Saturday, January 3, 2009

FIP vs. xFIP

One reader asked about this, and since I refer to xFIP from time to time, it's worth explaining. Less than ten years ago, a man named Voros McCracken made a startling discovery...pitchers had almost no control over what happened to balls in play. Greg Maddux was no better at "preventing hits" or "inducing weak contact" than anyone else. While his ideas have been debated and refined greatly since then, his original observation remains (in my opinion) the single most useful discovery in the history of baseball research...at least for fantasy players. You can do a better job predicting a pitcher's future ERA by looking at past K/9, BB/9, and HR/9 rates than by looking at past ERA. Ten years ago, nobody realized how important that was. McCracken's original formula for "DIPS" (Defense Independent Pitching Stats) was a little cumbersome to calculate, so Tom Tango came up with FIP...a simple formula that approximated DIPS and allowed easy calculation of what a pitcher's ERA should have been.

FIP = (HR*13+(BB+HBP-IBB)*3-K*2)/IP, plus a league-specific factor (usually around 3.2)

One of the refinements to the original DIPS theory was the discovery that (to a large degree) pitcher's didn't even control the home run rate they allowed, other than their ability to impact the rate of groundballs vs. flyballs. xFIP was a statistic that replaced the home run rate in FIP with a 'normalized' value that assumed that all pitchers allow the same number of home runs per flyball.

It's worth mentioning that while these formulas are excellent predictors of the future, they remain oversimplified models that don't incorporate everything we know about predicting pitcher success. The pitcher does have SOME control over what happens to various types of balls in play...in other words, their is a small degree of 'skill' in this area...but it's so small that for most practical purposes we can do fine if we ignore it.

1 comment:

Dave said...

Awesome, awesome, awesome. That is EXACTLY what I was looking for! So it's FIP with a refinement for HR rate, normalized relative to FB%, which is the component that is much more in the pitcher's control.

Thanks for posting that. I have read up on Voros and Bill James, and am very aware of Tom Tango, but sometimes his explanations of things leave me hanging a bit.

Much appreciated!