That's definitely useful, but not quite what I was looking for. Basically, there has been all this talk about flawed pitching stats so I started to think about WHIP. Now I know this is by no means revolutionary and there is probably already a stat for it but here's what I was thinking of.
Basically, a pitcher has some control over what kind of ball is hit off of him (see Wang) but for the most part cannot control where the ball is hit, how hard, etc. and thus there is a tremendous amount of luck involved in whether any particular batted ball is a hit or an out. Thus, instead of using (Walks + Hit)/IP I was looking at something along these lines for a "projected WHIP".
((GBXGB%) + (FBXFB%) + (LDXLD%) + BB - IBB)/IP
GB, FB, and LD% would be the league average for those types of balls for being hits, whereas GB, FB, and LD would be the amounts of those balls a pitcher has given up. This method of thinkng is correct right? Or am I crazy.