Add new comment

Not to the place saberdogs would like, but you'd have really useful information ... it is exactly the "vs expectancy" paradigm you describe that is the only key (I know of) to unlock these kinds of single-season puzzles ...
Of course if you start trying to analyze them across multi years then you get the problem of the Patriots "exceeding" 12-4 expectancy, then failing to "exceed" 13-3 expectancy ... and usually when studying platoon RH hitters vs LHP's, we need much more than a single season's worth of data ...
Not questioning your approach amigo - it's the one I'd use - just quibbling for the benefit of the few who don't quite get how complex such a problem is ...
........
Any thoughts on how you would attack the confounding variables also?  You've got it pretty well covered in terms of how the players themselves would be affected by their roles, but as to how the OTHER players on the roster are affected by a player extra or less, the $$ question that confounds, etc...  Benihana posits a 2-3 WAR loss, right off the top, based on the invested roster spot -