On 6/21 he gave up 4 runs in the 1st (36.00 ERA), 9 runs in the first 2 innings (40.50 ERA), and a total of 13 in 5 innings (23.40 ERA) -- keeping him in did him a favor in the ERA department. On 7/19, his ERA after each inning was 18.00, 22.50, 18.00, 18.00, 18.00, and it jumped to 21.60 by pitching into the 6th. An exception is not warranted. You may as well take out his best 2 starts. When you suck, you suck. EDIT: You almost got me -- that 2nd game was actually on 7/18. He posted an ERA of 0.00 on 7/19 It may have done him a favor in his ERA for that day-but normally he probably would have been taken out after 6 runs or so-and 7 more in 4 innings does not help out his entire years ERA at all. Actually, yes it does. How did the extra 7 runs in 4 innings help out his year's ERA? Because it was inflated by giving up 6 runs in 1 inning. Say it starts at 4, goes up to 5.5 after getting shelled in the first inning or so. Instead of being taken out, he stays in 4 more innings giving up a run or two an inning, and his season's ERA ends up at around 5. No-the 6 runs would take it from 4 to 5.5, and then the 7 runs in the next 4 innings would take it up to around 6. It wouldn't rise as fast, but the total year's ERA would still continue rising. 7 runs in 4 innings is never going to drop a 5.50 ERA for the season, no matter what the person did the inning before. You're right, I was calculating wrong. Still, the point remains that those 2 starts shouldn't be removed just as his 2 best starts should be and that he actually performed much better by staying in that June game.