When we submitted the concept for "Diseño de una aplicación web para la predicción de resultados en NBA" at the Universidad Politécnica de Madrid in 2018, we had no idea it would still be running eight years later.

It started as a final year project for a Software Engineering degree at ETSISI — the Escuela Técnica Superior de Ingeniería de Sistemas Informáticos. The brief was to design a system that could estimate NBA match results using statistical data. The 2018 work focused on methodology and concept — what factors matter, how to weight them, and what a prediction tool for basketball could look like in practice.

The original idea

The core question was simple: can you use publicly available basketball statistics to predict who will win a game, and by how much?

The answer, it turns out, is yes — imperfectly, but meaningfully. The 2018 thesis identified the parameters that matter most: team win percentage, points scored and conceded, pace of play — and critically, which players were actually available to play. That last factor, player availability, would become the obsession that shaped everything that followed.

The research was published through UPM's open access repository and remains there today as the earliest written record of what eventually became Buzzer Beater.

From concept to app

Two years later, during a Master's degree in mobile devices and applications, the concept became a real product. NBA Community — as it was called then — launched on iOS with a backend and database built from scratch.

The Master's thesis expanded the vision considerably. Live scores, injury reports, community chat, personalised notifications, team favourites. The academic abstract described it well: *"an application that provides users with the latest information of the American basketball league, as well as statistics about players, teams and matches."*

But the community angle was always there too. The idea was never just a data dashboard — it was a place for fans to talk about the game, share opinions, and make predictions together.

What changed, and what didn't

The app went through several names and several infrastructure changes over the years. Each time the NBA updated their data formats or endpoints, we adapted. The notification system evolved from simple alerts into something much more considered — grouping updates intelligently so users get what they need without being overwhelmed.

The web platform at communitynba.com launched in 2025 to reach fans who aren't on iOS. Live scores, standings, player and team pages, game chat — all of it built to match the mobile experience.

But the prediction engine from that 2018 thesis? That's still the destination. It was the original idea and it remains what we're building toward. The beta is targeting the end of the 2025–26 season.

Why we kept going

Honestly, because the community kept asking for things. Every time the project could have been shelved, someone would get in touch asking for a feature or pointing out something they cared enough to track down.

Side projects live or die on whether anyone actually uses them. Buzzer Beater has users, and they're the reason it's still here.

If you want to follow what we're building next, the about page has the full timeline — and the predictions feature will be the place to watch when it launches.