A pitcher, for the most part (please take care to note this italicized bit), doesn’t have control over the things that happen to a pitch he has thrown after it has been contacted by the hitter. Defense is one part of the story – teams with poor defenses will allow a lot more balls in play to become hits. Luck is an even bigger part. Fly balls usually do not fall in for hits, unless the hitter gets lucky and drops one in between the first baseman and right fielder, or some other no man’s land. A pitcher is not at fault for any of these things, which is why we pay close attention to their BABIP (Batting Average on Balls In Play), which measures the percentage of balls contacted by the hitter that fall in for a hit. The overall league average is around .300. If a pitcher’s BABIP substantially higher than that (unlucky) or substantially lower (lucky), it can be expected to regress to back to the average. In either case, it can dramatically effect the pitcher’s conventional statistics.

There have been plenty of admirable attempts to create statistics that eliminate the distorting factors of luck and defense from a pitcher’s performance. Most of them center on the “three true outcomes,” the three plate appearance results that pitchers have complete control over, and that correlate the most with pitcher effectiveness: strikeouts, walks, and home runs. The most widely accepted of these is FIP (Fielding Independent Pitching), which is essentially a pitcher’s expected ERA based on these most important of components. The exact formula is (HR*13+(BB+HBP-IBB)*3-K*2)/IP, with some adjustment of the end result to fit the ERA scale. I use FIP a lot on this blog and in general baseball discussion, because it is a pure, defense-and-luck-neutral measure of a pitcher’s performance, and is a far superior predictor of future performance to ERA.

xFIP goes a step further in pursuing the pure measure of pitching performance. Atmospherics and ballparks can have a big effect on how many home runs a pitcher surrenders, through no fault of his own. To account for this, xFIP uses the same formula as FIP, but adjusts the HR component to the amount of home runs a pitcher would have surrendered if his HR/FB% (proportion of fly balls given up by the pitcher that were home runs) were the league average rate. This eliminates the effect of ballparks, etc. I frequently use xFIP too.