Sidebar... It's a pet peeve of mine with baseball and prediction models that use run differential as a projection tool. How can a model accurately predict anything when a team could go 6-1 in a week, winning 6 games by 2 (+12) runs each game and getting blown out by 10 runs (-10) in the one loss, have only a +2 run differential and say they're playing essentially at a .500 clip? Run differential alone is the worst predictive model IMO. I think because these kinds of things even out over time. Teams will go through streaks of losing close games, or blow someone out by 10 runs. And thus it just sort of evens out in the end. The more games you play, the more 'accurate' run differential tends to be. It's not perfect, but it's generally pretty accurate in the long run. If the same players were used in wins and losses, then this makes sense. So I'll buy the runs scored half of run differential since the offense is basically the same every game win or lose. it's crazy to assume that "runs against" are basically the same players pitching in wins and losses, however. Crappy bullpen guys who rotate through the back half of the bullpen throughout the year as well as POSITION players are used in blowouts, and those blowouts losses anyway are extremely unlikely to be started by any of our 3 aces.