September 5, 2012

Pebble Hunting

The Rockies' Rotation, Before and After

by Sam Miller

Printer-friendly

Contact Author

In June, you'll recall, the Colorado Rockies announced that they would be going to a four-man rotation, with each pitcher limited to 75 pitches. Josh Outman was the first pitcher to start in the new format, and the consequences of the Rockies’ shift were immediate: lots of people became aware that Josh Outman was pitching for the Rockies now. In the Seth Smith trade? You don’t say!

It’s still too soon to say what the four-man rotation—with 75-pitch limits on the starters—has wrought, and will wrou... uhh... whatever the heck the infinitive of wrought is. Work? It seems to be work. Wrought seems to be the past tense of work. Forget it.

What we do know is that the Rockies say they will continue this experiment into next season, so perhaps truth will emerge. In the meantime, we know this: the Rockies allowed 5.74 runs per game before the switch and 5.32 runs per game since. Done! Fixed! Go away Ervin Santana, teams need only four starters now.

Except that it’s obviously not that convincing. Some regression from 5.74 runs per game was already likely. The Rockies probably wouldn’t have made this move if they weren’t performing at a level far below expectations. They’ve also made plenty of personnel moves, so it’s a comparison between two different groups of pitchers. And 0.4 runs per game isn’t all that much. The Rays and the Mariners, for instance, will likely each have bigger first half/second half splits than that, when it’s all done. 

I was curious, though, to see where in the pitching staff the four-man format would bend or break. As Colin Wyers noted back in June, the rotation is just the front of the staff; any innings shirked by starters must be picked up elsewhere. So the Rockies plan could:

a) work, by removing the worst starting pitcher and by keeping starters’ workloads manageable and by keeping starters from getting hit the third time through the order, or
b) fail

and if it failed, it could
a) fail because the starters struggled on short rest, or
b) fail because the relievers picking up the middle innings would be worse than the starters, or
c) fail because the relievers picking up the middle innings would no longer be available in the late innings, and the relievers picking up the late innings would be worse and/or exhausted.

In other words: the Rockies could get worse early in games, or in the middle of games, or late in games. So, with the massive SSS caveat, let’s see how many runs, on average, they have allowed in each inning:

So there’s a table. The first two innings are basically a push. The next two benefit the post-switch Rockies. The fifth, sixth, and seventh are kind of a mixed bag, and the eighth breaks dramatically in favor of the early-season Rockies. I left out the ninth for a few reasons, mainly that most ninth innings that matter have been handled the same way before and after the switch, but there’s no real difference there, either.

Could we explain these differences? Of course we could. There are ways to build a narrative around any result. We’d say that the Rockies do better in the third and fourth innings now because their pitchers aren’t holding a bunch of energy in reserve, and because the worst starter has been excised from the line. We’d say they do better in the sixth and seventh because they are turning to fresh relievers, rather than pushing their starters through the opposing lineup for the third or fourth time. We’d say they struggle in the eighth because the set-up portion of the bullpen has been spread thin by needing to cover the sixth and seventh innings, and because there are no dominant starters pitching dominant starts finishing the game. These are the narratives we would use, and they’re tempting because they’re exactly the way this plan is supposed to work if this plan works. Of course, there are easy narratives that could have explained the exact opposite results. It’s way too early to say this plan works, or doesn’t, for any reason, and that chart is a good 100 games shy of what I’d want to put my faith in.

Other notable parts of the Rockies' rotation so far:

• They are actually going with a five-man rotation at the moment. Since Jhoulys Chacin returned from the DL three starts ago, the Rockies have kept Tyler Chatwood, Drew Pomeranz, Jeff Francis, Alex White, and Chacin on a traditional schedule, but with the same 75-pitch restrictions.

• The 75-pitch restriction is, and has been, a bit overstated. The Rockies starters have actually averaged 76 pitches per start since the switch; there’s a 92, an 89, a 97, and a 91 mixed in there. There was an 18-start stretch in July in which the Rockies' starters—well-rested because of the All-Star break and off-days—topped 80 pitches 13 times.

• Rockies starters have averaged 4.44 innings per start, down from 5.09 innings per start before the switch. They have actually been, as a group, pretty much unchanged. They have thrown strikes on 60.8 percent of their pitches (up from 60.5 percent), struck out 6.3 batters per nine (down from 7.1), walked 1.78 batters per nine (improved from 1.88) and allowed 1.5 homers per nine (from 1.6). But, of course, the starting rotation has changed plenty since then, and continues to change. Only three pitchers have made at least five starts in each format: Alex White, Christian Friedrich, and Drew Pomeranz. White has been quite a bit worse. Friedrich has been somewhat worse. Pomeranz has been a bit of everything: fewer walks in the new format, more homers, lower BABIP, more runs. Sample size strikes again.

Sam Miller is an author of Baseball Prospectus. Follow @SamMillerBB
Click here to see Sam's other articles. You can contact Sam by clicking here