Historical Time-of-Game Data
Joe West’s recent comments prompted me to look at some data I’ve always been curious about. We all know (or think we know) that games have gotten longer, but exactly how has game duration changed over the years?
The Retrosheet Game Logs contain time-of-game data, but the data is very incomplete prior to 1956. (There are also a few data errors, for example, a game that supposedly lasted 413 hours in 1947. I considered everything over 500 minutes to be a data error and ignored it.)
The graph below shows the average game time through the years. All games, extra-inning and less than nine, are included.

What surprised me here was the huge increase from 1944 to 1955, where average game time increased 32 minutes over just eleven years! What happened? More pitching changes was a factor, but I don’t think that can account for it all.
After 1955, things settled down, then in 1979 times started to zoom up again, reaching a new plateau in the late ’80s.
I wanted to look at normal, nine-inning games, but it’s not easy to filter those from the Retrosheet data prior to about 1955. The Game Logs contain the number of outs in the game, so my definition of a normal, nine-inning game is one with 51 to 54 outs. The graph below shows the data for nine-inning games, giving a little more insight into the more recent trends.

Actaully, the average game time has been relatively stable for the last twenty years. Still too long IMHO.
