Prospective records · wins and losses retained
Public results ledger
Every shared prediction is timestamped before the off and retained after settlement. The ledger starts at zero rather than borrowing fictional history.
This is a platform-controlled public record, not independent third-party proofing.
Forecast revisions remain immutable in storage and the paid revision trail, but scoring uses only the terminal settled revision for each race and model contract. One race therefore contributes at most one observation to a contract's evidence.
- Terminal race forecasts scored
- 0
- Top-pick wins
- 0
- Top-pick win rate
- —
- Current-version Brier
- —
- Brier skill vs market
- —
- Current VALUE signals
- 0
- Evidence tier
- limited
- Locked price vs SP
- —
- Record policy
- Prospective only
One per race and contract · table shows latest 0
Lower is better · race-analysis-v11
0 matched locked races · positive is better
Locked before the off · race-analysis-v11
Descriptive product threshold · not independent proof
0 starting-price comparisons
Finish-band evidence
Prospective Harville top-two and top-three estimates are scored only when every forecast runner has an official result. This platform-controlled evidence is not independent validation; non-finishers are negative outcomes and dead heats can put more runners inside a nominal band.
Current-contract VALUE ledger
Each row uses one paper unit at the timestamped price observed with the prediction. An observed price is not proof a customer could transact. Returns exclude availability, commission, deductions, limits, tax and customer execution. Dead heats divide the stake under standard paper-settlement arithmetic.
No settled predictions yet
Real results will appear after the first provider-backed predictions have run and settled.