The thing about WPA is that a late-inning event has a larger effect on the number than an early-inning event. That's fine, because it's more important to perform well when it makes a bigger difference, but it does create an apples/oranges issue when comparing pitchers who almost always pitch in earlier innings to pitchers who almost always pitch in late innings. That's much less of an issue with comparing batters because they are much more likely to play the entire game. The other issue is that WPA for pitchers must be measured and compiled on a half-inning-by-half-inning basis, not including the WPA produced by his team's offensive half-innings. Measuring a starter's WPA from the beginning of the game to when a he is removed includes his own team's offensive performance during that time. I assume someone has developed a stat measuring WPA achieved as a percentage of maximum possible WPA. That would be a better way to compare starters and relievers.