// SETTLEMENT AUDIT — PREDICT.FUN

Predict.fun settlement audit,
recomputed from the archive.

Since 2026-09-30: 11,520 markets recomputed, none disagree with the official outcome.

Each Predict.fun crypto Up/Down market in the shipped archive, settled again from the price stream it names and compared with the outcome its published start and end prices imply. Three results, counted over all of them: agrees, disagrees, or cannot be settled from the archive — with the reason. Not a sample, and not only the markets that check out.

Last 10 days · markets audited11,520
agree with the official outcome11,32598.3%
disagree00%
cannot be settled from the archive1951.7%

Markets that closed on the last 10 archived UTC days, 2026-09-30 to 2026-10-09. Of the markets the archive can settle, 100% agree.

Not settleable means a second the verdict needs is not in the archive — an upstream stream gap, an early market without a strike, or an outcome not yet published. It does not mean a value disagrees: disagreements are counted on their own.

Not in this audit: 750 markets that settle on a Binance price, which the archive does not collect.

How each market is checked

  • Predict.fun publishes a start and an end price for each market, and the outcome follows from them: Up when the end is higher, Down when it is lower, a tie when they are equal.
  • We take the same two prices from the Chainlink stream Predict.fun forwards, at the exact opening and closing seconds, and compare them at the precision Predict.fun publishes. A missing boundary second is listed as not settleable, never filled in from neighbouring ticks.
  • Markets whose price source is Binance rather than Chainlink are listed as not settleable: that price is not in the archive.
  • Whether a market has settled is read from its end price, not from its status field, which can lag behind it.

The rule, the files and a script to run it yourself: How a market settles →

Why some markets cannot be settled from the archive

Every market in the audit that is not counted as agreeing or disagreeing is here, grouped by cause.
WhyLast 30 daysAll history
Upstream stream gap192 · 1.7%192 · 1.7%The settlement stream stopped publishing around the market’s opening or closing second, so the second the verdict needs is not in the archive. Nothing is filled in from neighbouring ticks.
Outcome not yet published3 · 0.026%3 · 0.026%The archived market record has no official outcome yet.

Not in this audit: 750 markets that settle on a Binance price, which the archive does not collect.

Each reason, all history
96 · 0.8%The settlement stream was flowing, but its report for the closing second is not in the archive.
96 · 0.8%The settlement stream was flowing, but its report for the opening second is not in the archive.
3 · 0.026%The archived market record has no start or end price yet.

Every disagreement

Listed as recomputed.

No disagreements in the audited days.

Since the archive began

All 10 audited days since 2026-09-30, month by month.
MonthMarketsAgreeDisagreeNot settleable
All history11,52011,325 · 98.3%0195 · 1.7%
2026-1010,36810,197 · 98.4%0171 · 1.6%
2026-091,1521,128 · 97.9%024 · 2.1%

By asset and market period

MarketMarketsAgreeDisagreeNot settleable
BNB 5m2,8802,833 · 98.4%047
BNB 15m960942 · 98.1%018
BTC 5m2,8802,833 · 98.4%047
BTC 15m960942 · 98.1%018
ETH 5m2,8802,833 · 98.4%047
ETH 15m960942 · 98.1%018

By day

Each day’s file lists every market with its verdict and the boundary ticks it rests on.
Day (UTC)MarketsAgreeDisagreeNot settleableVerdicts
2026-10-091,1521,14606JSON
2026-10-081,1521,116036JSON
2026-10-071,1521,128024JSON
2026-10-061,1521,122030JSON
2026-10-051,1521,14606JSON
2026-10-041,1521,116036JSON
2026-10-031,1521,134018JSON
2026-10-021,1521,14903JSON
2026-10-011,1521,140012JSON
2026-09-301,1521,128024JSON

What this covers

Predict.fun crypto Up/Down markets only. Polymarket has an audit of its own: Polymarket settlement audit →

Related