Decision Science
Odds Research Lab
Settled 2025 archive
28 observations
Calibration
Actual settlement rate versus average first-crossing price. Perfect calibration follows the diagonal.
Time To Expiration
Observed success and average paid probability grouped by days remaining.
Expected Value Desk
Sizing uses binary-contract Kelly: (p − q) / (1 − q). It is zero when estimated edge is negative. Spread, fees, slippage, and model error are not included.
Evidence Quality
D
Correlated evidence
What this test supports
First archived trade at or above 60%, converted to YES probability, then held through binary settlement.
What it does not support
Persistence, executable ask, volume confirmation, path drawdown, delayed fills, or walk-forward model selection require denser snapshots and order-book history.
Observation Audit
One row per unique Polymarket contract. Click a column heading to sort.
| Asset | Contract | Type | Entry ↕ | Crossed ↕ | Days left | Settlement | P/L ↕ | Trades |
|---|