The football model proof record is the product's trust layer.

Public picks are locked before kickoff.

GoalsProof uses the public proof page to make the model accountable. Selections are recorded before games start, then settled after full time. The key point is not one day's result; it is whether the record holds up across a meaningful sample.

The main metrics are simple.

• Landed rate: how often the selection outcome happened.
• Average locked odds: the average price available when the pick was saved.
• Breakeven rate: the hit rate needed to justify the average odds.
• Level-stake return: the profit or loss if every public pick was followed equally.

Proof is useful because it includes the limits.

Football remains uncertain. Context, late team news, weather and market movement can all matter. The public record does not pretend otherwise; it gives users a way to see whether the model is improving and whether different markets should remain live, stay in the lab or be paused.