It is easier to find four starting pitchers than five.

– Earl Weaver (from Weaver on Strategy, 1984)

Much discussion these days is given to a team’s lineup choices. But what about the other side of the ball — that is, the pitching rotation? What is the optimal way to order the pitching staff?

With Adam Wainwright, Chris Carpenter, Kyle Lohse, Todd Wellemeyer and Joel Pineiro slated as the Cardinals’ starting rotation having been "set well before a game is played," we’re tempted to play the old Sesame Street game: Which of these things is not like the other? Let’s look at their projected ERAs (averages of Bill James, CHONE, Marcel and ZiPS systems):

From their projected ERAs alone, the anticipated rotation order does not appear to be optimized. Furthermore, the Cardinals will likely continue an approach that has become the norm for three decades, but which does not seem to be an optimal way to use starting pitchers: The five-man rotation, which includes, of course, Pineiro.

Several years ago, Baseball Prospectus’s Rany Jazayerli wrote a compelling argument for the old four-man rotation, contending that the five-man confers neither the assumed health benefits nor improved performance on the field. In his research, Jazayerli found that one major reason that the five-man rotation fails is simply because it takes away starts from better pitchers and gives them to worse ones. Here’s his chart of average starts by rotation slot, using 1973 and 1999 as the examples:

Roughly eight starts were taken from a team’s top three pitchers and given to fifth or worse starters. So the question then is, does that distribution make a difference in the Cardinals’ case? To find out, we’ll have to estimate how many innings each starter will pitch at each slot. First, let’s do the numbers for the projected rotation based again on their average projections for Games Started and Innings, which will give us innings per start. We’ll assume that the mysterious sixth man will have an ERA worse than Pineiro (which we’ll put at 5.00), though admittedly, that is a big assumption (after all, Brad Thompson figures to have a 4.54 ERA).

Now that we have innings per start, we can apply those numbers to the number of starts each pitcher will make in each rotation, based on those 1973 and 1999 numbers:

Now we can estimate Wins Above Replacement (using the indispensible Sky Kalkman spreadsheet). First, let’s do the anticipated five-man rotation:

All right, not bad with 12.2 WAR. But what if the Cardinals reordered their starters by expected ERA and went with a 1973-style four-man rotation?

The four-man reordered rotation would yield 12.9 or .7 more wins that the current group. While an improvement, it’s not a very significant upgrade. Why? It would seem that the team lacks an ace pitcher who can work deep into games. Carpenter may offer excellent ERA, but he figures to go only 5.5 innings per start, whereas the league’s top pitchers, like Dan Haren and Tim Lincecum, lasted an average of 6.5 or more in 2008. However, on second look, Wainwright may actually fall into that category; his projections give him an unrealistically low IP/GS because he was a reliever in 2006. What if we assumed Wainwright were able to continue his average from 2007-08 of 6.4 IP/GS?

Now we’re talking, with 13.4 WAR, an improvement of 1.2 wins. It’s still not earth-shattering, but that’s about the difference that finagling with a lineup can make on the offensive side, so it’s worth considering. For all his cogitating over who is going to bat ahead of and behind Albert Pujols, Tony La Russa may want to spend some time on the pitching side of planning.

This entry was posted by Pip on Tuesday, March 24th, 2009 at 1:27 am and is filed under analysis. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.