One Prompt, One Match: Inside the Small Boom in Tennis Prediction-Market Side Projects

Why a 130-year-old scoring system became the best sandbox for AI-assisted builders.

Over the past several years, as technological tools have evolved at an alarmingly rapid rate, the online culture around them has shifted significantly. One of the biggest shifts is the rise of prediction markets and online casinos. While these tools and platforms have existed for decades, this new, hyper-aware online culture has pushed them into new ubiquity. Online casinos and prediction markets are now incredibly popular, and one of the newest applications of these platforms is a pretty surprising sport: tennis.

Prediction market platforms let users bet money on what they think the outcome of a given situation will be. In recent years, as these platforms have seen a massive influx of users, they've covered everything from the resolution of a given geopolitical tension to the average review scores of a given movie or TV show.

However, the most popular kind of prediction is often the outcome of a sporting event. This taps into something that has long been popular, even before it necessarily leaped into modern online spaces. People have bet on big sporting events for centuries, so it was only natural for it to carry over this way. Still, when it comes to betting on something like the NFL Super Bowl, the predictions largely consist of who will win. In tennis, prediction markets can go much deeper, essentially chronicling the match as it happens live and matching it point for point.

As such, building on tennis prediction markets is an active trend among AI-assisted developers. This is also partly because a computer program can understand tennis with surprising ease, and its point intervals and ebbs and flows are much more like computing than many might realize.

Developers who build with AI coding assistants have a new favorite side project: tools that watch a live tennis match and a prediction market at the same time. For these developers, these tennis-related projects look less like traditional quantitative finance and more like a weekend software project. But tennis has become an unusually accessible sandbox for builders working with live data.

In tennis, points are scored one at a time. The structure is strictly hierarchical, with point, game, set, match, and win probability having textbook closed-form solutions. As such, it is the rare sport that a developer can model on a laptop and genuinely understand. One prompt to an AI assistant now produces a working live match watcher. The appeal isn't simply that tennis is popular, but that its mechanics are unusually legible to someone building a model on a laptop.

This is all the more potent when paired with modern AI-powered tools, as what once needed a data team to chronicle effectively is now little more than a Saturday project. AI coding assistants such as Claude Code, Cursor, and Replit change the practical workload. These systems serve as live match watchers, break-point alerters, paper-trading bots, and even backtest leaderboards. As a result, most builders in this space are solo developers rather than professional quants.

On August 3, 2026, Polymarket became the Association of Tennis Professionals (ATP) Tourʼs official prediction market provider, covering roughly 20,000 matches a season with live streaming for registered US users. As a result, both parties gained attention from audiences that wouldn’t necessarily have been inclined to get involved. Tennis audiences were exposed to prediction markets in a new way, along with the possibilities of live sports betting.

At the same time, Polymarket users were exposed to tennis in a new way, even incentivized to engage with the ATP Tour to better bet on points and match outcomes. As such, tennis gained a mainstream in-play market almost overnight. The official ATP partnership means real liquidity and a real in-play venue, making it interesting for many solo builders to build against it.

Notably, this freshly formed relationship reached new heights at the end of August, when the US Open began. The US Open is among the most-watched tennis events each year, and its arrival so soon after the initial boom in tennis-related prediction markets helped fortify this new interest. Thus, the US Open effectively put the new market in front of its largest audience of the year.

A new, emerging public-built culture is forming around tennis market tools. The focus remains on experimentation, modeling, data plumbing, and software. These markets are young and thin. However, many find undeniable appeal in turning a live tennis match into a tractable software problem.