The bottom line was, you'll recall, that there was a small, insignificant difference between the chance that a run scores in an inning when a sacrifice is executed and in an inning when a bunt could have been executed but wasn't (a so-called "bunting situation"). Likewise, there was a small but insignificant difference between the average number of runs scored in those innings.
As I mentioned then, additional data might help to clarify any differences (i.e. to give statistical significance to the differences), hence I went ahead and expanded the analysis to include all the games played in National League parks in 2012. I also revised my analysis somewhat: instead of looking at the total runs scored in the inning that included a bunting situation, I focused the look only on what happened after the situation (including the bunting/non-bunting play itself). Thus, no credit would be given for runs scored earlier in the inning.
There is always a danger with expanding a data sample in this manner: as the saying goes, with enough data you can prove *anything*. Collecting all the play-by-play for all of the games played in National League parks last year is A LOT of data (there were well over 100,000 plays made last season!) As we saw in the "small" sample of games at Nats' Park, the differences were small, and even if they were significant, they didn't seem really all that important. If a sacrifice costs the team a percent or two chance to score, it's sub-optimal, but is it really worth the angsty "STOP BUNTING!!!" yell?
Well, in any case, here are
There were 9336 Sacrifice Bunt situations. In 974 of them, a Sacrifice was successfully executed. Home teams executed 590 of these bunts (this is expected due to the way "the book" reads about late-game strategies).
Teams that executed the sacrifice subsequently scored in the inning 47.4% of time. Teams that eschewed the bunt scored 51.8% of the time. This represents a small, WEAKLY significant difference (at the 90% level).
Teams that bunted ended up scoring an additional 0.87 runs in the inning following the sacrifice on average. Teams that did not bunt, scored an average 1.02 runs in the inning following the non-bunt. This represents a significant difference (at the 95% level) and is, really, pretty substantial. NL teams averaged a bit more than 60 sacrifices last season; with an estimated difference (decline) in runs following a bunt of 0.15, that suggests teams may have ended up costing themselves around 9 runs in the season due to the practice.
Here's the revised distribution chart:
Again, it seems likely that one actually does the opposite of what really wants to do by bunting: score at least one run. This is because, as we already strongly suspected, it's also much less likely that a team will end up scoring more than one run, even if it appears that a team might be ever-so-slightly more likely to score *exactly* one more run.
Aren't that much different from before. I still think that the bunt is a sub-optimal strategy, but only slightly so. I'm not fond of the bunt but I won't pull my hair out too much when it's used, except when guys who are clearly better suited to hit the ball LONG and FAR do it.