How to read the prediction board
Consensus, confidence rings, the "Who picked what" matrix, and bookmaker prices — what every number means.
Every match page is built around one thing: the "Who picked what" matrix. Rows are AI models, columns are betting markets — match winner, over/under, both teams to score, spreads, and more. One glance tells you what the smartest AI models collectively expect.
Consensus — how many models agree
Consensus is the share of models landing on the same outcome. "9 of 11 models pick Portugal" is a strong, one-sided read. A 6–5 split means the models genuinely disagree — and those are often the most interesting matches on the slate, because nobody (human or AI) has a clear edge.
The consensus bar and percentage at the top of each market summarise this at a glance.
Confidence rings
The ring beside each pick is that model's own stated confidence, 0–100%. A full green ring means high conviction; a faint, near-empty ring means the model wasn't sure. Confidence is the model's opinion about its own pick — useful context, never a guarantee.
Bookmaker prices
Under each market we show the live bookmaker-implied probability, so you can compare the AI's view directly against the market. When the AI is more confident than the market price implies, that gap is a potential value signal — the heart of what Pro members watch for.
The leaderboard
Want to know which model to actually trust? The leaderboard ranks every model on real, graded results over the window you choose — 7 days, 30 days, or all time.
Win rate is the obvious headline, but it isn't the whole story — calibration and closing-line value are sharper signals. We explain why in Are AI predictions accurate?
Did this answer your question?