I use Fielding Independent Pitching stats rather frequently in my pitching analysis, which can be useful or terrible depending on how transparent I am about how I’m using the stat. So, here it is, all of my assumptions and prejudices about FIP.
(1) I prefer to break FIP into its separate parts. See, FIP isn’t truly a stat; FIP weighs the value of HR, BB, and K against the expected runs allowed for those outcomes, and then expresses those elements in a ratio. (I’ve seen versions with HBP and IBB, as well, but I’ve not seen much agreement about how to calculate those other elements. Therefore, I use an extremely simple version of the stat: ((13HR+3BB-2K)/(IP)). I call this, “FIPratio,” since it is the bare element for calculating FIP).
(2) Once you calculate the FIPratio, you can express FIP on any number of scales. For instance, FanGraphs calibrates the stat to ERA; I prefer to use Runs Average, because then we can directly relate a pitcher’s expected performance to their actual runs allowed. (This is my prejudice: I prefer to use FIP to analyze defensive performances, which include unearned and earned runs alike).
Anyway, this is the step that places a pitcher’s fielding independent performance in front of a defense. It’s important to note that this shifts from year to year sometimes, as does general K/BB/HR ratios. For instance, although FIPratios remained similar between 2011 NL and 2012 NL, defensive efficiency declined significantly in 2012; a pitcher that allowed the same K/BB/HR ratio in 2012 might be expected to allow three more runs over 200 IP simply due to fielding decline.
FanGraphs calls the step of calibrating FIP to ERA or Runs Average, “FIPConstant.” Here, you take the LeagueERA or LeagueRunsAvg, and subtract “FIPRatio” from that number. VOILA! You have a constant number that you can apply to any pitcher’s K/BB/HR ratio for that season; by adding FIPConstant to a player’s FIPratio, you can see how many runs that pitcher might be expected to allow with average defense.
(3) SHORTCOMINGS: Of course, you should know that there are shortcomings with FIP.
(a) In general, FIP is not typically park-adjusted to a park’s ratio of K/HR/BB. This is especially problematic when comparing pitchers that work in, say, Great American Ballpark or Miller Park, with those that work in, say, PetCo Park or Coors Field. Not all parks require the same dependence on defensive efficiency, and not all parks encourage the same relationship between HR, K, and BB.
(b) In general, FIP assumes equal defensive distribution within a given league. Of course, this almost never happens because (1) teams do not have equal defensive efficiency, and (2) within any given team, fielders do not distribute their efficiency evenly between different pitchers. You might call this, alternately, the Randy Wolf or Zack Greinke issue.
(c) This is a personal prejudice, again. Sometimes it seems that K are valued much more than BB or HR. For instance, pitchers that don’t strike out a ton of batters might have below average FIPratios, but they could actually perform well by limiting the BB and HR damage. This happened to Wolf in 2011, and Bronson Arroyo (most notably) in 2012. Of course, over the course of a career, we’d expect a pitcher with great strike out rates to perform better than those with below average strike out rates (and perhaps that’s the point of FIP over several seasons), but sometimes those pitchers that walk the line with solid BB and HR rates can outperform their expected runs allowed.
However, if you understand the shortcomings of this stat, and you don’t try to do too much with it, FIP can be a great way to analyze: (1) a pitcher’s performance against his own team’s fielding, (2) a pitcher’s performance against his defensive efficiency for several seasons, (3) a team’s performance against their expected runs allowed, etc.
And now, you can calculate FIP for any AL or NL pitcher in the last 7 years, and you can judge them specifically against league K/BB/HR ratios, or include the tricky defensive element.
FIPratio: (13HR+3BB-2K) / (IP)
Runs Average: (9R) / (IP)
FIPConstant: (RunsAverage) – (FIPRatio)