How it works
The full pipeline, end to end
Five stages between your screenshot and a structured prediction.
1. Upload your screenshot
JPG, JPEG, PNG or WEBP up to 8 MB. Coupons, fixture lists, virtual football screens — anything where team names are legible.
2. AI vision extraction
A vision model reads each row and returns team names, league, kickoff time, current state, score and odds where visible. Nothing is invented — unreadable rows are flagged instead.
3. Match identification
Extracted names are normalised and matched against our match database. Low-confidence rows are handed back to you to confirm before anything is analysed.
4. Statistical prediction engine
Team form, home and away splits, scoring and conceding rates and points-per-game feed a Poisson goal model. Market probabilities are computed, not asked of a chatbot.
5. Prediction output & history
You get a primary call, probability spread across markets, expected goals, likely score, confidence, risk level and written reasoning — saved to your history and tracked against results.
What the engine actually uses
Where an external football data provider is configured, real team statistics are used. When none is configured, the platform falls back to internal statistical estimates — and every prediction produced that way is labelled so you always know which data backed it.
Predictions are AI and statistical estimates, not guarantees. Past performance does not guarantee future results — you are responsible for your own decisions.