Tango: A brief history of WAR

From Tom Tango at TangoTiger.com on December 20, 2013, with mention of SABR members Bill James and Pete Palmer:

The kernel of WAR goes back to most clearly in Bill James’ Baseball Abstract 1987, when he compared Roger Clemens to Don Mattingly 1986 seasons (as well as Rice/Guidry in 1978).  It was one of those things where the end result wasn’t the important part, but rather the process he used to answer the question.  You can even trace the idea of WAR a bit earlier to around 1982 or 1983, when Bill would ask “what are the odds that a .400 win% player put up this player’s estimated Won-Loss record”.  It was brilliant in its question, because it is in effect using regression toward the mean and replacement level at the same time.  Bill I don’t think appreciated the genius of it.  And I think most people have even forgotten about it.  Just a saber-historian, like Patriot, would really appreciate what Bill did. Again, the process was more important than coming up with the exact formulation.

Pete Palmer also almost got there, but he was so focused on WAA (wins above average), that he was just one step away from adopting Bill’s idea. 

My next exposure to the idea was from Clay Davenport, via WARP.  But Clay made a huge misstep, one that he simply did not accept as seriously as the rest of us (that he would do performance above replacement offense, plus performance above replacement defense).  He eventually did, but that was many years later.  It was frustrating because Clay has all the skills and temperament to have produced a state-of-the-art WAR structure.  But he held on too hard.  As a result, we’d end up with conversions of wins to dollars that were horrible to consider.  His about-face in one of the later BPro Annuals was wonderful to behold, in not only recognizing and admitting he was wrong, but then correcting himself.

Read the full article here: http://tangotiger.com/index.php/site/comments/history-of-war

This page was last updated December 20, 2013 at 11:33 am MST.