Dispute of the week · #2 · 31 October 2026

The Russia-Ukraine peace talks market

Vote: too early
Yes proposed twice and disputed twice, UMA vote round 10375 returned "too early" on 26 Oct, question reset on Polygon. No new proposal as of 30 Oct 23:27 UTC. Signed record · verify it

Predge, 31 October 2026. Prepared on 23 October from data read on 22 October 2026, 21:20 to 23:40 UTC; the vote result and the state after it were read on 30 October 2026, 23:10 to 23:40 UTC. All times UTC.

Market: "Russia-Ukraine peace talks by October 31, 2026?" (Polymarket market 4118937, condition id 0x1d78389c…cf5192). Yes needs an in-person diplomatic meeting between official representatives of Russia and Ukraine (indirect meetings through authorised mediators count) aimed at a ceasefire, peace agreement or other substantive de-escalation, by 31 October 11:59 PM ET (03:59 UTC on 1 November), publicly acknowledged by either government or reported by a consensus of credible media. Talks limited to humanitarian issues, calls and remote meetings do not count.

Status on 30 October: not resolved. A Yes proposal on 21 October was disputed, re-proposed and disputed again within four minutes, and the outcome went to UMA vote round 10375. On 26 October the vote returned "too early", UMA's answer for a question that cannot be settled yet. That does not resolve the market: the adapter reset the question and asked for a new proposal. At 23:27 UTC on 30 October nobody had proposed again, and Yes was at 0.009 in the last CLOB sample (22:30).

Timeline

Time (UTC)What happened on chainSource
2 Sep (block 93,092,916)Question initialized on Polymarket's UMA CTF adapter. Reward 5 USDC.e, bond 500 USDC.ePolygon tx 0xf96e47fe…0666
21 Oct 23:33:05Proposal: Yes, proposer 0x6c16…48bfa0x24b0d730…2cf0
21 Oct 23:33:15First dispute, 10 seconds later, disputer 0x4c64…c531. Question reset, new request in the same tx0x104dfa4e…587c
21 Oct 23:36:05Proposal: Yes again, same proposer0x98781089…2095
21 Oct 23:36:53Second dispute, 48 seconds later, from a different disputer 0xbcb1…fa37. The outcome now goes to a UMA vote0x127e282b…aa74
22 Oct 00:10:47First disputed request added to UMA VotingV2 on Ethereum, round 10375 (settles the first proposal's and dispute's bonds)Ethereum tx 0x45190cf5…0342
22 Oct 00:15:35Second disputed request added to VotingV2, round 10375. This vote decides what happens to the questionEthereum tx 0xbc2ec592…d859
23 Oct 01:37:05Bulletin board, question creator: quotes the in-person rule and says the 20-21 October meetings a US special envoy held separately with Ukrainian officials in Oslo and with Russian officials in St. Petersburg do not qualify, because representatives of both countries must be present at the same meetingPolygon tx 0xa10dc38b…4ee4
26 Oct 00:00:00Round 10375 ends (getRoundEndTime(10375)), 96.4 hours after the second disputeeth_call
26 Oct 00:00:23Both requests resolved on VotingV2, round 10375, with price -2^255 (the minimum int256). Under UMA's yes/no rules that value means "too early": the request came before the question could be answeredEthereum tx 0x6cbc5494…550e, block 26,273,138
26 Oct 00:41:26QuestionReset on the adapter. "Too early" does not resolve the market; the adapter sends a new price request to the oracle (same 5 USDC.e reward, 500 USDC.e bond, 250 USDC.e final fee)Polygon tx 0x406ec830…0ef7, block 96,179,943
30 Oct 23:27:15The new request is still in state Requested (getState = 1 on the oracle): no proposal yeteth_call at block 96,464,919

From first proposal to the second dispute took 3 minutes 48 seconds. From the second dispute to the reset took 97.1 hours, and the question ended up back where it was before 21 October: open, waiting for a proposal.

Price and size

What settlement logic should assume

None of this is a view on Yes or No.

  1. A proposal is a claim, not a result. Someone put up a 500 USDC.e bond saying Yes more than ten days before the deadline, while the market priced Yes near 0.05. A proposal can arrive before the end date; it settles only if it survives its window.
  2. A dispute in seconds is a signal, not an answer. Both proposals were challenged within a minute. Fast disputes say someone is watching this question; they do not say who is right.
  3. A vote does not always end the question. Here the vote returned "too early", and the market went back to waiting for a proposal. An agent that treated "vote finished" as "market settled" would have been wrong on 26 October and still be wrong today.
  4. After the second dispute, plan for days. The vote round ended 96.4 hours after the second dispute. Across Polymarket disputes from 1 January to 2 October 2026, the median from a second dispute to on-chain resolution was 88.5 hours. A "too early" vote adds the time to a new proposal and its 2-hour window on top.
  5. Read the chain, not only the venue's status field. At 23:10 on 30 October, nearly five days after the reset, Gamma still reported umaResolutionStatus: "disputed". On chain the question had a fresh request and no proposal.
  6. Do not move collateral on the spike. Yes went from 0.05 to 0.17 on a disputed proposal and came back. An agent that read the proposal as news would have bought the top of a claim the vote later called too early.

Context: from 1 January to 2 October 2026, 853 of 2,543 settled disputed Polymarket markets (33.5%) settled differently from the disputed proposal. That is a base rate, not a forecast for this market.

What the signed record returned on 22 and 30 October

GET https://api.predge.io/v1/settlement-risk/4118937 (free, no key). Read at Polygon block 96,004,564 (22 Oct 23:36:57), while the vote was open, it returned:

Read again at Polygon block 96,464,464 (30 Oct 23:15:52), the same URL describes the new request after the reset:

The 22 October snapshot record-4118937-2026-10-22.json (body sha256 5a39a9a6fd6b140dc992d26e2cf01cc6563dac3b13fd18ad1c5347d90466f80e) passes all 15 checks of verify_record.py, and the 30 October snapshot record-4118937-2026-10-30.json (body sha256 ef0ce45ba152aa3312e42e4b3129e79f1b7b660e6b6782bdbd81b2db6e1a6cfb) passes all 16, including the ed25519 signature over the record and over the whole HTTP body. Both are signed with key 3b6551a6439b9bda, our current key after the rotation of the old one.

How to verify:

pip install cryptography
curl -sD headers.txt -o record.json https://api.predge.io/v1/settlement-risk/4118937
curl -sO https://data.predge.io/verify_record.py
python3 verify_record.py record.json --body-signature "$(grep -i '^x-predge-body-signature' headers.txt | cut -d' ' -f2- | tr -d '\r')"

Or paste a snapshot into https://data.predge.io/verify.

Next week

"Will Trump say "Two Percent" during his Midterm Rally in Lubbock, Texas?" (Polymarket 6279816). A No proposal was disputed at 00:43:38 on 28 October and a Yes proposal at 04:21:53; both requests are in UMA vote round 10378, which ends 1 November 00:00 UTC. At 22:00:04 on 28 October the question creator posted on the bulletin board that the disputed "49.2%" in the rally stream qualifies for Yes. Open interest 198,006 USDC on 30 October.

Sources

Files

Not advice and not a view on any market. Predge reads public chain data and signs a verifiable record of what it read.