Live Tennis APITennis hackathon projects, with what each team built
18 public tennis projects from hackathons between 2014 and 2026, each linked to its own page. About half work from video or motion sensors. Others build on match results, rankings or point data.
Each name links to the team's own page. "Data it used" is what that page says, in short.
| Project and hackathon | Year | What it built | Data it used | Code |
|---|---|---|---|---|
| TennisIQHacklytics 2026 | 2026 | Takes match video, finds the court, ball and players, cuts the match into point clips and exports a PDF report for a coach. | Uploaded match video | GitHub repo |
| What Wins Tennis Matches?AQX Sports Analytics Data Bowl 4.0 | 2026 | A dashboard that splits how much seven serve and return stats explain men's match results, by surface. | A public dataset of men's tour results, 2015 to 2024 | GitHub repo |
| Event-Driven Gain in Expectancy (EDGE)PhysTech 2026 | 2026 | A player metric that weights aces, winners, double faults and unforced errors by how much each changes the chance of winning the game. | Point-by-point match data | GitHub repo |
| RallyPointMuslimHacks | 2026 | Court tablets scored by voice and one organiser screen, synced over the local network, for running club tournaments. | Players' spoken score calls | GitHub repo |
| Real Time Tennis AnalysisHackGT 12 | 2025 | Tracks the ball and the player's pose in match video, with trend dashboards and coaching tips. | Match video and historical men's results files | GitHub repo |
| Para standing Tennis EvaluatorTechnica 2025 | 2025 | Para standing players record a swing with a camera and Arduino sensors and get suggestions for the body parts they chose. | The player's own video and motion sensors | No public code |
| TennisShotClassifierAI with MATLAB, UNIWA | 2025 | A neural network that tells forehand, backhand and running apart from phone motion data, with the phone standing in for the racket. | Their own phone sensor recordings | GitHub repo |
| TRACYQHacks 2024 | 2024 | Tracks the players and ball in match footage and turns it into coaching summaries, heatmaps and statistics. | Match footage | GitHub repo |
| Tennis BoleHacklytics 2021 | 2021 | A crawler and a desktop tool that merge public player ratings and rankings into one list, to help a college team find young players. | Ratings and rankings from public tennis websites | GitHub repo |
| Queens of TennisPearl Hacks 2021 | 2021 | A web app that charts women's player data and compares two players' chances of winning. | Public women's match data | GitHub repo |
| CourtFind!uOttaHack 4 | 2021 | Pick an area of a map and a computer-vision backend marks the tennis courts in the satellite images. | Satellite map images | GitHub repo |
| Single Player TennisBackyardHacks | 2020 | An Arduino circuit, simulated in Tinkercad, that turns a backyard practice wall into a scored game against a virtual opponent. | None | No public code |
| From_AO_to_AIFrom AO to AI | 2018 | A LightGBM model that predicts whether a point ended in a winner, a forced error or an unforced error. | Tracking data from the organisers | The page is the code |
| Predicting How Points End in TennisFrom AO to AI | 2018 | Notebooks that compare classifiers, neural networks and ensembles for the same task, and settle on XGBoost. | Tracking data from the organisers | The page is the code |
| Curious GeorgeMakeMIT 2018 | 2018 | A robot that finds tennis balls with computer vision on an Nvidia Jetson and picks them up with a foam claw. | The robot's own camera | GitHub repo |
| Project TennisMHacks 8 | 2016 | A multiplayer VR tennis game in Unity with an Alexa skill as scorekeeper, driven by phone and armband motion data. | Motion data from a phone and an armband | No public code |
| BDPebbleTennisHackRU Fall 2015 | 2015 | A Pebble smartwatch app that classifies each stroke and counts forehands, backhands, topspin and slice in an iPhone app. | Smartwatch accelerometer data | GitHub repo |
| SpinTennisSamsung Gear App Challenge | 2014 | A Samsung Gear 2 watch app that counts forehands, backhands and overheads and shows the power of each shot. | The watch's own motion sensors | No public code |
The 2018 Australian Open competition
In 2018 Tennis Australia's Game Insight Group ran From AO to AI, an online competition. Entrants had to predict how each point ended: a winner, a forced error or an unforced error. The Australian Open says 750 people from 55 countries took part and sent 2,731 solutions, working from 10,000 points of tracking data. The winning model was right 95 times out of 100. Two of the public solutions are in the table. Read the Australian Open's report.
Can we build one on the free plan?
Yes, for a weekend project. A free API key needs no card. It gives live scores, players, fixtures and tournaments, at 30 requests a minute and 100 a day per key.
100 a day goes fast if the whole team shares one key. Give each member their own, and cache what you fetch. A live scoreboard that asks every 15 minutes stays under the daily limit.
To get going, try the coding exercises. They start with a first request and end with a live scoreboard, and the tests run offline. If your project uses an AI assistant, the MCP setup page has the config for your client.
Is your project missing?
Send the link to hello@livetennisapi.com. We add tennis projects whose page is public, whatever data they use.
Every project page was checked and live on 8 November 2026.