I am just wondering how significant a 0.5 differential is between FIP and xFIP when considering career norms, especially for someone who has logged lots of innings. I guess it well might be but I am just not sure and just not a statistician.The simple answer is that it is a fairly significant differential, particularly when you are talking about multiple seasons versus a single season.
I pulled the data from Fangraphs on all qualifying starting pitchers from 2002 (the earliest year Fangraphs keeps the stat, unfortunately) through 2009. Of the 669 seasons of 162+ innings, 105 produced an FIP that was greater or less than 0.5 in a single season. Out of these 105 seasons, only 20 pitchers put up an FIP greater or less than 0.5 more than once. Some of these pitchers put up an FIP > xFIP once followed by an xFIP > FIP once, which further dilutes the pool, and leaves us with the following pitchers:
FIP higher than xFIP 2+ times: Brett Myers (4), Derek Lowe (2), Greg Maddux (2), Eric Milton (2), Ramon Ortiz (2), Nate Robertson (2).
xFIP higher than FIP 2+ times: Matt Cain (3), Joe Blanton (2), Cliff Lee (2), Tim Lincecum (2), Mark Redman (2), Chris Young (2), Barry Zito (2).
These data further support (in my opinion) Derek Carty's assertion that FIP is a volatile stat. It is difficult to jump to a serious conclusion over two Major League season that "proves" that Cliff Lee or Tim Lincecum do a better job of keeping fly balls in the park (though it's tempting to jump to that conclusion based on how good they are). Likewise, it's hard to say that Nate Robertson or Eric Milton do a bad job of keeping the ball in the park because of two seasons of data.
The exceptions, though (which is what Gypsy Soul was asking about in the first place) are what fascinate us endlessly.
Is Matt Cain extremely lucky or is there something about the way he pitches that suppresses HR and keeps his HR/FB% low? (as an aside, I'm curious to know why he wasn't one of the two pitchers mentioned in Derek's study?)
You can argue that AT&T Park suppresses HR, but this doesn't completely explain it; with the exception of Cain's rookie season, he's actually given up more HR/9 on the road than at home. He's a fly ball pitcher, so it's not that he's keeping the ball on the ground (and in the park). His career strikeout rate of 7.4 per nine is very good...but certainly not in the uber-dominant territory that makes me think that everything Cain throws is impossible to hit.
At some point, I have to look at Matt Cain and conclude that the exception will hold. He's doing it again this season (a 0.9 differential in favor of Cain's FIP versus his xFIP), which makes four years out of five where his FIP is better than his xFIP. Cain is keeping fly balls in the park.
So, yes, Gypsy, a 0.5 career differential is significant, and would make me believe that more than mere luck is involved.