How we analyse football matches and generate data-driven predictions
Our predictions are model-driven statistical forecasts built from recent competitive matches. They are intended as analytical insights, not advice.
Not all fixtures have the same type of recent historical data available. To reflect this, each match is assigned a data source tier, which describes how the recent match data was gathered for the underlying analytics.
Fixtures where both teams have sufficient recent data from the same competition and match context.
Fixtures where competition-specific data is supplemented with additional relevant matches.
Fixtures where limited recent data exists in comparable competitions.
For every fixture, the system evaluates how much recent data exists for each team in the same competition and match context as the upcoming game.
The tier is determined by the least available context-specific data between the two teams.
Each published prediction is placed in a confidence bandbased on the model's probability for that market.
Highest model probability band for that market. Exact cutoffs differ by market (full-time vs first-half overs).
Mid model probability band for that market.
Lower model probability band still above the publish threshold.
Predictions below the publish threshold are not shown. Dashboard accuracy for confidence bands uses only these three groups.
Predicts the match to produce 3 or more goals.
Used when both teams show strong attacking profiles and goal-heavy tendencies.
Predicts the match to reach 2 or more goals.
A more conservative full-time prediction.
Included automatically for matches that already qualify for Over 2.5.
This ensures filtering captures all relevant games. Accuracy is always measured against the primary prediction, not the auto variant.
Predicts 2 or more goals in the first half.
Based on rapid-start teams and strong early scoring patterns.
Predicts at least 1 first-half goal.
A more conservative first-half prediction.
Included automatically for matches that already qualify for 1H Over 1.5 Goals.
This ensures filtering captures all relevant games. Accuracy is always measured against the primary prediction, not the auto variant.
Our analysis focuses on recent form, matchup dynamics, and scoring patterns, using a large set of competition-relevant statistics.
Bookmaker odds are collected for research. They are not shown per fixture and are not used to rank or filter matches.
On settled fixtures, Pro compares each confidence band's hit rate with market-implied probability after removing overround — an aggregate dashboard view, not a fixture-level score.
We don't promise results.
We promise clarity, consistency, and honest data-driven analysis so you can make informed decisions.