era, and creating era-norms not for the average player, but for the best players. The "dominant" players we will say are 1 per team.  From 1900 to 1909 there were 152 seasons.  We focus, then, on the 152 best 19-year averages from within that era—10 of which were posted by Honus Wagner and 10 by Nap Lajoie, 10 by Sam Crawford, many by Cobb and Cy Young and others, but just the 152 best one-season figures (each one representing a 19-year-window) from within the decade.  As I said, it doesn’t quite work, but it sort of works. 

We can call it a dominance level.  The 152 best players of the 1900 to 1909 era had an average score by this system of 22.69, meaning that the best player on a team in that era typically earned 22 to 23 Win Shares per season over a long period of time.  That’s the dominance level.  This number increased to 22.99 (22.9993) in the 1910s, the increase being accounted for by the lengthening of the schedule; the 154-game schedule didn’t arrive until 1904. 

You could adjust that games-played problem out of existence in a more careful study; this was just a first effort.  But since 1900, the dominance level has generally gone down steadily over time.   After peaking at 22.99 in the Ty Cobb decade (1910 to 1919), it dropped to 22.1 in the 1920s, up a hair to 22.2 in the 1930s, down to an all-time low of 19.8 in the 1940s, due to the war.   Fewer seasons by dominant players, many or almost all of whom spent two or three years in uniform. 

The level went up to 21.8 in the 1950s, and then up to 22.4 in the 1960s, due to (1) the longer schedule, 162 games rather than 154, and (2) expansion, which diluted the population a little bit.  Then in the 1970s it dropped to 21.0; in the 1980s, to 20.0 (20.05.) In the 1990s, due probably to steroid use by some players, it went back up to 20.6, and in the 2000s. to 20.2   It’s impossible to figure it (by this method) for the 2010s, since the data to make a long-window evaluation of players from that era is not yet in existence.

Babe Ruth in 1923 is marked at a multi-year level of 40.25, meaning that a typical season for Babe Ruth in that era is 40.25 Win Shares.  That’s against a dominance level for the 1920s of 22.137.  40.25 divided 22.137 is 1.8182, so Babe Ruth for that season is 82% better than the dominance level.    For the ten best seasons of his life, Ruth is 72% better than the (decade determined) dominance level.  

This is still the highest figure of all time, but the list now is quite different: 

By comparing the player to the era dominance level, Barry Bonds has moved up from fourth to second.  Stan Musial has moved up from six slots behind Ty Cobb (third vs. ninth) to one spot ahead of him.   In the top 20 most dominant players, we now have seven whose peak seasons were within the last 50 years—Bonds, Pujols, A-Rod, Schmidt, Morgan and Rickey Henderson.  Given an estimate of the rest of his career, Mike Trout would also be there, probably somewhere in the top ten.  

The problem is "dominance of the all time lists by players long retired."  As I said, this method doesn’t SOLVE that problem, but it makes a little progress against it.   Given more work and more refinements, we might have something here. 

Another thing we can do with the long-window stabilized career records is to take another look at aging patterns.  A player’s best season may be at age 27, but if his production at ages 27 to 30 is greater than his production at ages 24 to 27, then his peak would tend to measure at 28, rather than 27, so this approach places the center of a player’s career typically at 28.   This research studies all players in the years 1900 to 2014 who had a peak season of at least 10 Win Shares, and identifies the age of the highest multi-year value.   There are 4,209 players included in the group, of whom 619—15%--had their peak season at age 28.  This chart gives the number of players who had peak production at each age: 

The one player who had his peak at age 41 was Connie Marrero, a Cuban pitcher born in 1911.  Signing with Havana of the Florida International League at the age of 36, he went 25-6, 20-9 and 25-8 with ERAs of 1.66, 1.67 and 1.53 before getting a shot at the major leagues at age 39.  He had his best season with Washington in 1953, going 11-8 with a 2.88 ERA at the age of 41.

This chart adds the percentages of players hitting the center of their peak period, with the cumulative percentages:

As you can see, 84% of players have passed their peak by age 30.  One-half of one percent have their peak seasons after age 36.  Most of those are pitchers (17 out of 22), as are most of those who have their peak seasons at very young ages. 

I am certain there are other things we will find that we can do with these stabilized player values.  Thanks for reading.