How It Works
From data to evidence.
Every published probability follows a structured process. Every outcome is recorded. Every mistake helps us improve.
The process, step by step
Data
We gather historical information on teams, players, and competitions. No noise, no narratives.
Analysis
Our system processes the information using mathematical models calibrated on thousands of matches.
Probability
The model generates a quantitative probability for each market. No intuition, no luck.
Publication
We publish the probability before the event. It's recorded in the history, impossible to modify.
Outcome
When the match happens, we record the actual result. Good or bad, it stays there.
Evaluation
We compare prediction vs. reality. We measure performance with transparent metrics like Brier Score.
Improvement
We accumulate evidence. Each new model version is built on real history, not assumptions.
What does a probability mean?
A probability is not a certainty. It's a quantitative estimate based on the data available.
When we publish that a team has a 72% probability of winning, it means in similar scenarios the model estimates that outcome would occur roughly 72 out of 100 times. It does not guarantee any certainty in a specific match.
A model with 70% estimated accuracy will be wrong 30% of the time. That is the statistical reality.
Transparency as a principle
All published predictions are recorded. We don't delete the ones that fail. We don't rewrite history. Our track record is built with every event, correct or not.
Explore the predictions
Everything we publish, with its full record.
See probabilities →