Wednesday, May 21, 2008

Extra Innings, Part Deux

A little while back, I posted on extra inning baseball games and how two teams can end up in a scenario that neither like and both would probably agree to avoid ex ante if such a possibility were possible. After posting, Frank Stephenson, of Division of Labour fame, shot me an email that he liked the cut of my jib, and sent a dataset of all of the baseball games from the 2007 MLB season. I like baseball, I'm an economist, I do empirical work...clearly, this was a fantastic email.

So I set out to prove what I'd put forth in the initial posting. Here's what I got:

- The best result is that teams are less likely to play an extra inning game if they have a game the next day. This holds true with home and visiting teams, as well as if they are playing the same team or a different team. In other words, teams are most likely to play an extra inning game if they do not have a game the next day.

- The number of pitchers used in a game is not a function of whether the team has a game the next day or not. Well, it is, but it's positive, but that's against the theory put forth. It's the same for visiting and home teams, as well as whether they play the same opponent or a different one the following day. Perhaps teams would rather use a large number of relievers in short roles than a few in long roles?

- Nothing comes from analyzing the length of extra inning games; models alternate between positive and negative, and all are insignificant. One could argue that this lends a bit of credence to my extra-innings-are-perpetual-short-run-scenarios argument. If they weren't, teams would want to reduce this number as much as possible. Or, maybe extra inning games really just don't have that much of an impact on teams the following day. I'm a bit hesitant to accept that, though.

Thanks to Frank for the data! If anyone has any spare data lying around, by all means, send it along...

1 comment:

Bryce said...

It's interesting that extra inning games are more likely prior to a day off. I would be interested in seeing if this is a consistent trend over a number of years.