Handling the “Oracle Gap” in Hybrid Prediction Markets: Settlement Latency vs. UX Consistency

We’re building a hybrid prediction market (off-chain order book, on-chain settlement) on an EVM L2. We’ve solved most of the core matching engine issues (price-time priority, self-trade prevention), but we’re stuck on a specific problem related to outcome resolution.

The “Oracle Gap.”
Specifically, after a market event ends, there is a delay between when the off-chain oracle (e.g., a Chainlink Function or a trusted API) knows the outcome and when the on-chain settlement contract is resolved and users can claim payouts.

For a 5-minute resolution window, this is fine. But for fast-moving events (e.g., “Will ETH close above $3200 in the next hour?”), a 2-5 minute gap creates a horrible UX. Users see the “Yes” positions in their portfolio as winners, but can’t claim or trade out for several blocks.

We’ve looked at optimistic oracles (UMA), but the dispute window (2-6 hours) kills the product for short-term markets. We’ve considered a centralized pre-image commit, but that breaks our decentralization goals.

How are production platforms handling this latency between “real-world outcome” and “on-chain finality” without resorting to a fully trusted model? We need a mechanism that allows for sub-60-second settlement finality while maintaining a fallback for disputes. Any smart contract patterns or off-chain coordination tricks you’ve seen work?