Why Don’t We Have FIP for Hitters? A Need and Attempt to De-luck Hitter Performance

To date, hitting sabermetrics have lagged behind pitching metrics in the removal of luck. While most sabermetricians recognize the pervasive impacts that batting average on balls in play (BABIP) can have on a hitter’s batting line, attempts to precisely adjust players’ “triple slash,” OPS, or weighted on-base average (wOBA) for luck remain relatively uncommon. Thus, under existing metrics, luck disproportionately distorts assessment of hitters, particularly when performance differs from expectations. Given the number of hitters each year who “break out” or “decline,” the ability to recognize when these incidents stem from luck versus fundamental changes in play is an essential part of player evaluation. In this paper, I propose a method to estimate de-lucked hitter performance by estimating rates of individual hit types using underlying peripherals (e.g. fly ball percentage and fly ball distance). These rates can be used to compute expected versions of many additional statistics, as well as expected counts for each hit type. With a slight modification for home runs, expected versions of batting average, on-base percentage, slugging, isolated power, OPS, and wOBA all outperform their realized values in predicting subsequent-year performance. This aligns with existing findings for pitchers, and hints at the potential to successfully evaluate hitters on a more luck-neutral basis. Tags: hitter luck; expected hitting; xHitting; xWOBA; xBABIP ∗This paper was originally submitted to the 2014 MIT Sloan Sports Analytics Conference, and has been expanded since. I am grateful to Dan Brooks and Chuck Korb for a poster opportunity at Saber Seminar 2014, as well as Matt Goldman, Lester Lusher, Kevin Ray, Ling Shao, Alex Smith, Matt Swartz, Igor Vaynman, and various readers on FanGraphs.com for comments and feedback. Any errors are my own.