Monday, July 15, 2019

ALPB TrackMan Follies - A Neck-High Strike

Long before the term "robot umpire" entered baseball's mainstream lexicon, I wrote that the electronic strike zone had a significant vertical blind spot. One day after the first Atlantic League TrackMan ejection, MLB/ALPB's computer called a neck-high strike on a batter squaring around to bunt in what nearly turned into the second TrackMan-related ejection in as many days.

Stop & Think: The computer-generated strike zone graphic placed this pitch well within the strike zone. This is yet another of dozens of examples of what we've seen based on less than a week's worth of the Atlantic League's automated ball/strike system (e.g., the All-Star Game had a 37.6% discrepancy rate, and the aforementioned neck-high called third strike isn't even part of the discrepancy group since the graphics & TrackMan call agreed that the neck-high pitch was a "strike").
Related PostReviewing Atlantic League's Automated Strike Zone (7/11/19).

Now extrapolate that to MLB's FoxTrax and K-Zone graphics, care of PitchCast...this failure or error rate is precisely why the technology is nowhere near ready for the big leagues and, as a consequence, precisely why casual observers routinely think umpires aren't great at plate calling while the league privately tells the umpires that they are doing a fine job (fans are influenced by broadcast graphics that are routinely incorrect).
Related PostCall for Umpire Accountability & the 97% Plate Score (4/19/19).

The 3rd strike.
The Play: With none out and one on (R2) in the top of the 6th inning of Rockers-Patriots, Rockers batter Giovanny Alfonzo showed bunt on a 0-2 pitch from Patriots pitcher Julio Perez before pulling the bat back as the ball sailed past Alfonzo, approximately neck-high.* After the customary TrackMan delay, HP Umpire Kyle Fecteau called Alfonzo out on strikes.

The Non-Ejection: As beleaguered batter Alfonzo flipped his bat and threw his arms in the air, a brief discussion ensued between the umpires and High Point coaching staff, including Frank Viola, who was ejected earlier in the series for arguing a TrackMan ball/strike call.
Related PostHistory - Baseball's First Ejection Due to TrackMan (7/12/19).

Pitcher: Throws neck-high. TrackMan: "Strike"
The Overrule Provision: An umpire is empowered to override the TrackMan decision on an overwhelmingly and clearly incorrect call: mainly the rather black-and-white issue of a pitch that bounces in front of home plate and crosses at a point above the hollow of the knee, where TrackMan's "strike" call is clearly wrong. Much less clear is a situation where the properties of the batter's stance obfuscates the lower and upper limits of the strike zone, as in a bunt attempt, juxtaposed alongside an electronic system the league wants to pass off as accurate and legitimate.

The look on umpire Detweiler's face says it all.
The Counterargument: Whereas High Point argued that umpire Fecteau could have overruled TrackMan (as Viola did with umpire Tim Detweiler on Friday night), had Fecteau disregarded TrackMan in favor of his own call, opposing Somerset would have entered the dispute with the counterargument of why the umpire overruled the technology, given ALPB and MLB's investment in its implementation—the "no win" proposition.

The Problem: As previously stated, vertical strike zone error with MLB's TrackMan (and StatCast/PitchCast) has been an ongoing problem. FiveThirtyEight caught on when MLB switched from Pitch f/x to StatCast in 2017, as have many other analytic outlets, and baseball academics have been aware of the issue ever since. As I wrote in another column, the mainstream has invested far too much in the technology to make much of it, at least not yet.
Related PostMLB Ejection 077 - Jeremie Rehak (4; Brad Ausmus) (6/9/19).
Related PostUEFL f/x vs K-Zone and the Player-Umpire Disconnect (10/4/18).
Related PostRob Manfred Talks Robot Umps - Tech is "Way Up" (5/30/18).
Related PostGil's Call - Best Practices and Avoiding Faulty Journalism (7/5/17).

Big League Example: When Ron Kulpa ejected Houston Astros coach Alex Cintron and Manager AJ Hinch in April, PitchCast, in real-time, indicated Kulpa's call was incorrect. The ejections generated much fanfare when Kulpa responded to Hinch's instruction not to look at the dugout with, "I can do anything I want" (for the future, PR-friendly response is, "I can look where I need to in order to do my job; if someone in the dugout is distracting me from my duties, I must look in their direction to warn and/or eject the offender"; make sure you say that over the span of about two seconds).
Related PostMLB Ejections 007-08 - Ron Kulpa (1-2; Cintron, Hinch) (4/3/19).

Kulpa was right, but Hinch didn't know it.
In any case, a day after the game ended, good old MLBAM processed the game's numbers and credited Kulpa with a correct call after all. This is a routine postgame process that ordinarily benefits umpires (because the humans adjust to vertical zones in real-time while computers do not), that the public rarely if ever hears about. We've known for awhile that the strike zone computers are generally unable to execute real-time vertical calculations, and those chickens are coming home to roost with the Atlantic League TrackMan experiment, which will continue its bumpy ride off into the dog days of summer.
Related PostBad Computer Umpire - Faulty Pitch Data Defames Kulpa (4/6/19).

Once the ALPB experiment proves problematic past a certain point, MLB may admit the error of the Atlantic League's ways and divest itself from the problem through a PR-laden campaign that instead of scapegoating an umpire, may very well end up scapegoating a computer. Who else could possibly be left to blame?

*SIDEBAR: Why was Alfonzo bunting with two strikes? Because one of the other Atlantic League rules changes allows a foul bunt with two strikes to be deemed a foul ball, instead of a third strike.
Related PostAtlantic League Debuts New Rules, E-Zone (7/10/19).

Video as follows:

Alternate Link: Alfonzo squares to bunt and is called out on strikes after pulling the bat back

0 comments :

Post a Comment