UMA Dispute Resolution Time: How Long Does It Actually Take?
If your Polymarket capital is locked in a UMA dispute, the only question that matters is: when do I get my money back? The official UMA documentation describes the mechanism but doesn't publish empirical timing data. We've measured it across 1,654 actual disputes from our own dataset. Here's what the numbers say.
TL;DR — Resolution Time Distribution
| Percentile | Time to Resolve | Reading |
|---|---|---|
| p50 (median) | ~2 days | Most disputes clear here |
| p75 | ~4 days | Subjective questions sit in this range |
| p90 | ~7 days | Slow voter participation, hard evidence |
| p95 | ~14 days | One in twenty |
| p99 | 3+ weeks | Re-disputes, escalation, edge cases |
If you're trying to plan around a frozen position: assume 2 days, prepare for 7, accept the small chance of 3+ weeks.
The Four Phases of a UMA Dispute
Total resolution time is the sum of four sequential phases. Knowing them helps you predict where in the process a dispute is and how much longer it'll take.
Phase 1: Liveness Window (2 hours)
After a proposer asserts the outcome, there's a 2-hour window during which anyone can dispute. If nobody does, the market resolves. If anyone disputes, we move to Phase 2.
You can't do anything during this phase. Most market resolutions clear here without ever becoming a dispute.
Phase 2: Commit Phase (~24 hours)
Once disputed, the question goes to UMA's Data Verification Mechanism (DVM). UMA token holders have ~24 hours to commit a hashed vote (YES, NO, or "too early to tell").
This is fixed-length. It doesn't matter how easy or obvious the answer is — every disputed market sits here for ~24 hours.
Phase 3: Reveal Phase (~24 hours)
Voters reveal their committed votes. Another fixed ~24-hour window. The reveal must match the commit (using a salt) — this prevents voters from changing their mind based on others' votes.
At the end of Phase 3, the resolution is theoretically determined. But it isn't final yet.
Phase 4: Settlement / Re-dispute Window
The resolved outcome is settled on-chain. This is where most variance comes from.
If the result is uncontroversial, settlement happens within hours and your funds unlock. If a participant believes the DVM vote got it wrong, they can re-dispute, restarting Phases 2 and 3 from scratch. Re-disputes are uncommon but they're the main driver of the long tail.
Best Case vs Worst Case
Best case: ~2 days
Question is unambiguous (e.g. "Did the Knicks beat the Celtics on April 5?"), DVM votes cleanly in one cycle, no re-disputes, settlement happens quickly.
Typical case: 3–5 days
Question has some ambiguity but evidence is clear enough that voters converge. One DVM cycle resolves it. Settlement within a few hours.
Worst case: 2–4+ weeks
Multiple re-disputes. Question genuinely ambiguous (e.g. ceasefire definitions, "clear evidence of" claims). Voter quorum issues. Each re-dispute adds another ~2-day cycle.
Real example: Some 2024 ceasefire markets were disputed twice and took 18+ days to resolve.
What Drives the Long Tail
From our analysis of the slowest disputes, three factors dominate:
- Subjectivity. Questions where reasonable people disagree on the answer. Voters split, narrow margins, re-disputes likely.
- Evidence quality. Markets resolved by "media consensus" require voters to actually read sources. Markets resolved by FIFA or the FOMC don't — the answer is one click away.
- Stake size. Larger markets attract more attention and more disputes. Whale positions get re-disputed more often because the financial incentive to challenge is higher.
How to Estimate When Your Specific Dispute Will Resolve
Check three things on UMA's voter app (vote.uma.xyz):
- What phase is it in? Commit → +24h to next phase. Reveal → +24h to settlement check.
- Is the question objectively answerable? If yes, expect resolution at the end of Phase 3. If no, expect at least one re-dispute.
- What's the current vote split? Tight votes (55/45) tend to be re-disputed. Lopsided votes (90/10) usually settle cleanly.
How to Avoid Disputes Entirely
The cheapest dispute is the one you never enter. Across our dataset, three patterns predict 90%+ of disputes:
- Subjective thresholds — "clearly," "significant," undefined judgement calls
- Ambiguous key terms — "ceasefire," "regime change," "officially declared"
- Weak resolution sources — "according to media reports" instead of a single authoritative source
Markets with two or more of these signals are disputed at 8–10% — vs the 1.14% baseline. Skip them, or use an automated tool to flag them before you enter.
The Bottom Line
Median UMA dispute resolution is about 2 days. The 95th percentile is two weeks. The worst 1% can drag on for a month or more, almost always because of re-disputes on subjective questions. If you've got capital frozen, assume the median, plan for the p90, and check the dispute phase on UMA's voter app to refine the estimate.
OracleMangle scores every Polymarket market for UMA dispute risk before you enter. Free Telegram bot — check any market in seconds.