Is Hawk-Eye Inflating Extensions?

Along with all the madness going on in the 2020 MLB season, another change has been the use of a different ball tracking system. The Hawk-Eye system is debuting this season as the replacement for the Trackman system. Comparing the data of 2020 with the past, there appears to be inconsistency between the two systems.

Through August 13th, league wide extensions – how far the pitcher releases the ball from the rubber – have seemed inflated. Both Trackman and the ball tracking system before it, PITCHf/x, agreed the league average extension was just around six feet. Hawk-Eye disagrees tacking on about an extra third of a foot. That may not seem like a lot, but in a game that comes down to fractions of seconds, taking that extra millisecond of decision making away from the hitter can make all the difference.

What if this is due to changing conditions? If new players are suddenly coming into the league with Glasnow-like extensions, that could help explain the table above. To test this, I looked at returning pitchers from the 2019 season. If the distribution of extension changes is around zero, then the aggregate changes observed are probably due to new players in the league or players with poor extensions pitching less often.

Of the 480 pitchers to pitch in both 2019 and 2020, only 18 have seen their extensions shrink. The average player gained .315 feet of extension in the offseason. Clearly the average shift is nowhere near zero. Unless everybody shook hands during quarantine (not recommended) and decided to stage an extension revolution, something else is going on.

Looking at 2019’s month by month median extensions serves a few purposes. First, it proves extensions do not fluctuate all too much within a year. So even if players were feeling particularly flexible, such a significant jump in league wide extensions would be unlikely. Second, it shows extensions trend upward as the year progresses. This means extensions are probably sitting lower than they usually would, not higher. So if anything, the Trackman-Hawk-Eye extension difference is being slightly understated at this point in the year.

Looking at returning player differences by ballpark, it can be seen the problem is widespread. It is not isolated to a few ballparks jacking up results around the league. Instead it is a systematic problem. Some locations, notably Tropicana Field and Angel Stadium, diverge from the pack, but none manage to approach zero.

The disagreement between Trackman and Hawk-Eye on extensions seems relatively consistent meaning the problem should not be too painful to fix. Whichever system is right, adjust the data from the other system accordingly. With PITCHf/x and Trackman having agreed in the past, I default to them being correct, but that’s not the real problem.

The real problem comes in using Hawk-Eye’s 2020 data with and the previous data concurrently. Had I not looked around the league, today’s article would’ve been about Trevor Bauer’s extension increase (.487 without adjustment, 85th percentile). With the month-to-month and park-to-park variation, the accuracy of that measurement is too unclear to analyze.

Trackman and PITCHf/x both went through their own growing pains when they were first introduced. PITCHf/x had calibration issues and Trackman struggled with the vertical position of balls. Ball tracking systems need time to work the kinks out. Until all these inconsistencies have been found and rectified, be wary of 2020 data being compared to previous years; it might not be apples to apples.