How pitchers tire is one of the great unknowns left in baseball. Ideally, you could measure fatigue by ERA inning by inning, but the problem is not only does fatigue factor in, but so do adjustments by hitters who have now seen the pitcher multiple times.

So when you hear an announcer say something like "The third time through the order Joe Pitcher has a .320 batting average against" it is hard to say if the cause is pitcher fatigue or hitter familiarity. A better method would be to track the pitcher's stuff and see how it changes the deeper in the game he goes. This completely removes the batter from the equation and we can isolate pitcher fatigue. With piles of PITCHf/x data rolling in, we have the perfect information to begin to isolate pitcher fatigue. ...

hat-tip Tango, BBTF thread