Early resolution has been part of HivePredict for a while. If a market outcome is clear before the official close, users can submit proof and admins can resolve manual markets early.
That part works well.
The issue was timing.
The awkward situation could run into
A market outcome could already be known in the real world while the chain and app had not caught up yet.
That lag can happen for normal reasons:
- the person with proof is offline,
- someone is gathering links/screenshots,
- admins are asleep or in another timezone,
- indexing/review just takes a bit.
During that lag, people could still place bets.
Most people were acting in good faith. But the system itself allowed a bad outcome: someone could place a bet after the result was already known, then still participate in settlement as if it was a real prediction.
At that point, we’re no longer rewarding forecasting. We’re rewarding timing and information asymmetry.
What we changed
We added one explicit field to proof submission:
outcome_known_at
This is the approximate timestamp when the outcome became known in the real world.
At resolution time, that timestamp is treated as a cutoff:
- Bets placed before the cutoff are valid.
- Bets placed at or after the cutoff are invalid for settlement and fully refunded.
Simple rule, big impact.
Why this is the right tradeoff
There are really two fairness questions here:
- Is it fair to accept bets made while uncertainty still exists?
- Is it fair to accept bets made after uncertainty is gone?
We want a strong “yes” for #1 and a clear “no” for #2.
Without a cutoff, those two cases get mixed together, especially when proof is delayed. With a cutoff, we separate them cleanly.
That does not punish legitimate users. It protects them.
A concrete example from HivePredict
This is the exact kind of market we are talking about:
It was also the first early market prediction on HivePredict.
The market was set to run through the end of March 2026, but the real-world event happened much earlier. As of February 28, 2026, the market was already resolved YES.
External reporting on February 28, 2026 described a joint U.S.-Israel operation against Iran (for example, AP reporting here and AP photo coverage here).
That creates a timing window:
- the outcome becomes clear in the real world,
- proof is still being gathered/submitted,
- the market may still look open to users.
Without an outcome-known cutoff, bets placed in that window could still get counted in settlement. That is the unfair edge we wanted to remove.
With this update, proof includes outcome_known_at, and settlement can treat bets placed at or after that point as invalid and refund them.
One clarification: sports markets are not where early resolution applies, so this US-Iran market is the right real example for this rule.
What this means for each group
Bettors
If you’re betting while the outcome is genuinely unknown, nothing changes for you.
If a bet is placed after proven outcome time, that bet is refunded instead of being treated as a valid position.
Market creators / proof submitters
When using early resolution on active manual markets, include outcome_known_at in your proof.
Think of it as adding one extra line of context: “this is when uncertainty ended.”
Why we didn’t just “close betting faster”
A fair question.
Could we just auto-close instantly based on external signals? In theory, maybe. In practice, markets span different categories and evidence quality varies. A rigid auto-close system can create false positives and even more disputes.
This update keeps human review where it matters, but closes the exploit path:
- proof establishes the outcome-known moment,
- settlement enforces it deterministically.
Reliability and safety details
A few implementation notes that matter:
- Active early-resolution manual markets now require
outcome_known_at. - Timestamp input is validated (must parse, and can’t be future-dated beyond a small tolerance).
- Proof APIs/UI now surface this timestamp for transparency.
- Resolution records include metadata about cutoff-based refunds.
- Refunded post-cutoff bets are excluded from winner/fee math.
That last point is important: we don’t just refund those bets; we also remove their influence from payout distribution.
What this change is really about
This isn’t just about one field or one handler update.
It’s about enforcing a principle:
Prediction markets should reward being right under uncertainty, not being late after certainty.
Early resolution still gives us speed.
Now it also gives us better fairness guarantees.
And when those two align, the market is healthier for everyone.