I'm a huge fan of sabermetrics and by extension, I'm a huge fan of one of the fields greatest minds, Tom Tango. Tango is always positing new ideas at his blog, and today he toyed with the idea of something he called 'FutureFIP' (can we call it fFIP?) that is meant to be a better predictor of future performance than regular FIP/xFIP. As Tango notes, FIP was never meant to be a predictor of future performance, even though it does a decent enough job - certainly better than something like ERA.
Anyways, it's worth checking out if you're a stat-head.