Methodology, data sources and how often it changes

Where the data comes from

Fixtures, line-ups, statistics and odds arrive from a data provider. Odds are taken from several bookmakers, so that a single outlier doesn't set the value of a whole tip.

From that data the model calculates probabilities for every market on offer. Each published calculation is stored with its timestamp, so numbers are never overwritten — a new version is added and the old one stays traceable.

The closer to kick-off, the more often

A match a week away is recalculated on a scale of hours; a match an hour away, on a scale of minutes. There's no point calculating more often than new data arrives. Every match shows when it was last calculated, so you can tell whether you're looking at a fresh number.

What the model can and cannot do

The model weighs historical quantities for both teams — goals, corners, cards, home and away form — and turns them into probabilities for individual markets.

It doesn't read the news, doesn't know the mood in the dressing room, and has no idea the manager fell out with his star player this morning. It works with what's in the data, and everything else is beyond its reach.

Its weak spot is the small sample. A promoted side with no history, a team after a transfer window, or a competition early in the season — estimates are shakiest there. When data is missing for a match, the app says so rather than producing a number with nothing behind it.