Update on Statcast, PitchFX Change: Are Pitch Movement Comparisons Valid Across Years?
Dave Cameron of FanGraphs jolted the baseball numbers community when he published a post about Statcast’s takeover of PitchFX. This means any baseball site that used PitchFX will now be posting Statcast data instead. As a consequence, the past 10 years of PitchFX data cannot be confidently compared to 2017 numbers.
About 10 hours after I voiced my concerns about the Statcast takeover, Dan Brooks, one of the chief operators of Brooks Baseball, wrote a post defending the baseball data website. Soon thereafter, I asked him and Alan Watts, another baseball statistician, to elaborate on the inconsistency between the two systems. I frequently compare season-to-season differences in horizontal and vertical pitch movements, and I wanted to know if I can still do this with confidence. Their responses:
If you want the short version to my question of whether we can compare 2017 pitch movements to previous years, Alan fortunately explained it in layman’s term:
While my question was sufficiently answered by Dan and Alan, I’m left feeling unsatisfied. In biological sciences, if one outcome measurement is being acquired across multiple sites, researchers control for that site difference, even if those differences are small. However, we can’t control for Statcast and PitchFX differences in movement since we don’t know the extent to which the two technologies differ. While the two methods should shoot out similar pitch-movement measurements, we can’t confidently compare this season to previous seasons.