The headline is an obvious statement, but I had yet to see an actual number put on the gap between starting and relieving until now. Tom Tango said the following:
The replacement level pitcher as a starter has a .380 win%. Move that starter to relief, and his win% goes up by about .09, or .470 win%. That’s it.
The average starter has a win% of .490 and the average reliever has a win% of .520 (more or less, and by win% I mean based on his pythag component ERA). As you can see, the average reliever is not that much better than the replacement-level pitcher as reliever. That’s why we say relievers are a dime a dozen. So, the average starter is +.11 wins per 9 IP and he uses up two-third of the innings. The average reliever is +.05 wins per 9 IP and he uses up one-third of the innings. If you follow along, the average starter gives you twice the value, per inning, as the reliever, and he gives you twice the innings. That sets the value of the average reliever of 25% of the average starter (1/2 times 1/2). This number goes up a little when you add in the leverage impact of relievers.
When people bring up Joba Chamberlain and suggest he belongs in the bullpen, I frequently explain that starters are significantly more valuable than relievers, such that it makes sense to give him every chance to succeed out of the rotation. Even if Joba is a top reliever and simply an average starter, his value is almost certainly going to be greater taking the ball every five days. Unless he tanks entirely in that role, the “bull in a china shop mentality” and all of that psycho-babble garbage that gets spewed to support moving him to the pen should be viewed as largely irrelevant. The job of the team is to extract as much value as possible from Joba, and having him in the rotation is the best way to do so.