There wouldn't be a point to taking out his best starts, because what you really want is a guy who consistently goes out there and puts up quality starts. I don't know that Marquis fits that description, but I find it reasonable to remove a few real stinkers and leaving his best starts in there as an evaluation tool. To look at it from an extreme, a pitcher who shuts out his opponent 30 times in a season, but has 5 starts where he gives up 30 runs each is actually a pretty valuable pitcher, even if his ERA is going to be way up there because of those 5 starts. Why exclude his 5 best starts if you are just attempting to see what his body of work looks like if you exclude a small handful of really bad games? Pitchers are going to have really bad games that are outside their norm, but they're also going to have really good games as well. If you want to exclude the 2 worst performances as some sort of an aberration, then the 2 best should be considered the same. It's not very often that a pitcher allows 12 ER in a game and it's even more rare that he has a performance like that twice in a season. This is beyond an aberration. That being said, I wasn't trying to justify Marquis' contract or say that without those two starts he was a good pitcher. I was simply showing how much those two starts affected his ERA for the season. His ERA jumps by 0.69 just by him giving up 5 runs in those games. EDIT: His 2006 season was flat out bad no matter which way you look at it.