I’m a couple of days late on this one, but Colin tested run estimators at Statistically Speaking last Friday. Instead of testing each estimator against the number of runs a team scored in a season, he compared run estimators to the number of runs a team scored in specific innings.
This is a great idea, because the differences you see in innings are more like the differences you see between batters (the differences between teams in a season tend to be much less extreme). Colin’s conclusion is no surprise: Base Runs is the best formula. This is an important post that should be getting a lot of attention.