How Coach Bola Makes Picks

How a pick gets made

Coach Bola sweeps every fixture on the calendar through a three-tier model. Tier 1 is a statistical baseline. Tier 2 layers on 11,879+ data points — form, lineups, market context, weather, travel — and produces a probability for each outcome. Tier 3 (Coach Bola itself) compares that model probability against live market prices, and only recommends a pick when the model thinks the market has it wrong by a meaningful margin.

No pick is recommended on every fixture. On most matches, the model agrees with the market and stays silent. Picks only surface when there's a measurable edge.

What “confidence” means

The confidence number on every pick is the model's implied probability that the recommendation lands. A 70% confidence pick should, over a large sample, win about 70% of the time. We measure this every day via the public ledger — the calibration card at the top shows the actual hit-rate inside each confidence band. If the model's 80–90% bucket isn't hitting 80%+, we don't hide it; the badge auto-degrades to “rebuilding”.

What we don't claim

• Past results don't guarantee future returns. The ledger is a public receipt, not a promise.
• No pick is a sure thing. Confidence is a statement about long-run frequency, not a certificate.
• We don't hide losing periods. Every settled pick — win, loss, push, void — sits in the ledger the same way.