Digest Polymarket & UMA Discord

Daily digest from Polymarket & UMA Discord. Disputes, bugs, clarifications, announcements.

[Digest Polymarket & UMA Discord @predictdigest]

POLYMARKET DIGEST — 27 OCTOBER 2026

⚠️ Bugs & Issues

API/CLOB latency crisis (ongoing): Order placement and cancellation latency has been consistently above 500ms, with some users reporting over 1000ms from Dublin. The issue has worsened since an October 21 infrastructure incident. Multiple market makers report the situation is fatal to FAK orders and unpredictable tail latency makes market making a "lottery." vader1208 confirmed their team has halted trading entirely, stating it is "just impossible" to operate. piercethecircle noted other large accounts have also stopped trading.

skulldoom97 (Polymarket dev) provided an official update outlining three mitigation approaches:
1. WebSocket improvements (coming soon, but won't fix core CLOB slowness)
2. Further DB tuning (limited options without architectural changes)
3. Rate limit adjustments as a last resort, with prior notice given

skulldoom97 attributed the root cause to increased load from the fee holiday and sports season returning, and confirmed lag is especially severe on high-activity markets (e.g. BTC 5m). A new exchange is in development with much higher throughput; it will roll out in stages (sandbox → beta → broader release). Formal announcements are expected within 1–2 weeks. vader1208 requested that current maker-tier rate limits be preserved in the interim.

LP rewards accrual interval changed: muzzammil2 reported the rewards page switched from showing accumulation every 1 minute to every 5 minutes, asking whether this was intentional.

Typhoon market inconsistent resolution sources: brassbonanza4655 flagged that in the new "How strong will In-fa be?" market, the "Violent" outcome uses JMA as the resolution source while "Super Typhoon" uses JWTC — two agencies that measure wind speed very differently. This could lead to contradictory resolutions and may be a market setup bug.

RTDS v2 activity topic not migrated: pinner reported not receiving v2 trades from the RTDS activity topic. kartojal (dev) confirmed the activity topic has not yet been migrated to v2 and is investigating.

📋 Clarifications

NBA Kings market possible misresolution: dipsybitsy and crypt0blonde raised a case where a Kings NBA quarter-spread market (Kings 3Q -1.5) appears to have been resolved incorrectly via UMA, with the score feed corrected after the proposal was submitted and the dispute window too short for anyone to react. crypt0blonde argued Polymarket should issue a refund as platform responsibility, regardless of UMA's role.

If latency keeps market makers offline, does Polymarket risk losing liquidity depth permanently before the new exchange launches?

[Digest Polymarket & UMA Discord @predictdigest]

UMA DIGEST — 27 OCTOBER 2026

🔥 News & Announcements

A massive wave of disputes hit Bolivian soccer markets. A proposer submitted results for multiple matches (including "potos"-tagged markets) within one hour of game end instead of the required two hours, and the resolution source had not yet updated. rex416. filed disputes backed by double archives (Wayback Machine + https://t.co/SW2ljJBBlN), making the "cached page" defense unavailable. im.the.wizard noted the proposer is already down roughly $34k with an additional ~$7.5k in pending golf losses. All exact-score proposals were disputed; the proposer continued submitting new proposals even after disputes were filed.

Earlier in the day, two Indonesian football markets (idn1-bpi-mak-2026-10-26 and idn1-ps-mau-2026-10-26) were also flagged as early proposals, with the resolution source showing both matches still live at proposal time.

⚠️ Bugs & Issues

The UMA bot for proposal requests was reported not working by elie220 on the morning of Oct 26. infinity_d confirmed it was resumed and fixed later that day.

NBA stats inconsistency: im.the.wizard flagged that official NBA stats briefly showed James Harden with 22 three-pointers before correcting to 19. mawyoo confirmed other sites also adjusted. No dispute needed.

📋 Clarifications

A market asking whether Democratic House incumbents lost primaries (capped at September) was proposed resolved "No" on the basis that reaching 13 losses is mathematically impossible — 7 lost votes, 4 filed then left the race, and the Louisiana special election is a new election outside the market scope. justinnocent_ stated the proposal was vetted with multiple proposers and confirmed as good by Cruz. armanh01 and laurleo agreed it appears fine.

⚡ Disputes

A heated exchange broke out between exquisite_wolf_79651 and alexchunus, with alexchunus accusing exquisite_wolf_79651 of "intentionally and knowingly misresolving oracle disputes" and front-running, while exquisite_wolf_79651 called alexchunus a "convicted scammer." exquisite_wolf_79651 suggested a video meeting be arranged with alexchunus (Penguin), joes49572 (0xWater), jason.k7 (Jason), and nickc01457 (NickC), implying multiple accounts belong to the same person. abhiz69 noted the wallet in question has been doing the same thing with multiple wallets and was previously removed by UMA. No resolution to this dispute was reached.

When resolvers have a history of multi-wallet tactics and accusations of intentional misresolution fly, at what point should UMA consider stricter identity verification to maintain oracle integrity?

[Digest Polymarket & UMA Discord @predictdigest]

POLYMARKET DIGEST — 26 OCTOBER 2026

⚠️ Bugs & Issues

Multiple traders reported significant API and infrastructure degradation starting around 21:00 UTC on October 25. 001_raghav noted POST /order latency jumping from ~220 ms to 500–1,500 ms since 12:00 UTC on Oct 25, potentially linked to the Oct 22 CLOB update. berabuilder and berth6833 confirmed p99 cancel latency exceeding 2 seconds on esports markets, with WebSocket delays also severely impacted. r33nter and pool6225 reported order kill confirmations taking 600 ms to 10+ seconds. The issues appear to be known but unresolved.

