It began with two brothers and a weekly football challenge.
GoalsProof started with a simple idea between two brothers: pick a short run of football fixtures, follow the story through the weekend, and see whether we could build a little momentum.
We jokingly called it Roller, or more accurately, Rollaaaaaa. It was never just about the fixtures. It was the weekly debate, the screenshots, the optimism, the post-match messages, and the familiar feeling that this time we had found the perfect set of games.
The problem was simple: we were spending hours researching, but still making too many decisions on instinct.
We had pieces of logic, but no real process.
Each week followed the same pattern. We would look through fixtures, check recent results, scan head-to-head records, glance at team form, and convince ourselves we had found the right angle.
Sometimes the thinking was sensible. Sometimes it was selective. A team had scored in five of its last six, but against who? A league looked goal-heavy, but was that still true this season? A previous meeting finished 3-2, but did that actually tell us anything useful about the next match?
The more we looked at it, the more obvious it became: football decisions need structure. Without it, you end up remembering the evidence that supports your view and ignoring the evidence that weakens it.
So we started building the board we wished we had.
The first question was not, "Can football be predicted perfectly?" It cannot. The better question was, "Can the research process be made clearer, faster and more honest?"
That became the starting point for GoalsProof: a daily football intelligence board focused on Over 1.5 goal candidates. Instead of jumping between fixture lists, form tables, market screens, weather checks and old notes, the product brings the useful signals into one workflow.
The model looks at scoring patterns, recent goal behaviour, home and away splits, league baselines, market support, match context, weather risk and historic evidence. Then it turns that into a shortlist that is easier to review.
The mission is better decision-making, not perfect prediction.
GoalsProof is not built on the idea that football is certain. It is built on the opposite belief: football is noisy, emotional and easy to overread.
The goal is to make the uncertainty clearer. Which fixtures have the strongest goal profile? Which ones are supported by the market? Which ones look tempting but carry context risk? Which parts of the model are actually proving themselves over time?
That is why GoalsProof is designed around a simple loop: scan the board, understand the reason, track the decision, and review the proof.
Proof became the centre of the product.
The more we built, the more obvious it became that a model is only useful if it is willing to keep score. Anyone can talk confidently after a good day. The real value is in showing the record over time, including the misses.
That is why GoalsProof separates public model proof from private user tracking. Public shortlists are timestamped before kickoff and results stay visible. Users can also track their own choices separately, so they can compare their personal record with the model record.
No deleted misses. No hindsight edits. Just a clearer way to learn whether the process is improving.
At its heart, GoalsProof is still inspired by the same thing that started it: two brothers trying to make sense of the weekend football slate together. Only now, the challenge has a bit more data, a bit more discipline, and a lot more proof.