I'll start off by saying I'm new to MLBAM's Pitch F/X data. I haven't compared him to other pitchers to see if his pitch selection patterns are normal, but I'll show you what I found. In Tuesday night's game, Joe Saunders pitched 5 innings of batting practice, giving up 8 hits, 5 runs, and 3 homers.
I think his problem was that when he got behind in the count, he threw only fastballs. He did at least mix these up between his four seamer and two seamer, but the hitter was not going to be fooled by speed. Saunders threw 14 pitches on a 3-0, 3-1, or 2-1 count. All but two were fastballs, and the two non-fastballs came on 2-1 counts, where he at least had some leeway.
Contrast this with his pitch selection with 2 strikes and less than 3 balls. He had 16 pitches in these counts, and mixed them between 4 fastballs, 3 sliders, 8 curveballs, and a change. He may have been throwing the fastball behind in the count because he was afraid to walk people, but his strike percentage was about the same either way.
If I can find these patterns, then hitters know them too. When major league hitters know what's coming they are going to hit it a long way. One compliment often paid to the elite pitchers of the game is that they have confidence to throw any pitch on any count. You never know what's coming. Later I'll look at last year's data to see if Joe is usually this predictable or if he did a better job randomizing his pitches last year. I sure hope he does on Sunday when he takes the mound against the A's.