Version 1.0

Methodology

We sell a frozen research edition, not a promise. The proof system is designed to make quiet rewrites visible.

Exact markets

Every forecast states an objectively resolvable market. Match products use team to advance, including extra time and penalties. The outright product resolves to the officially declared tournament champion.

Evidence grades

A means captured publicly before the event. B means an independent archive proves pre-event visibility. C is retrospective embedded-media evidence. D is context-only uncertainty. X is a conflict, repair, recap, or integrity failure and is never counted as a hit.

Commit–reveal

Before sales, the versioned packet is canonicalized, combined with a fresh 32-byte salt, hashed with SHA-256, and signed. The public ledger exposes the commitment without revealing the selection. Buyers receive the packet and salt. At kickoff, both become public.

Scoring

Every edition receives a terminal state: revealed and resolved, withdrawn before the event with a reason, or integrity failure. Markets are scored separately. We do not silently overwrite misses or blend different market definitions into one accuracy number.

Limitations

Source consensus can be correlated, samples can be small, and past performance does not guarantee future results. Confidence is editorial research judgment—not certainty, investment advice, or a demonstrated wagering edge.