Additionally, pog_is_pog flagged a PnL discrepancy — their balance increased by ~$300 with no trades, redemptions, or deposits. satatardy noted PnL display in the UI has been broken for a long time with no fix.

bluvv_.xo flagged a potential rounding error in weather market rules: a reading of 16.5°C rounds to 16°C under half-to-even rounding, not 17°C as the rules appear to assume.

hellbot632 reported the UI cracking with more than 2,000–3,000 resting maker orders.

crypto_hamster raised a detailed concern about the taker delay mechanism on BTC 5-minute markets: held orders are released inconsistently (100 ms to 1+ seconds rather than a flat 150 ms), cancellation of held orders via precomputed order ID is refused, and minimum expiration is enforced at 180 seconds. One wallet (0xcd30) appears first in the post-hold queue ~1,500 times in September, 4–7 ms after a trade clears, raising questions about queue ordering (submit time vs. other criteria).

📋 Clarifications

Cape Town October 18 temperature markets have been updated. Due to a National Weather Service outage, all Cape Town high-temperature markets (9°C or below through 19°C or higher) will now resolve using the Weather Underground Daily Observations Table. The previously identified erroneous 19°C reading (originating from another airport and incorrectly assigned to FACT) remains excluded from resolution across all markets. The consensus view, confirmed by hans32312, is that the actual peak temperature was 16°C and markets should resolve accordingly. ak504111 confirmed the clarification was posted, and hans32312 subsequently submitted the resolution proposal.

velvet_soulb raised several open questions about the Protocol V2 migration (scheduled for Nov 2 for net-new markets), including: whether recurring BTC/ETH/SOL/XRP 5-minute Up/Down markets will launch as V2 from day one, slug naming conventions, auto-redemption behavior for deposit wallets, whether fee schedules and the 150 ms taker delay carry over, and whether a discovery endpoint will exist for upcoming recurring markets. 2773380 echoed broader frustration about insufficient advance communication around deprecations, noting that real money is at stake.

Given the multiple infrastructure issues and unresolved bugs, do you think Polymarket should pause new market launches until core API and UI performance is stabilized?

[Digest Polymarket & UMA Discord @predictdigest]

UMA DIGEST — 26 OCTOBER 2026

🔥 News & Announcements

Brazil's presidential runoff was called by the superior electoral court, with Flávio Bolsonaro refusing to concede and demanding annulment. Brazil election markets on Polymarket were proposed by non-whitelisted users and settled without dispute. armanh01 noted the market became resolvable when the electoral court made the official call.

7h0nk3r requested that extensions on the Trump Michigan rally speech brackets be undone, noting a typo on the page (October 26 vs October 25) caused extensions, while several brackets — including 3 Yes brackets — had already settled. He also argued that the "No Qualifying Event" bracket should now settle No since it cannot logically coexist with settled Yes brackets.

⚠️ Bugs & Issues

im.the.wizard flagged a misresolution on the Kings 3rd quarter lines market, which kept flipping between 29-29 and 29-31. The spread market was extended and then unexpectedly settled, while the total was disputed. Users were directed to contact Polymarket directly for a refund on the incorrectly settled spread.

📋 Clarifications

Archived Brazilian LGBT markets: Three Polymarket markets on LGBT federal deputies in the 2026 Brazilian elections were found to be archived yet proposed. 7h0nk3r clarified that archived markets can still be proposed if the outcome is determinable from the ancillary data. laurleo confirmed proposals were correct per the resolution source. The markets were likely archived due to their sensitive subject matter. One proposer self-disputed after realizing uncertainty in one bracket, saving ~$500.

Israel/Iran ceasefire markets: A dispute arose over early proposals on ceasefire markets following the Sirik explosion. armanh01 and 7h0nk3r argued markets should stay open given the unclear attribution of the strike and the active, escalating conflict. A US/Iran clarification confirmed that all related markets should remain open for 3 days from the incident. armanh01 noted that UMA Rocks voted against this and were slashed as a result — notably the first round in which Rocks chose not to publicly post their vote.

Color vs. Ex-Zero esports markets: nopsuke raised a concern that several Color match markets were proposed before HLTV (the designated resolution source) had confirmed the walkover result. The debate centered on whether the 2-hour rule applies from the moment the forfeit was declared by the TO or from the scheduled match start time. alexchunus noted the rules require only 2 hours from the event's conclusion, and since the walkover was declared 7+ hours before proposals, most participants leaned toward the proposals being valid. nopsuke maintained that the resolution source should have been waited on; HLTV did eventually update confirming the result.

⚡ Disputes

im.the.wizard vs. committee member (corruption allegation): im.the.wizard publicly accused a committee member (not named but tagged) of corruption and manipulation, stating: "I'm glad you are changing your process to not post anymore, but I hope you can recognize it's still a large problem to have a fraudster and manipulator on the committee." No formal response was recorded in the channel.

When resolution sources like HLTV conflict with market timing rules, should the community prioritize waiting for official confirmation or stick strictly to the 2-hour window from event conclusion?

[Digest Polymarket & UMA Discord @predictdigest]

POLYMARKET DIGEST — 25 OCTOBER 2026

⚠️ Bugs & Issues

Significant API and latency issues were reported throughout the day. Starting around 14:00 UTC, multiple users reported extremely high place order latencies — hft_pro noted P50s above 500ms on sports markets, while phoenyx69 reported P50 of 300ms+ on 5-minute crypto markets. laucyph3r described it as "a massacre" on 5m crypto latency. Later, exuberant_lynx_38171 reported Taker P50 of 550ms and P90 as high as 1700ms. skulldoom97 confirmed the team was investigating. Latencies reportedly improved somewhat around 15:52 UTC (hft_pro noted P50 returning to ~35–40ms on sports), but issues persisted for crypto.

Additionally, berth6833 reported recurring HTTP 500 errors ("order queue full for maker") and "Context deadline exceeded" errors, raising frustration about wallet queue limits despite respecting rate limits. pepesong reported random heartbeat errors, and later noted that Redeem is failing. berth6833 also flagged that user fill events are being sent with significant delays again.

📋 Clarifications

The Shenzhen lowest temperature market for October 21 remains unresolved. hans32312 (VIP) noted it should be resolved soon.

Regarding combo last look access for v2 migration, quantwifbag reported filling out the access form and meeting volume requirements but receiving no response, and asked whether the team reaches out via email/telegram or enables it automatically — no answer was provided yet.

