April 12, 2011
Early PITCHf/x Ballpark Speeds
Several people have asked me about the accuracy of the speeds reported by PITCHf/x systems around the major leagues in the early going. I usually wait about a month into the season before I run checks on the calibrations of the PITCHf/x camera systems installed in all the major league ballparks. We need a sample size of 10-20 games before we can be really confident that we are evaluating the cameras and not just the specific set of pitchers that have played in a given stadium or the weather they have been playing in. In my full set of PITCHf/x accuracy checks, I adjust for those things in order to improve the estimates.
However, in order to get a quick idea of how to evaluate the fastball speed readings we have seen so far, it is not necessary to chase accuracy down to the tenth part of a mile per hour. I took the average fastball speed of every pitcher in every game and compared to his average fastball speed over the 2008-2010 seasons. Then I averaged these differences across all the fastballs thrown in each park this year. This result will include the effects of temperature and pitchers in early-season condition. In other words, we should expect fastball speeds that are slower than average because temperatures in April are colder and pitchers are not in mid-season shape.
Without further ado, here are the results by park:
The average (mean and median) recorded fastball speed is 0.8 mph slower thus far in 2011.
Let me remind you of a few things before I go. First, changes in fastball speeds can be due to things other than a change in pitcher performance. As mentioned, temperature has an effect. Pitch classification can also have an effect. (Joakim Soria is Exhibit A.) Second, last year's speed measurements at a pitcher's home ballpark may have been slightly off (again, see Soria). Third, I do not consider these values an accurate assessment of PITCHf/x camera performance given the temperature bias, point-in-season bias, and relatively small sample size. Please do not consider this data an evaluation of Sportvision or Major League Baseball Advanced Media data quality. Rather, it is intended to help people gauge early season fastball speeds in context, to the extent that is possible two weeks into the season.