Much has been written about the ever-increasing length of ballgames. Attention has recently turned to the impact of expanded September rosters on game length and “level playing fields for both teams”. Here I’m only focusing on the game length issue vis-a-vis September callups. Using game length data culled from the Retrosheet game logs for 1920 through 2015, here are the month-by-month average game lengths in minutes, by decade (March and October games excluded due to their relative small number of data points). The final column here, “APR-AUG vs. SEP” takes the average game lengths for the first five months and compares it to the average for September. Negative numbers mean that September games were shorter on average.
|DECADE||APR||MAY||JUN||JUL||AUG||SEP||APR-AUG VS. SEP|
|% change 1920s -2010s||52%||51%||51%||54%||59%||66%|
According to MLB official historian John Thorn, September roster expansion dates back to the 1910s. Also, Thorn notes that with regard to early season rosters, extra men were permitted for 30-60 days (with many variations) as early as 1901 and as late as 1990. Clifford Blau has a handy rundown of roster limits through the years here. April games were ever-so-slightly longer than May games from the 1930s through the 1960s.
Now, on to September. When we compare September to August (when rosters would be most comparable in terms of similar roster makeup), September games didn’t last any longer until the 2000s, and during the 1920s-1960s, games were actually shorter in September. One can therefore postulate that the advent of bullpen segmentation (LOOGY, ROOGY, seventh inning, setup men and closers) and increased reliever usage on the whole has driven game length increases since the 1970s (along with increased offense, commercial time and other aspects), and this exerts its most impact in September.
If you look at the last row, you’ll see that the largest increase in game length since the 1920s has occurred in September games. And looking at the last column, since the 2000s, games in September have for the first time lasted longer on average than the rest of the year. September rosters are littered with “extra” relievers, and it appears managers are utilizing them more than ever before. Here are the yearly total number of pitchers utilized since 1998 (when the Majors expanded to 30 teams). Look at the sizable jump from 2014 to 2015, and note that we’ve already set a record in 2016, with two weeks left in the season. Arguably some of this could be pitcher injuries, but I think September roster additions play a large part also.
For completeness sake, here are the total number of players utilized since 1998. The overall increase in recent years is smaller than for the pitcher subset, but its still an increasing trend.
Post-script: Here are the average game lengths for each month this season, since June:
|SEPT thru 9/12||189.1|