How much latency is too much before you'd consider moving your trading volume to another platform?

[Digest Polymarket & UMA Discord @predictdigest]

UMA DIGEST — 25 OCTOBER 2026

⚡ Disputes

Israel–Iran Strike Market (Sirik explosions): A dispute is active over whether the Sirik explosions qualify as an Israeli strike. jenia9419 proposed YES resolution was pushed through, arguing there were zero credible reports implicating Israel and that Polymarket's "Too Early" clarification on the US–Iran ceasefire market is irrelevant here. gum5683 disputed, arguing there was a non-zero chance Israel was responsible and that the "Too Early" clarification for the US–Iran market cannot be applied to the Israel–Iran market, where YES remains appropriate. armanh01 noted that Polymarket has accepted the Sirik strike's attribution as unclear, and criticized the proposal for potentially scamming NO holders in a high-volume market. The dispute appears likely to resolve in jenia9419's favor as gum5683 missed the window to dispute a second proposal while sleeping.

Ukraine targeting Moscow on October 21: A NO proposal was put forward on the market asking whether Ukraine targeted Moscow on October 21. tranquil_dolphin_46729_72282 contested this, citing AP, TASS, and Kyiv Post all reporting that 630 drones were directed toward the Russian capital from 20:30 on October 21. The NO side argues that the drones near the city were only intercepted after midnight on October 22. tranquil_dolphin_46729_72282 sought others to dispute but admitted having only $300 in their wallet — insufficient to fund a dispute — and was largely dismissed after it emerged they were using a purchased Discord account. The market resolved NO without being disputed.

András Schäfer 1+ Shots market: ostrich_knight flagged that a market was proposed NO despite ESPN, Flashscore, and Sofascore all recording a shot for András Schäfer in the Augsburg vs Union Berlin match.

⚠️ Bugs & Issues

NBA score data was reported as unreliable, with scores randomly changing back and forth across sources including the official NBA website. whoooo8888 noted the NBA site was in a particularly bad state. Separately, bromley86 flagged that the Polymarket resolution UI is showing markets as finally resolved when they are still in review, and even appears to allow share purchases after a final outcome is displayed — confirmed as a cache/display issue.

When attribution is genuinely unclear but market volume is high, should disputes prioritize protecting majority holders or setting strict evidentiary standards?

[Digest Polymarket & UMA Discord @predictdigest]

POLYMARKET DIGEST — 24 OCTOBER 2026

📋 Clarifications

NATO x Russia military clash markets: Polymarket issued a long-awaited clarification on the "NATO x Russia military clash" series (Oct 31, Nov 15, Nov 30, Dec 31). The October 17 incident — two Romanian F-16s shooting down two Geran-family drones over Romania's territorial waters — does not qualify as a NATO-Russia military clash. Per the rules, interceptions of munitions targeting a third party (in this case a civilian/commercial vessel, not Romania or its forces) do not count, even if the vessel was within Romanian territorial waters. Order books were cleared at 4:00 PM ET on October 23.

Russia-Ukraine diplomatic meeting markets: All "Will Russia and Ukraine hold any diplomatic meeting by…" markets (Oct 31, Nov 15, Nov 30, Dec 31) received a bulletin: the October 23 meeting between Ukrainian Ombudsman Dmytro Lubinets and Russian Human Rights Commissioner Yana Lantratova in Belarus qualifies as a diplomatic meeting under the rules. These markets are expected to resolve YES.

US x Iran ceasefire market: The "ceasefire continues through October 22?" market remains open due to conflicting sourcing around reported October 21 explosions in Sirik. The market will stay open until October 24, 2026, 11:59 PM Iran Standard Time for the 3-day attribution window to expire.

Landskrona BoIS vs. IK Oddevold corners markets: All corners-related markets for this Superettan fixture (Total Odd/Even, Home O/U 4.5, Home O/U 5.5, Total O/U 7.5, Total O/U 8.5) will resolve using the official Superettan governing body website as the primary source.

2026 Spanish general election markets: Language on both "Will PSOE win the most seats?" and "Will PP win the most seats?" markets has been updated for clarity.

⚠️ Bugs & Issues

Severe CLOB latency degradation: Multiple users reported a significant and ongoing spike in taker order latency over at least two days. Typical figures cited: median around 544–623ms, P90 at 1,200–1,400ms, P99 reaching 3+ seconds — compared to a median of ~297–300ms and P90 well under 1s the previous week. habitual_revers confirmed the team is investigating the edge networking stack. papakarlo8, purpletyoma, dust7829, and exuberant_lynx_38171 all independently confirmed the issue persisting from EU-west-1 and other regions.

FAK order status unknown after taker delay: the_purple_person flagged that Fill-and-Kill (FAK) orders in markets with taker delay (e.g. sports) leave no order event and return null from GET /data/order, making it impossible to reliably determine whether the order filled without polling with an arbitrary timeout.

Self-trade prevention undocumented for predictions: dean_weez asked how self-trading is handled for prediction markets (as opposed to perps). hft_pro noted from observation that the taker order is not matched against the user's own maker order, and the maker order is ignored rather than cancelled — but this behavior is not officially documented.

V2 trades missing from RTDS stream: pinner reported that the RTDS activity/trades stream on wss://ws-live-data.polymarket.com appears to contain no Protocol V2 trades. Out of ~3.7 million RTDS trade records from October 23, none matched 99 V2 transaction hashes returned by the Data API. It is unclear whether V2 trades appear on a different subscription or are simply not yet included. Some RTDS payloads also have a populated asset field but an empty conditionId, which is unresolved. gasuajex confirmed investigating and testing.

Unrealized PnL non-zero despite closed positions: up0rabn1k reported the user-stats API returning a non-zero unrealized_pnl (-35.48) despite all positions being closed. Cause remains unknown.

WebSocket 1013 "slow consumer" errors: up0rabn1k flagged persistent 1013 slow consumer: send buffer full disconnections from Polymarket's WebSocket, reproducible across Java, Go, Python, and C clients, suggesting a server-side issue.

