FIP or xFIP

I used xFIP instead of FIP in my A.L. rookie write-up, and JEM1776 wasn't sure if I should have.

Just a saber query: shouldn't you be using FIP instead of xFIP? xFIP is the windshield of the car as we gaze at the future, i.e. the 2012 season, while FIP is our rear view window looking at what actually happened in 2011. We're deciding about what actually happened in 2011, not what should happen in 2012.

Yes, I probably should be using FIP and not xFIP in cases like these. However, it's important to note that while FIP might be somewhat more descriptive of what happened in the past, it is also generally regarded as a predictive metric as well.

Fangraphs and others have already done a great job with the definitions of both metrics, so I'm not going to rehash that here.

To jem's point, though, FIP and xFIP are extremely similar for most pitchers.

ERA, FIP, and xFIP Differentials: MLB Pitchers with 100+ IP, 2011

DIFF	ERA/FIP	ERA/xFIP	FIP/xFIP
0 to 0.3	56	48	90
0.31 to 0.6	47	43	42
0.61 to 0.9	25	27	10
0.91 to 1.2	10	14	2
1.21 or higher	7	13	1

That 0.3 differential isn't arbitrary. To create a “perfect” corollary between ERA and FIP, you would also create a hypothetical number of earned runs that a pitcher allowed. The number of pitchers in the 0.3 differential is nearly identical to the number of pitchers whose real ER versus their hypothetical ER are five or less. When analyzing a pitcher, 0-5 earned runs isn't a significant number in the overall analysis.

So while I should have used FIP, the difference between the two metrics isn't as significant as the difference between either ERA and FIP or ERA and xFIP. For pitchers who give up a significantly high number of ground balls or fly balls AND either give up a lot of home runs or very few home runs the difference in xFIP and FIP can certainly loom large. But as the chart above shows, pitchers with a large FIP/xFIP differential are the exception and not the rule.