Fox10 Line Calling Wrong

[glacierguy]

For the second time at the Astana tournament I've seen the Fox10 line calling appear completely wrong. The second time is in the Tsitsipas v Hurkacz match, first set tie-break, 4-3 to Hurkacz, Hurkacz serves wide on the ad side, linesperson shows in, Tsitsipas appeals. The computer generated ball impact shows in by about 1cm, the actual video shows ball nowhere near line, ie out. Plus when they stop both displays to show "contact point" the ball in the video is still in the air, whereas the computer generated ball has already impacted and deformed. Tsitsipas disputes it, but of course nothing happens.

In my opinion that system does not work reliably.

[jimlosaltos]

For the second time at the Astana tournament I've seen the Fox10 line calling appear completely wrong. The second time is in the Tsitsipas v Hurkacz match, first set tie-break, 4-3 to Hurkacz, Hurkacz serves wide on the ad side, linesperson shows in, Tsitsipas appeals. The computer generated ball impact shows in by about 1cm, the actual video shows ball nowhere near line, ie out. Plus when they stop both displays to show "contact point" the ball in the video is still in the air, whereas the computer generated ball has already impacted and deformed. Tsitsipas disputes it, but of course nothing happens.

In my opinion that system does not work reliably.

There are several such instances of bad calls by Fox10 captured as ani-GIFs online. In every one I have seen the ball was called in when it landed slightly out.

Is Fox10 not as good as HawkEye? We have no idea. Fox10 supplies live video, Hawkeye does not and doesn't even have cameras at ground level. There is no third-party verification of either system that I am aware of. HawkEye stated that its own review showed hardly any errors and those were entirely by human operators. But what did they check against if they have no on-the-deck cameras?

Remember when I did the article accumulating high MPHs and RPM measurements from the InfoSys tabulations? John noted how a USTA official told him that while the line calls were accurate, the highest velocity measurements were (or was it might be?) off. I can't fathom how a system can create a computer simulation of an impact if it is measuring a ball going at, say, 100 MPH as going at 130 MPH. I don't need a computer to see that is a 30% error.

The main reason I have faith in HawkEye is that Major League Baseball uses a far more complex version of that system in every ball park. They replaced one with a combination of doppler radar and cameras. So, they checked the cameras vs radar and publish not only ball velocities to decimal points, but speed of motion and reaction for every player on the field.

One time, they gave TV a live track of a squirrel running across the infield with its instantaneous speed <g> Hit 23 mph, I believe. Not bad.

Unlike tennis, MLB is transparent and providing massive amounts of data to fans. They're measuring exit velocities of home runs, and reports of how much and what kind of spin pitches have. It's as if the ITF/ATP/WTA are using abacuses and baseball is using computers.

[johnyandell]

It's a known fact that error margin with Hawkeye is 3mm.

[jimlosaltos]

It's a known fact that error margin with Hawkeye is 3mm.

Sony has long advertised accuracy of 3.6 millimetres. It might even be true.

But I remain skeptical until and if we see some independent testing that's done transparently.

More recently the company is basically claiming 100% correct calls in a study, absent human error. HE's release claimed zero computer errors, every single erroneous call from HawkEye was caused by a human, such as picking the wrong shot to track. How were those shots checked?

I assume ELC is more accurate than human line callers, or at least more consistent. But I hope you'll forgive my skepticism in numbers presented solely by a vendor.

And when they move to claiming 100% accuracy, the hair stands up on my neck. It's like Elon Musk claiming back in 2016 that Tesla's "self-driving software" was already safer than human drivers. His stats proved it!

As far as I can tell tennis has never done an independent test of the system, or if it has but it has not released the results nor explained the methodology.

Baseball is more transparent and its information is reassuring about the accuracy of HawkEye. But MLB uses a more complex version.

MLB switched in 2020 to a more sophisticated version of Hawk-Eye than that used by tennis, installed in every ballpark, but even that with that "the league was unable to avoid variances in data collection based on each ballpark. When Statcast is unable to accurately record exit velocity data for a batted ball, either because of ballpark factors or some other reason, it imputes a value in its place"

https://www.baseballprospectus.com/news/article/50763/statcast-exit-velocity-a-statistical-assessment-of-park-effects/

Please forgive my skepticism but I've been around tech companies for too long to take claims at face value.