Which resolution criteria surprises you more — that drone interceptions don't count as NATO-Russia clashes, or that a human rights meeting qualifies as diplomatic engagement?

[Digest Polymarket & UMA Discord @predictdigest]

UMA DIGEST — 24 OCTOBER 2026

🔥 News & Announcements

A massive MLB proposal incident occurred when approximately 400 proposals were submitted while a game was still in progress following a call overturning. Most game-related proposals (excluding player props and overs) were disputed as P4. The UMA settlement bot processed and settled nearly all of them within ~15 minutes because at settle time the game was officially over, leaving disputers very little time to act. im.the.wizard confirmed being the proposer and acknowledged the loss as a "quarterly mistake."

Russia-Ukraine diplomatic meeting markets were extended well beyond the standard challenge period (~8 hours), frustrating many traders with large positions. infinity_d confirmed that extensions are applied by bot or manually, and that removing them is always manual. The markets were eventually held pending a Polymarket clarification; armanh01 noted the clarification came through confirming the meeting counts as YES. tsybka criticized the lack of communication during the ~8-hour wait, and armanh01 agreed that pending clarifications should be disclosed to all traders equally rather than kept silent.

⚠️ Bugs & Issues

Two Sao Paulo temperature dispute markets ("16°C" and "22°C" on Oct 19) showed "unable to settle" in the Oracle UI while all other similar disputes resolved normally. alexchunus attributed this to the UMA settlement bot experiencing lag due to high proposal volume over the weekend and advised waiting until morning.

📋 Clarifications

Superettan corners dispute (IK Oddevold vs Landskrona Bois): ghostmbao and eip15942 argued that the official https://t.co/autmTSsbgs source recorded only 4-2 corners due to a tagging flaw — corner kicks tagged as "goal" or "shot" are not counted in the stats widget. Six major tracking sites (FotMob, Sofascore, Flashscore, etc.) all show 7-2 (9 total corners). eip15942 noted the official source contradicts itself, showing 0 second-half corners for Landskrona despite their own match events feed and video showing a corner-kick equalizer at 51'. alexchunus argued against the "6 corners + 3 direct kicks from corner" interpretation.

Russia-Ukraine meeting market inconsistency: .rustyn raised a question about why a previous market (created Aug 19, meeting on Aug 21) resolved NO while the current market resolves YES under apparently the same conditions. tsybka and raskell1 clarified that the previous market always resolved off the Lubinets–Lantratova meeting, and that the prior NO resolution was due to a timing issue — the referenced meeting predated that market's creation window.

MLB spread proposals mid-game: laurleo and im.the.wizard discussed the edge cases around proposing spreads before a game ends. im.the.wizard noted that once 5 innings are complete, a game is official (including playoffs), and that spreads can technically be proposed once mathematically determined — e.g., a home team losing by 5+ cannot cover -4.5. laurleo confirmed he will vote P4 on the Brewers -4.5 dispute as nonsense, while cautioning proposers to be certain before attempting such proposals due to their complexity.

⚡ Disputes

P4 dispute involving im.the.wizard (esports/sports market): A major ongoing dispute centers on whether a market was proposed before the resolution source updated. im.the.wizard presented an archive snapshot and a video recorded in a fresh browser session showing the score was not yet published at proposal time. exquisite_wolf_79651 (cruz), a https://t.co/XaqaTFIn9q voting committee member, argued the archive may be inconclusive due to potential CDN caching and that the site was known to serve stale copies. im.the.wizard, abhiz69, joes49572, and alexchunus all accused exquisite_wolf_79651 of voting based on a personal grudge rather than evidence, pointing to his prior voting history where he accepted P4 on far weaker soccer disputes. abhiz69 stated: "Cruz, everyone knows it's because of your personal beef — everyone can see your past votes." exquisite_wolf_79651 maintained his reasoning was based purely on the evidence posted in dispute threads. 7h0nk3r stated he will vote P4 (supporting wizard), arguing the archive should take precedence over the video given trust concerns, but noted CDN caching is a real possibility. The dispute remains active and heading to a community vote.

Why should markets with ambiguous resolution sources allow proposals mid-game when the outcome isn't mathematically certain yet?

[Digest Polymarket & UMA Discord @predictdigest]

POLYMARKET DIGEST — 23 OCTOBER 2026

🔥 News & Announcements

Istanbul Temperature Markets – Dispute & Resolution
A large group of Istanbul temperature markets (highest and lowest temperature brackets for Oct 20–22) faced disputes after Weather Underground returned "The location supplied is invalid" for Istanbul Airport (LTFM). Per market rules, markets with no data by 11:59 PM ET the day following the observation date resolve to the lowest bracket. Order books were cleared at 11:00 AM ET on Oct 22.

For the Oct 20 highest-temperature market specifically, a dispute over whether coordinate-based WU data showing 21°C qualified as valid LTFM data was rejected — the market resolved 13°C or below. Polymarket confirmed no valid data existed under the rules, and the disputer lost their bond.

Rule change: All temperature markets created after Oct 22, 11:59 PM ET will resolve using METAR data. Existing markets remain under current rules. The community criticized the lack of a precise WU URL/station definition in the original rules; the team acknowledged the ambiguity.

ETH Price Markets – Oct 22 Resolution
The Binance ETH/USDT 1-minute candle closing at 12:00 ET on Oct 22 settled at $2,398.91. Key implied resolutions: Above $2,300 → Yes; Above $2,400 → No; Between $2,300–$2,400 → Yes.

Russia–Ukraine Diplomatic Meeting Markets
U.S. Envoy Steve Witkoff's separate meetings with Ukrainian officials in Oslo and Russian officials in St. Petersburg on Oct 20–21 do not qualify as a diplomatic meeting under market rules, as both countries' representatives were not physically present at the same meeting. Third-party mediators do not count as representatives of either country. This clarification applies to all related markets with deadlines from June through December 2026.

Tampa Bay Rays vs. Milwaukee Brewers MLB Markets
Rules clarified: if an inning is not played or the game is cancelled, inning-by-inning and first-five-innings markets resolve to Draw.

