David Pinto brings the news that the Rockies are going with a four-man rotation. This is an interesting idea. I believe the current five-man plan employed by most teams during the regular season began in the 1960s. Though some teams experimented with different rotations, since 1980 the five-man rotation has been the norm. This article by Rany Jazayerli at Prospectus has short history. So, which is better: the four-man or the five-man? I will join the list of people who have attempted to answer this question.
First I want to identify the trade-offs of the choices. If you go with the four-man, your best pitchers will pitch more regularly than the five-man, but they may be tired and therefore pitch less per start. With the five-man you have to throw an inferior pitcher out there once every five games; however, each starter is able to throw more pitches. In Pinto’s post he cites that the Rockies plan to make the trade-off by limiting starters to lower pitch counts. So, basically going to a four-man rotation has the effect of switching a certain number of middle innings from starters to relievers. I am not sure if starters are more valuable in the 5th-7th innings than in innings 1-4, and I am not sure to figure out which is more valuable.
For simplicity I am going to assume that all pitcher innings are the same. This is not such a bad assumption given that most starters will be relieved when their performances drop below bullpen replacements. Therefore, I will focus on whether or not starters are pitching more or less innings with the five-man than with the four-man. Using a basic pitch count estimator (BPCE = 3.3 x BIP + 4.8 x K + 5.5 x BB) I estimate the average pitch counts for pitchers who started at least 10 games in a season from 1950-2003 (excluding 1981 and 1994).
The five-man count appears to be associated with a jump in pitch counts in the late-1960s through the mid-1970s. There was a slight dip in late-70s and early-80s, but now pitch counts are back to all-time highs. The five-man rotation is associated with high pitcher counts per game than the four-man, although it is not that large of a difference (approximately 10 pitches per game). This is not surprising; however, over a season how does this affect overall innings pitched by starters? The following figure graphs average total innings pitched per season for the same group of pitchers.
Interestingly, though starters are pitching longer with five-man rotations, they are pitching fewer overall innings. I suspect that much of this has to do with increased reliever usage for other reasons, but the difference in average total innings pitched from the early 1970s and the present is about 30 innings. If starting pitchers are better than non-starters and the decline in starter innings is due to the five-man rotation (I acknowledge these are two big “ifs”), then the four-man rotation may be a better way to configure your pitching staff in a season. On the other hand, fewer innings per season may prolong the entire career of a pitcher, which might offset the loss in one particular season. But really, what incentives do managers have to increase the career length of his starting pitchers by a few years? I suspect mangers are very myopic in their use of players given their incentives. I look forward to seeing how the Rockies experiment goes.