‹ Jerry Shi

Try the player analytics

Serve and return numbers for top US players, built the same way. Open full screen ↗

How Hawk-Eye tracks one ball

Every US Open court has run Hawk-Eye Live since 2021, with no line judges. Each court has 12 ball-tracking cameras and 6 foot-fault cameras. The cameras film at 340 frames a second. In every frame, at least two cameras see the ball, and the system triangulates where their sight lines cross. It then fits a flight curve through those points to find the bounce.

12 + 6tracking and foot-fault cameras per court
340frames per second, every camera
3.6 mmmean error. The ITF allows 5 mm
0line judges on US Open courts since 2021

From the cameras to the player box

At the USTA I was in the stadium working on pro players' data, for players such as Ben Shelton and Coco Gauff and for their coaching teams in the player box, coaches like Brad Gilbert. The feed gives speed, spin, bounce and position for every shot. My job was turning it into the few things a coach can use between points. Coaching from the box has been allowed at the US Open since 2022, with short phrases at the same end and signals otherwise. So the insight has to fit in one sentence.

Ben Shelton, Coco Gauff and Brad Gilbert are likenesses generated with Meshy AI from Wikimedia Commons photos (CC BY-SA); I am generated from my own photos. The points they play are real Hawk-Eye tracking from another player's matches (Rudy Quan, 2025 Indian Wells qualifying, shared with permission), not their own matches. The opponents and the crowd are generic Meshy figures.

What is real here