clock menu more-arrow no yes mobile

Filed under:

The Changing Landscape of WAR: 5+ WAR Seasons, 1911-2010


(Click to enlarge)

Adam's fantastic wWAR series got me thinking about how today's star players have a much higher hurdle to clear to prove their Hall of Fame credentials. Now, there are lots of reasons for this, not the least of which has to do with the lack of uniform criteria amongst voters. However, I think another factor has to do with the changing landscape of the game.* When looking at position players, peak performance is much harder to spot due to the decreasing distance among elite players.

If we look at all position players that have posted 5+ WAR seasons (using Rally WAR here) since 1911, we see far fewer seasons occupying the right tail (i.e. >=10 WAR) and less variation amongst seasons the closer we get to today's game. The graph above (a screen shot from the interactive visual below the fold) plots the frequency of 5+ WAR seasons by decade in increments of .5 WAR.

Let's take 1921-1930 and 1991-2000 as one example. The former decade witnessed the rise of Ruth and the power game. The latter decade is remembered as an offensive renaissance. While both eras saw their fair share of 5+ WAR seasons, those seasons were not distributed in the same fashion.

During the 1920's 89% of all 5+ WAR seasons fell between 5 and 10 WAR. Compare that to 98% during 1990's. If we look at the right tail of the distribution we see that 19% of all 5+ WAR seasons during the 1920's included 8+ WAR seasons, while the 1990's only saw 10% of their total rise above 8 WAR. Furthermore, 11% of 5+ WAR seasons (14 seasons) during the 1920's included seasons of 10+ WAR, whereas the 1990's only saw 2% of their 5+ WAR seasons (4 seasons) hit that mark.

Comparing other decades yields similar results.

What is also interesting is that the the standard deviation amongst those 5+ WAR seasons has steadily decreased since the 1910's. During the 1920's, the standard deviation for 5+ WAR seasons was 2.1; for the 1990's, that number shrank to 1.22. If we look at the number of players elected to the Hall of Fame in terms of the decades in which they played the pattern is quite similar.

Now, partially this is the Babe Ruth affect. During the 20's, Ruth owned 7 of the 14 10+ WAR seasons (Hornsby had 5 and Gehrig the other 2). Moreover, Ruth posted 12+ WAR four times during the 20's. Barry Bonds is the only player since Joe Morgan in the 1970's to post a 12+ WAR season (he did it three times during the 2000's).

So it appears there are a few things going on. First, we have generally witnessed the decline of extreme outlier performances among position players since the 1920's. Second, the variance between great players--and great seasons--has seemingly decreased over time. Combined, you end up with more players putting up 5-10 WAR seasons, less putting up 10+ WAR seasons, and great players lacking separation from each other. The less distance between performances lends itself to judgements of greatness that are more likely to be influenced by single moments and observational bias (i.e. "clutch plays", playoff performance, etc).

The past 10-20 years did provide something of a resurgence in outlier performances and variance amongst seasons, but that was mainly due to one player--Barry Bonds. So we end up with great offense performances being downgraded historically because of the lack of variation, and when players do perform as outliers that performance is more and more explained away as "chemically enhanced".

That being said, if we are interested in a Hall of Fame that rewards those players that dominated during their own eras, the WAR data presented here would suggest that position players have become progressively less dominant over time.

(*This might seem obvious to some, but as a relatively new saberist and someone that has to see things for themselves I had to look into the numbers myself)


Below is the interactive version of the WAR frequency data. The first tab represents 5+ WAR frequency as a percentage. The second shows the raw count data. You can highlight multiple years using CTRL. You can also isolate specific years by using the check boxes.

Raw data by decade can be found in this folder.