Moneyball in the Starting Rotation

I agree that you want your best pitchers pitching as much as possible, which will result in fewer runs allowed. I disagree however, with your conclusion that innings are not being transferred from the better pitchers to the worse pitchers (and in fact you suggest it might be more). Consider the following:

In 2011, AL starting pitchers averaged 6.06 innings per start (I’m focussing only on the AL, because in the NL pitchers are sometimes removed for PH in critical situation, not because they necessary should be removed on pitching merit). That means that the average team had 982 innings pitched by starters, or about 196 per spot in the 5 man rotation. Now, you suggested each starter under your scheme would pitch 3-4 innings, so let’s just call it 7.5 innings between the two starters. That would mean 1215 innings pitched in total over the year, or around 202 innings per spot in this 6 man rotation. Now, of course, the better pitchers pitch more innings than average, and the back end guys less, but the same would be true under your new system, so sticking to averages works. So in other words, the average workload would be roughly the same, but now we’ve introduced a 6th starter to soak up some of these innings, and this 6th guy would presumably be the worst starter. So essentially, we’re redistributing innings from 1 inning relievers to a below average starter, and I can’t see that increasing the overall pitching level (though in theory, through chaining, you’re displacing the worst relievers, not the better ones, so it might be close to a wash. That’s the best case I can see though). The key is, you’re introducing a 6th starter into rotations, who by definition is going to be a very marginal starter since 5th starters are (in aggregate) quite marginal.

I can’t see how the front line guys would get more innings. Right now, front line guys average around 7 innings over 30+ starts, or around 210-225 innings. Under this system, to get into this range across 54 starts, you have to average 4 innings. This is basically the limit, since under your system, you can’t push guys much further than that since they have much less recovery time. I certainly couldn’t see any starter getting to the 266 innings Halladay put up in 2010, I’d imagine 225 would a ceiling.

Another factor is exactly how much pitchers would improve. Relief pitchers outperform starting pitchers because they go all out for one inning. Starting pitchers have to pace themselves currently to throw 100+ pitches, and you suggest they would be able to put in some effort in shorter stints. I didn’t address this above, but I’m not sure this is the case. Throwing 4 innings still requires a certain stamina level, plus the pitchers will have a lot more recovery time. There would be some benefit, because hitters improve each time through the order (part of it is pitcher fatigue, part of it is just having seen what a pitcher has): the OPS against in 2011 the 1st/2nd/3rd time through the order was 700/740/776, so by cutting off the third time through essentially, you’d see some better results.

This leads me back to the copying aspect. Of course that fact that others will copy you isn’t a reason not to do something, but if everyone does this, you ultimately get no competitive advantage while introducing weaker 6th starters, which ultimately dilutes the pitching. So from a game theory perspective, I’m not sure your proposed idea is an equilibrium, because I don’t see it netting a benefit at all. But ultimately, I think the biggest problem is just the fact that all starters are brought up in the current system – it would be almost impossible for a team to switch over, just as it’s basically impossible to go back to a 4 man rotation. It’s like the QWERTY keyboard – terribly inefficient and suboptimal, but it’s the standard.

Moving on, you maintain starters will have new rhythms. Sure, no doubt. But what I was worrying about was from a physical standpoint, will their arms have enough recovery time? To be honest, I don’t know the answer, and I know there there is a debate in baseball about optimizing contributions from starting pitcher (Nolan Ryan, for example thinks pitchers could pitch more innings without larger injury risk). 100 years ago, pitchers started 50 games a year, but it was a very different game – the stress modern day pitches put on arms is a lot higher. But the burden of course, is on proving your system would be better than the current system, and I’ve heard or no evidence to suggest this. More importantly, the second starter will not necessarily know when he’s coming into the game, whereas now they know exactly, since warming up takes a certain amount of time. Now, to a certain extent, you ca mitigate this by saying to guy, you’re coming in for the 5th, whatever happens. But what if the first pitcher is doing really well, and you want to leave him in for the 5th – and your second guy has always started warming up? What if your first starter gets shelled and you take him out after 3? And let’s your offense is scoring, so the games still close – you ca bring in a reliever, but you’d ideally want your best guy going into the tight game. So there’s a lot of practical issues to work out in terms of second starter prep that I think you’re glossing over.

Finally, to address your facts about why relievers are better – yeah, you’ve basically nailed the reasons. I’ve discussed about how I don’t think you’d get much extra ability to go all out or that they’d be much fresher, you’re still going through the order so there’s no platoon or match up advantage. The only thing is, less times though the order, and I see that advantage as being counteracted by having poorer starters. Why do starters get pulled earlier in the playoffs? Simple, because it’s a short series and you basically play to win every game, more or less at whatever cost, forget what effect it has on tomorrow and worry about tomorrow tomorrow. In the regular season, you just don’t do that, you have to think about keeping guys fresh and in routine.