Skip to main content

True KAP Computation

Feature Request

Many people like Anna Bright and RealClearStats take a much stricter view of what Kitchen Arrival Percentage is, at least for Serving KAP which is the most important. Roughly KA Success is both servers arriving at the line STABLE, or winning the point outright on 3/5/7 shot on the way in. PBV’s current Kitchen Arrival metric is far off - it often shows rallies where a serve team member doesn’t even get in. At high levels, this is one of the most important stats: in fact, Ben at RCS has data showing a clear correlation between DUPRs and KAP. What would it take to get a much more useful KAP metric? I’m happy to expound further and to even provide examples.

Posted by Randy Parker ·
Status: In Progress6 comments

Log in to comment and vote

Comments6

  • PB Vision Team

    Team•

    Sep 1

    @Randy Parker this is great feedback. The KAP algorithm is very feasible for us to update. The main work is agreeing on a precise definition we can consistently translate into code.

    Are you focused specifically on Serving KAP? Today, PB Vision counts a serving-team opportunity once four non-fault shots have been completed (giving that team a fifth-shot opportunity). It then checks whether each player reached the kitchen (arrived within a few feet). Winning the rally does not itself count as an arrival, which explains why our metric can differ significantly from the stricter definition you described. That was by popular request since people didn’t want to see kitchen arrival stats muddled into when a team wins on a serve or third for example (so short rallies are excluded from denominator).

    Could you email support@pb.vision and mention Ryan asked for more detail on KAP. It would be great if you could spell out how you would score situations such as:

    • Both servers arriving and becoming stable

    • Winning on the 3rd, 5th, or 7th while transitioning

    • One player arriving while their partner does not

    • Both players arriving briefly but being pushed back

    • Rallies ending before a fifth-shot opportunity

    We will need a measurable heuristic for "stable," perhaps both players staying within three feet of the kitchen line for at least one second. A few video links and rally timestamps showing what should and should not count would be incredibly helpful.

    You can also DM me in our Discord: https://discord.com/invite/N73jSVDpwK. Roadmap comments are easy for me to miss, but I would like to work through this with you and get the definition right.

    ~ Ryan

    • Randy Parker

      •

      Sep 2

      @PB Vision Team Hi Ryan, thanks so much for following up. Serving KAP really is such an important stat. I’m going to take some time to do a thorough writeup answering those questions and more. I’ll also pull some example rally examples.

      But, just to get this out there: I realized in reading your message that I, in fact, compute an even more nuanced version which is essentially: TEAM KAP BY WHO HIT THE THIRD.

      Why Team KAP: because I think this is fundamentally a team metric, especially at higher levels. If I arrive and my partner doesn’t, we haven’t accomplished our objective which is to neutralize the serve-team disadvantage.

      Why Hitter of Third: this is key to figuring out why a team isn’t getting in. I have opponents in Senior Pro that I target for returns because I’ve noticed their KAP is lower when that person hits the 3rd.

      I’ll pull this together in the coming week and include all of the above. I want to see if I can get Anna’s take on it too as she is the one who originally showed me the wisdom of KAP.

      -Randy

      • PB Vision Team

        Team•

        Sep 2

        Thanks, @Randy Parker . Team KAP by third-shot hitter is a really interesting way to make the stat more actionable. I really appreciate you taking the time to write this up and pull examples, and I’d love to hear Anna’s take as well. Looking forward to digging into it when received

        ~ Ryan

      • PB Vision Team

        Team•

        Sep 24

        Hey @Randy Parker just checking in with you on this!

        • Randy Parker

          •

          Sep 25

          Thanks @PB Vision Team - I got tied up but last week I did produce a pretty good writeup to address the edge cases and revise some of the answers I gave earlier. I am awaiting feedback from several ppl including AB. I’ll update you next week. I’m on the Discord now too. P.S. Just did a ton of video analysis and still find this stat to be a real storyteller for high-level matches.

          • PB Vision Team

            Team•

            Sep 25

            Awesome, I’m very excited about what the stat can tell us and to feed the algorithm to our new AI Coach so everyone can start to use it. Feel free to DM me on Discord (I’m ryan@pb.vision)