---

⚠️ Bugs & Issues

CLOB Performance Degradation (Ongoing)
Significant API/CLOB issues persisted over the past 24+ hours:
- FAK/taker order latency rose from ~315ms (Oct 19) to ~670ms, with p90 reaching ~1.4–1.5s.
- Cancel acks remain fast (~20ms); GET round-trips unchanged.
- WS fill latency severely worsened; one user measured a ~150ms (>1.5×) median increase.
- EU-based traders reported WebSocket messages routing via US (JFK CloudFront) instead of Dublin, adding ~70ms per message.
- GTC orders receiving repeated "can't be canceled because it is pending/delayed" rejections for up to 3,350ms after submission.

Staff member skulldoom97 confirmed the issue is load-related, not caused by recent changes, and that the team is investigating tuning options. No ETA provided.

Additional issues reported:
- gamma-api IPv6: DNS not returning an IPv6 address; reported to devs.
- "Market opening rate limit" error on POST /orders: undocumented rate limit causing developer errors; documentation formally requested.
- Builder API / session-key 403 error: Developers receiving 403 {"error":"not allowed"} from the relayer; directed to DM staff.
- API returning HTML instead of JSON: Some bots receiving raw Cloudflare error pages instead of API responses.
- Intermittent order placement/cancellation failures via API (sig type 3) reported; skulldoom97 suspects a Cloudflare issue and is investigating.

---

📋 Clarifications

- WebSocket snapshot staleness: WS subscriptions sometimes provide outdated snapshots with gaps in updates. Community requested sequence numbers and a unified order_book channel with verifiable update continuity.
- SDK feature request: Event filter for list_account_trades endpoint requested (already available for list_trades and list_activity).

Given the Istanbul temperature market dispute and the shift to METAR data, are prediction market rules better served by being overly specific upfront or flexible enough to adapt when data sources fail?

[Digest Polymarket & UMA Discord @predictdigest]

UMA DIGEST — 23 OCTOBER 2026

🔥 News & Announcements

Crypto Daily Above/Below markets experienced a rule confusion: on Oct 2, rules were silently changed; reverted on Oct 8 to Binance 1-minute candle closing at 12:01 ET; then silently changed again around Oct 10 back to 12:00 ET with no announcement. lonfus stated the platform is now reverting again to 12:01 ET for all markets from Oct 8 onward (including Oct 6–7 live markets), and promised that any future rule changes will be announced well in advance. The chaotic rule flip-flopping allowed user fernando5495 to make an estimated $125k–$150k+ by positioning correctly on the "Ethereum above 2400" market.

⚠️ Bugs & Issues

A Superettan match score (4–2) listed on the official resolution source (https://t.co/dc3rHkWTCg) appears to be a bug in their video tagger, according to eip15942. rex416. flagged that proposals based on that score differ from what the official resolution source shows, while the proposer (identified as GhostMbao, with possible alt accounts including aartirai) cited Sofascore/Flashscore as corroboration. enormouz noted the dispute requests bot was down.

im.the.wizard also flagged a Getafe vs. Lugano UEFA Conference League market proposed as 2–5 when the official source did not yet confirm that scoreline; he attributed the proposal to "corgi" and called it at minimum a P4.

📋 Clarifications

NFL player props (Bills-Bengals game): laurleo disputed several NFL player prop markets proposed during live play as P4, citing the rule that they must wait for the official NFL gamebook (published ~10 minutes after game end). laurleo noted precedent from the Damar Hamlin incident where stats from an abandoned game did not count. im.the.wizard and jessicaonlychild agreed props should wait for official stats. Markets for Taylor Swift and Colston Loveland were proposed after game end but before official stats — both im.the.wizard and laurleo considered letting those slide but leaned P4.

NBA vs. Uruguay soccer rule inconsistency: abhiz69 questioned why some voters applied P4 to Uruguay half-time markets but P2 to NBA markets, flagging the inconsistency.

Regarding the soccer HT market disputed during a bombing alert interruption: exquisite_wolf_79651 noted it was proposed after HT but the match was interrupted in the final 2 minutes due to a bombing alert. 7h0nk3r said that case is "reasonably debatable," while calling the NBA dispute (proposed mid-second quarter) "one of the most open-and-shut P4s" he'd seen. whoooo8888 agreed it was clearly P4.

⚡ Disputes

im.the.wizard vs. exquisite_wolf_79651 (soccer score dispute): im.the.wizard disputes a soccer market proposed by a rival proposer, claiming https://t.co/1GisDXum6c evidence and video footage prove the official source had not yet updated at the time of proposal (9:58), with the site updating only after 10:04–10:06. exquisite_wolf_79651 counters that cache issues could explain the discrepancy and challenged the validity of the archive as proof. The dispute escalated into mutual fraud accusations: im.the.wizard alleged exquisite_wolf_79651 accepted an on-chain bribe of ~$750 to intentionally misresolve a prior tennis dispute; exquisite_wolf_79651 denied it and accused im.the.wizard of fabricating evidence with associate "Scout." rex416. flagged the matter to a moderator.

lonfus crypto market dispute: lonfus bonded a resolution on the "Ethereum above 2400" market at 95 cents (Yes), apparently based on the 12:01 ET rule, but alexchunus argued this market category (Daily Above/Below) was never officially clarified to follow the 12:01 rule — only the Daily Up/Down category was. The market resolved No (ETH did not close above 2400), making lonfus's bond a loss. Community consensus was that a platform bail-out/clarification in lonfus's favor would be improper given the written rules.

If UMA's rule changes keep catching even experienced traders off-guard, should the platform prioritize a single immutable rule version over the flexibility to adjust for edge cases?

[Digest Polymarket & UMA Discord @predictdigest]

POLYMARKET DIGEST — 22 OCTOBER 2026

🔥 News & Announcements

Protocol v2 is tentatively launching on new markets on November 2. Traders on BTC 5-minute markets are advised to prepare for migration immediately, as each 5-minute market is treated as a new market.

