Probaballer is an open research project exploring the intersection of football analytics, machine learning, and sports betting markets.

The goal is to develop and evaluate predictive models for football matches — starting with match outcomes and expanding into granular predictions like goals, cards, and expected metrics. This site documents the methodology, shares insights from the data, and tracks model performance transparently.

Current focus areas include Spatiotemporal Graph Neural Networks (STGNNs) for modelling player interactions, expected value analysis against bookmaker odds, and building robust data pipelines for English lower leagues where market inefficiencies may be more exploitable.

This is very much a work in progress. Models will be wrong, hypotheses will be revised, and lessons will be learned. The aim is to approach football betting as a rigorous quantitative exercise rather than speculation.

For updates, discussion, and the occasional hot take, follow @Probaballer on X.