The app combines statistical and ML models (Elo-style ratings, expected goals, scoreline probability distributions) with historical match data to generate pre-match context and analytical predictions. A core goal was to keep the output explainable and readable for fans, rather than overwhelming them with raw numbers.
Some features: - Pre-match win/draw/loss probabilities and xG estimates - Scoreline probability distributions - Live match events, lineups, and stats during games - User predictions vs model output - Match threads and discussion tied directly to fixtures
One motivation behind this was that many large sports apps try to cover everything and everyone, which can make them noisy and hard to digest. Scoping both the model and the UI to a single team allows for a tighter fan experience and, in theory, better-calibrated models for that specific context.
I’d really appreciate feedback on: - What information feels unclear or confusing - What feels missing or underdeveloped - Whether the single-team focus makes sense from a product standpoint
Happy to answer technical or product questions - thank you!