Maker rewards on weekly/monthly stock and commodity hit markets have been slashed to ~$1/day (roughly 1/10th of previous levels), causing wide spreads and reduced liquidity. Users warn the taker fee holiday is ineffective without adequate market maker incentives.

⚠️ Bugs & Issues

Order latency degradation (Oct 21–22): p50 latency rose from 38ms to 75ms; p99 from 513ms to 2,791ms; order timeout rate increased from 0.015% to 0.285%. Users baka_34910 and vader1208 noted zero communication from Polymarket. Separately, market WebSocket latency and random connection drops were flagged as an additional concern.

API & endpoint errors: Users reported HTTP 429 "market opening rate limit" errors immediately after new market announcements (lunmoonatik, miships), and 500 Internal Server Error ("order timed out") on the /order endpoint (pumpty_dumpty). The Gamma API was also reported down for some users on Indian IP VPS.

Weather market infrastructure: An NWS (https://t.co/rXv8V4e9Ds) outage is affecting non-US city observations. Fallback rules apply — if NOAA data is unavailable by 11:59 PM ET on the day following observation, Weather Underground (WU) Daily Observations will be used; if no data at all, the market resolves to the lowest bracket. Istanbul has no WU data available and is expected to resolve to the lowest bracket. A bug on WU causes LLBG (Tel Aviv) searches to redirect to Haifa airport; ak504111 is investigating. The Denver resolution station was switched from KBKF to KDEN without public notice.

Suspicious queue behavior: User crypto_hamster flagged wallet 0xcd30457c…2dd0 for consistently appearing first in queue on new price levels with ~5ms reaction latency — faster than physically possible from AWS Dublin (~40ms round-trip). Community speculation suggests pre-placed post-only orders rather than reaction-based trading. User bootypool separately alleged preferential queue positioning for certain wallets.

Other issues: joaocra2 could not view or place orders post-maintenance (workaround: clear cache/cookies or change browser); darko5586 reported missing taker rebates despite >$7,800 in trades and Bronze tier status; lvn_szn reported a missing maker rebate payment; edgod3714's KYC application has been pending nearly one month; Builder API approval requests from tchicochico (waiting since Oct 9) and inmind5624 were resolved after staff member gasuajex intervened.

📋 Clarifications

WNBA GSV vs. ATL (Oct 22): All associated markets (moneyline, quarter, half) will resolve based on the event dated Oct 21 (wnba-gsv-atl-2026-10-21).

CS2 — 9INE Academy vs CTRL Esports: CTRL Esports was disqualified from United21 Season 57 for suspected match-fixing and roster fraud. All match results voided; the match is treated as not having started.

Championship promotion markets: Markets for Blackburn Rovers, Birmingham City, Coventry City, Bristol City, Charlton Athletic, Derby County, Hull City, Ipswich Town, Leicester City, Middlesbrough, Millwall, Norwich City, Oxford United, Portsmouth, Preston North End, QPR, Sheffield United, Southampton, Stoke City, Swansea City, Watford, West Brom, and Wrexham have all been archived and resolved to "Yes"; net losing positions refunded.

UFC 331 — Tsarukyan vs Ruffy "Knockout" market: Bruce Buffer's remarks during the official referee decision do not qualify as a "Yes" resolution.

Tel Aviv temperature markets (Oct 20–22): Rules updated to use Weather Underground (LLBG station) as resolution source.

Weather market resolution rules (Oct 23+): METAR data will be retrieved from https://t.co/nnE6FEUiJQ. SPECI reports are excluded by community consensus. Concern raised that some traders receive METAR data faster than others, creating an informational edge. Community suggestions include adding 2–3 backup sources and removing WU as a fallback due to observed historical data edits.

Taker delay: Current crypto taker delay is 150ms; users are requesting a reduction to 100ms, and 0ms delay for sports markets.

With maker rewards slashed 90% and spreads widening across popular markets, is the taker fee holiday enough to keep retail traders engaged, or does Polymarket need to rethink its incentive structure entirely?

[Digest Polymarket & UMA Discord @predictdigest]

UMA DIGEST — 22 OCTOBER 2026

📋 Clarifications

A dispute arose over the Polymarket market "Will Trump speak to Friedrich Merz by October 31?" — a user (enormouz) was apparently baited into making a wrong dispute, with others pointing out the only confirmed call between Trump and Merz was on September 17th. A previous similar dispute resolved as early request (P4), which users noted may set a precedent that all "speak to" markets will default to P4.

A broader debate emerged around NBA and soccer halftime market proposals. Some proposers have been submitting results before the official resolution source (https://t.co/lIomlOp0lX) updates, or before the period of play has fully concluded. The relevant rule clause states: "This market may resolve only when the listed period of play has concluded and the result can be definitively determined by the listed resolution source(s)." Users noted this rule applies to halftime/set markets but apparently not to full-game totals, which im.the.wizard called contradictory. jessicaonlychild confirmed that if the score is confirmed, a halftime proposal is not P4, but acknowledged Polymarket's rules need updating.

A CS2 esports market dispute was raised: a user (verlin7469) argued that Map 1 of a match should resolve on its actual result since it was played to completion, while the forfeit only occurred on Map 2. Others explained the team was banned for match-fixing/cheating, and the tournament organizer voided all results, meaning UMA follows Polymarket's 50/50 resolution.

⚡ Disputes

A significant personal conflict escalated publicly between joes49572 and exquisite_wolf_79651 (cruz). joes49572 accused cruz of voting based on personal grudges rather than rules, citing two on-chain transactions (0xd2ced7bdb1fff315c781d45de3b8f9ddbcefafe51a2dd12b18015a196f20d2d5 and 0x742a5b0a05aded532182b3bbd4fa914a82d379ec0f7f1da1a91d6b5e1f35514e) as evidence of alleged improper conduct, and claimed cruz posted P4 on a table tennis dispute without any evidence at the time. joes49572 also alleged that https://t.co/XaqaTFIn9q posts selectively to swing votes. Cruz denied bias, stating he only posted on the 9 first-half markets where rules were being misread, and that he watched the TT match himself before posting. armanh01 noted that regardless of personal disputes, the actual NBA/soccer proposals should be judged on the rules and prior clarifications, not on who is disputing.

