Practice-native volleyball analytics
Every touch in every practice, tagged.
Skyball turns practice and match film into a searchable touch log — who touched the ball, when, and what skill. Built for college programs, starting with the film nobody has time to break down: practice.
We run the demo on your own practice film, not a canned reel.
Film in. Structured data back.
Upload a session and Skyball returns a touch log you can actually work: filter it, click a row to jump the video to that moment, build a reel from it, and export it.
A touch log, not a highlight reel
Contacts become rows — time, player, skill — that you can filter by player or skill with the video following along. Touches it can't confidently attribute are marked unknown rather than guessed.
Per-player breakdowns
Touch counts and per-skill splits by player within a session, so you can see who actually got the reps in that practice.
Film room
Heart a touch and it lands in a playlist. Playlists play straight through as a reel — no clipping, no exporting, no waiting.
Match tools break on practice film. This doesn't.
Practice is where the reps are and where analysis never goes — because practice film breaks the assumptions match tools are built on. There's no scoreboard, no clean rotations, and players usually aren't wearing numbers.
Identity without jersey numbers
In practice, numbers are missing or hidden under pinnies. Skyball groups each player's appearances and shows you the crops so you can confirm who's who — once — instead of guessing silently or giving up.
No score context required
Practice has no scoreboard to lean on, so the touch log stands on its own: what the body did, and what the touch was in the rally.
Built for volume
Practice is most of your film and almost none of your analysis. Skyball is built to take every session, not just the matches worth clipping by hand.
Match film too
The same engine handles match video and adds what the scoreboard makes possible: team and jersey identification, where players were standing at each touch, and possession-level value.
How it works
- Upload the sessionDrag in practice or match video from the camera you already use.
- It processes on GPUPlayers and the ball get tracked, every contact gets tagged. You don't have to sit and watch it.
- Confirm who's whoFor practice, match the identity clusters to your roster once. That's the only manual step.
- Work the filmReview touch by touch, pull player breakdowns, build reels, leave comments for your staff, export the data.
Straight answers about your film.
College film is sensitive and student-athletes are involved, so here is where we actually stand — including the parts that are still being built.
- We don't train on your film by default. Processing your video is the service. Using it to improve our models is a separate, explicit, compensated agreement — never the fine print.
- No facial-recognition matching. We don't run a face model that identifies your athletes. In matches, identity comes from jersey numbers; in practice, we group each player's appearances and show you the crops — including ones where faces are visible — so a human on your staff makes the call.
- You can take your data out. The touch log exports as CSV or as its underlying structured file, any time, without asking us.
- We show you what it couldn't identify. Every run reports how many touches got attributed and flags the ones it couldn't place. That's coverage, not a correctness score — we won't tell you an identification is right when we don't know that.
- Retention and deletion, in writing, before you send film. Source video currently sits in storage with a fixed retention window, so we won't pretend it vanishes on a phone call. We'll put the actual retention period and deletion process in your agreement, and we'd rather answer that question precisely than wave at it.
Bring us a practice. We'll break it down.
We're onboarding a small founding group of college programs. The demo runs on your film, so you can judge it on the sessions you actually coach.