I've seen analyses that say that overall, sac bunts don't increase your odds of winning. However, I've never seen an analysis of just the ninth inning and later in a tie game. I would like to see it, because intuitively, it seems like when manufacturing one run will win the ballgame, it might make sense to trade an out. I'm sure something that obvious has been controlled for in all the data, however.
Exactly. That intuition certainly battles against the prevailing math. I think James basically makes the point that outs are worth x and moving a runner from first to second etc is worth y, and that it's across the board not worth it. But what about when the out supposedly worth 'x' is the pitcher? His out is worth a lot less, right? What about the 9th inning? I'm not convinced.