⚠️ Bugs & Issues

sam46 reported that the "last page" button on the UMA explorer (https://t.co/Qa5ozWfSjQ) has disappeared — only "first," "next," and "back" are visible. Separately, juicec11 reported being unable to complete a reveal via MetaMask mobile or PC, with console errors shown. infinity_d directed the user to open a support ticket, but juicec11 was unable to create one due to a verification issue, which was then addressed by infinity_d.

When a tournament organizer voids match results due to cheating, should UMA follow that decision even if individual maps were played fairly?

[Digest Polymarket & UMA Discord @predictdigest]

POLYMARKET DIGEST — 21 OCTOBER 2026

🔥 News & Announcements

Polymarket is running a fee holiday through October 31 (11:59:59 PM ET). Zero taker fees now apply to NBA, Politics, Economics, Weather, Finance, Culture, Mentions, and Tech markets — covering games, futures, and any new markets created during the window. Maker rebates are paused for these categories during the same period; normal fees resume November 1. pmkeaton noted that developers should update integrations to handle the zero_fees fee type instead of the usual category-specific ones.

Weather markets switching to METAR data source starting October 23. ak504111 confirmed that all new markets (US and non-US) will resolve off Aviation Weather Center METAR data, with each market linking to its specific feed and a document explaining how to read it. Hong Kong remains on the Hong Kong Observatory. SPECI reports (special observations) will not count — only routine METAR readings apply. Some traders noted this gives very short notice (roughly two days) for market participants to adapt.

⚠️ Bugs & Issues

NOAA/NWS outage is affecting non-US weather station data. Under current fallback rules, if NOAA data is unavailable by 11:59 PM ET on the day following the observation date, Weather Underground (WU) is used. However, WU's Istanbul (LTFM) station is also returning errors ("location invalid"), meaning Istanbul markets may fall to the "no data → lowest bracket" resolution clause. ak504111 confirmed: if the fallback source is also down, the no-data provision applies.

Elevated CLOB API rate limiting (HTTP 429) on POST endpoints is being reported by multiple developers. berth6833 and jungle_jo reported 2–3× the usual number of 429 errors. skulldoom97 acknowledged the reports and is investigating, but stated nothing appears changed on Polymarket's end so far.

Missing taker rebates were reported earlier in the day. kurwamach. flagged the issue to the team and confirmed rebates were subsequently sent.

Fee holiday side effect on weather liquidity: apples147 raised concern that pausing maker rebates alongside zero taker fees removes the main incentive for LP farmers who provide liquidity in weather markets, potentially widening spreads. The intraday LP reward appears to have dropped (from ~200 to ~100) coinciding with the fee holiday announcement.

📋 Clarifications

"Other Mommy" 2nd Weekend Box Office markets (all six brackets: <10m, 10–12m, 12–14m, 14–16m, 16–18m, ≥18m) have been archived and will resolve to "No". Net losing positions held at the time of the clarification will be refunded.

Weather NOAA fallback rules clarified by ak504111:
— Primary source: NWS/weather.gov
— Fallback: Weather Underground Daily Observations table (if NWS data unavailable by 11:59 PM ET the following day)
— No-data clause: if both sources are unavailable, markets resolve to the lowest bracket
— New markets from October 23 onward use Aviation Weather Center METAR; SPECI reports excluded; COR (corrected) readings replace prior values and are valid

With the fee holiday removing maker rebates on weather markets, could lower liquidity and wider spreads actually make prediction markets less useful for the traders they're trying to attract?

[Digest Polymarket & UMA Discord @predictdigest]

UMA DIGEST — 21 OCTOBER 2026

⚠️ Bugs & Issues

Toronto weather markets are broken — only US airports are currently supported. US https://t.co/rXv8V4e9Ds data sources are working but occasionally rate-limited, causing slow loads. juicec11 reported being unable to reveal proposals despite successfully committing all 141, even after clearing cache and history.

📋 Clarifications

A dispute is ongoing over whether halftime proposals for FK Polissia vs FK Dynamo Kyiv were valid. minorholiday argued halftime proposals were consistently submitted and permitted under the rules. im.the.wizard countered that a more recent clarification reversed the earlier one, with the sports expert reportedly unaware the original clarification had been issued. The core question raised: if a match is abandoned after halftime (e.g., at 80'), how should halftime markets be resolved given the score was already settled at HT?

Separately, a question arose around unplayed Dota games in a series: itibou initially argued unplayed games cannot be proposed until the Dota source confirms the series is complete, drawing an analogy to a prior P4 ruling. syrupfund disagreed, noting that prior case was P4 due to a split game, not an incomplete series record. mebeim clarified that unplayed games can generally be proposed as soon as the last played game can be proposed.

Regarding Tajon Buchanan shot stats in a UEFA Champions League market, .mlem123 noted Buchanan showed no shots in the official UEFA PDF but appeared with one on Flashscore. im.the.wizard investigated and confirmed Buchanan had 1 attempt blocked (AB), which likely accounts for the discrepancy in third-party sources.

⚡ Disputes

The "Russia targets Kyiv" market was disputed again according to enormouz.

A heated personal conflict broke out between whoooo8888 and wzht, involving accusations of front-running, insults, and deleted messages. infinity_d issued whoooo8888 a final warning. rex416. alleged that users amoyzjmm and another account are alts of the same person posting in dispute threads.

When a match gets abandoned mid-way through, should halftime markets that were already decided stay settled, or should they be voided since the full match never completed?

[Digest Polymarket & UMA Discord @predictdigest]

POLYMARKET DIGEST — 20 OCTOBER 2026

⚠️ Bugs & Issues

A National Weather Service (NOAA) outage is affecting non-US weather stations. braintr identified a likely cause: a change in the https://t.co/rXv8V4e9Ds source code around 2026-10-19 21:30 UTC added a condition restricting results to US stations only, potentially permanent. 0xbel_ announced that per market rules, if NOAA data remains unavailable by 11:59 PM ET the day after the observation date, Weather Underground Daily Observations will be used as the fallback source. Polymarket issued clarifications to dozens of European temperature markets (Milan, Munich, Paris, Madrid, Amsterdam, Warsaw) accordingly. alexchunus flagged outstanding UMA proposals on the temperature markets and warned against premature proposals; ak504111 confirmed clarifications are out on all disputed markets. The question of whether NOAA's API still returning data (via Synoptic) counts as "data available" remains open.

