Cadence
Cadence describes how long after a game finishes an assertion is posted, and how that timing adapts to the game's volatility signal.
Post-game assertion timing
When the stats provider marks a game final, the oracle classifies active player/stat cells using a volatility assessment based on the play-by-play data. The result determines how long each signal bucket remains in internal validation before the short formal UMA window runs.
CLEAN cells are settled on a fast track: the assertion can be built after the T+15m stability buffer, internal validation, and Arweave archival. The formal UMA window is 15 minutes by default.
Non-CLEAN cells go through the additional 60-minute stabilization delay, then a signal-specific preclose schedule before cross-reference validation and assertion. LOW, MEDIUM, HIGH, and CRITICAL buckets use progressively longer internal validation windows, described in Settlement → Timing, followed by the same short formal UMA layer.
End-to-end timing:
| Path | Game end → finalized assertion |
|---|---|
| CLEAN | ~30 minutes |
| LOW | ~3 hours |
| MEDIUM | ~5 hours |
| HIGH | ~7 hours |
| CRITICAL | ~13 hours |
A consumer that resolves markets within the same session as the game ends can plan for roughly half an hour on clean cells and a few hours on low-risk cells. Games with detectable volatility signals — a disputed scoring decision, an error affecting scoring, a 2-out error extending an inning — take longer by design, because the internal validation period gives the league time to publish any corrections before the assertion is posted.
What happens if a game does not finish cleanly
Games that are postponed or suspended produce no market-resolution assertion until they later reach final. Explicit GAME_VOID assertions for cancelled games are part of the contract design, but are not live in the current beta posting path. If a postponed game is later rescheduled under a new scheduled start time, the oracle produces an assertion for the rescheduled game under a new canonical event ID, because the commenceTime input changes. Consumers that locked a market against the original event ID will need their own governance process to handle the situation; the oracle cannot retroactively map the old ID to the new game.
Games whose stats never stabilize within the hard cap on stability verification are routed to manual review. Consumer markets referencing those games remain unresolved until the operator intervenes — either confirming the correct values against a third source and asserting manually, or declining to assert if the data cannot be reliably determined.