Skip to content

Cool tennis-tracking app

Swupnil Sahai writes that he’s developed Swing, “the best app for tracking all of your tennis stats, and maybe we’ll expand to other sports in the future.”

According to Swupnil, the app runs on Apple Watch making predictions in real time. I hope in the future they’ll incorporate some hierarchical modeling to deal with sparse-data situations.

In any case, it’s great to see our former students having this kind of success.


  1. Jonathan says:

    It’s more than a neat app: the data can be used to teach. There’s a physicist at Yale I heard talk about golf. I believe he played at Div1 college level. He put sensors into a club so he could measure the forces. He found some extremely powerful things that can be taught. Two are: a) every good golfer stops at the top the swing and b) the energy transfer point is in a relatively small hitting area and the force translated, which he expressed in g’s, is expressed to a very large extent by the wrists snapping through. He saw the data and adjusted his swing and dramatically increased his power. It also increases the ability to impart backspin, etc. The problem with extending this knowledge is that you’d have to buy the sensor-filled golf clubs. There’s a certain usefulness in teaching this with a special club, though I think there’s much more to be gathered from using your own club with the watch. You should then be able to see how your swing power varies across the clubs, and how it can be improved in specific situations – like when you need to hit down more, are you actually snapping through as much as needed? By teach, I mean you should be able to build a community of results, by which I mean your own iterations and those of others, from which you can get specific advice. It should be able to tell you what to do by learning what works best.

    This kind of learning is specifically appropriate for sports where things are held. Running is tough: I can figure my stride length and frequency but not how I hit the ground, how my hips tilt, etc. This leads in a held sport to ‘you need to turn your wrists over more’ connecting to more general advice that you need to hold your hips properly to swing your arm properly, etc., knowing that those parts are less directly related to the physical action that can be measured. I find that mathematically fascinating: the usual is that a first order perturbation or effect is worth so much more, but it’s very difficult to identify the effects of higher orders filtering into what can be measured. That’s like UZR: the guy does or doesn’t get to balls, and then when he does, he catches them or not, hits them back or not, hits them back or not with speed, etc. If the sensors get good enough and the algorithm clever, it can identify your reactions better and from that say, for example, when you take this much effort to reach a ball, you hit it back this much ‘better’ or ‘worse’ than when you go this far, this hard, etc. There’s a lot in this ‘room’ which we’ve not yet begun to explore.

  2. My guess is that these things matter far more for those at the highest levels of the game. All the technology isn’t going to help you if you don’t get your racket back, bend your knees, and see the ball on the strings. Which most people don’t. Vic Braden, a tennis pro, used to say that when people asked him what racket they should buy (i.e., which technology was particularly suited to their needs), his answer was, “It doesn’t matter. They all play better than you.”

Leave a Reply