Model picks
The most likely scenario for each match, with how much confidence the model places in its own reading. Sorted from the firmest reading to the weakest.
No picks published right now
We only publish a pick when there is data on both sides to support it. As the next round enters the analysis window, the matches show up here.
How to read a pick
The pick is the HIGHEST-probability scenario according to the model — not a scenario that is probable in absolute terms. A match can have "home win" as its main reading and still see the home side lose in most simulations, if probability is split across three outcomes.
Confidence measures how well the model supports its own reading, not how certain the result is. High confidence on a small sample is still a small sample — which is why we only publish with at least 10 matches per team in the current season.
The model does not see line-ups, late injuries, fixture congestion or squad rotation. The official line-up lands up to an hour before kick-off and can invert the whole reading.
Nothing here is a betting recommendation. It is a statistical reading of recorded data, published with the outcome afterwards — including when it is wrong.
Every pick on this page comes from the same model, with the numbers checked against the database before publication. Each outcome is recorded in the track record, hit or miss.
21+. Gambling problem? Call 1-800-GAMBLER.