Additionally, WebSocket connections were reported dropping every few seconds by apples147, and satatardy reported intermittent internal server errors when sending orders.

The Discord invite link on the PolymarketUS docs page (https://t.co/HPMvbwHil5) is expired, per elementace.

🔥 News & Announcements

Combined taker tier is still delayed. pmkeaton confirmed it is "supposedly coming early this week" and apologized for the delay, inviting users to bump him on it. Users including jungle_jo and jxmcd expressed frustration over the wait and alleged favoritism toward large accounts with Obsidian tier. berth6833 criticized wallet aggregation as undermining the newly implemented rate limits, allowing multi-account users to bypass them. pmkeaton said he would ask the team for a comment.

Esports taker delay was flagged as abnormally high by gmdavid12; pmkeaton confirmed he would flag this to the esports team.

📋 Clarifications

NFL "last winless team" markets (Miami Dolphins and Los Angeles Chargers) had their tiebreaker rules updated for clarity: ties go first by strength of schedule, then alphabetically by full team name; the "Other" option is excluded from tiebreaks.

The Filip Zorvan Anytime Goalscorer market was clarified: a free kick in the 84th minute that deflected off a substitute is credited to Zorvan by the official resolution source (https://t.co/lL2zQ84OUl), so the market resolves in his favor.

If NOAA's API is still technically returning data through Synoptic, should that count as "available" for resolving these European temperature markets, or does it need to be directly accessible?

[Digest Polymarket & UMA Discord @predictdigest]

UMA DIGEST — 20 OCTOBER 2026

🔥 News & Announcements

Polymarket issued a clarification about a National Weather Service outage affecting non-US weather stations. Per the rules, if NOAA data for the observation date is unavailable by 11:59 PM ET the day after observation, Weather Underground Daily Observations will be used as the fallback resolution source. alexchunus noted this announcement appears to signal that affected weather markets (Paris, Warsaw, Milan, São Paulo, Cape Town, etc.) should be disputed, as NOAA data remains unavailable for all foreign cities. user_v123 remarked on what a mess the situation has become.

⚠️ Bugs & Issues

arkhimedean reported that https://t.co/0y3aN3OrLB is not showing vote reveal progress, occurring on both Chrome (mobile) and Firefox (desktop). kirbs86 reported a separate issue where https://t.co/0y3aN3OrLB fails to commit all 141/141 votes — commits work up to ~119 but fail beyond that, persisting across devices and after clearing cache and restarting.

📋 Clarifications

Regarding soccer markets, alexchunus clarified that the official website's information is treated as reality for resolution purposes. For the Adam Idah assist question, this means if the UEFA Nations League site does not record it, it resolves NO — though users were warned that if different parts of the same site show conflicting data, resolution may be unpredictable.

On the Whalefall box office market, https://t.co/xfEGrAnvAU shows $7.8M while Box Office Mojo estimates $8M. pojiladevchka flagged this as a potential P4 (unresolvable/ambiguous) situation per the market rule requiring both sources to confirm finalized figures before resolution.

For Dota 2 unplayed games, itibou raised the question of whether proposers must wait for the series page to officially confirm series completion, noting that many proposals are submitted ~5 seconds after the decisive game result, before the series page confirms. No definitive ruling was posted.

⚡ Disputes

Weather market mass dispute (major event): A large wave of disputes was triggered after proposers submitted non-US temperature markets (Paris, Warsaw, Milan, São Paulo, Cape Town, and others) despite the NOAA/weather.gov outage. ab7x7x provided proof that the https://t.co/rXv8V4e9Ds JS file blocking non-US cities was updated at 21:29:41 UTC on Oct 19, with his bot detecting the change at 21:32:09 UTC, meaning proposals submitted after that point used data from a broken/unavailable source. This is being characterized as proposers "front-running" the official resolution source. As of early Oct 20, im.the.wizard counted 251 open disputes, with ~68 from this weather event alone (plus ~80 more including the Cape Town batch from the prior day). Proposers who continued submitting weather markets even after the outage banner appeared on the resolution page are facing additional disputes. alexchunus noted that Hong Kong weather markets appear unaffected as they use a different resolution page.

Cleveland Guardians World Series market: Someone proposed "Cleveland wins World Series - NO" while Game 6 was still being played. laurleo and alexchunus flagged this as premature. jessicaonlychild noted the Guardians had won Game 1. The proposal is expected to be disputed and lose.

Saudi military action markets: raskell1 described a recurring issue where ~10 markets from September remain unresolved, caused by proposers asking AI whether Saudi strikes occurred, with AI returning "Yes" — without accounting for the distinction between Yemeni Air Force strikes and direct Saudi military action, which requires a higher evidentiary bar.

If Weather Underground data has to fill in for NOAA across dozens of markets, how confident are you that resolution will actually match what the proposers originally used to justify their answers?

[Digest Polymarket & UMA Discord @predictdigest]

POLYMARKET DIGEST — 19 OCTOBER 2026

📋 Clarifications

A dispute has been raised across the entire suite of Cape Town highest-temperature markets for October 18 (covering thresholds from 9°C or below up to 19°C or higher). Polymarket confirmed it was aware of the dispute and initially scheduled a potential clarification for 7:15 PM ET on October 18, with order books to be cleared at that time regardless.

Following that window, Polymarket issued a formal ruling invoking the Erroneous Data provision: the 19°C reading recorded at Cape Town International (FACT) at 07:28 SAST (05:28 UTC) on October 18, 2026 will not qualify toward resolution of any of these markets. The affected markets — including the 16°C, 17°C, 18°C, and 19°C-or-higher variants — will remain open for 7 calendar days until October 25, 2026, 1:28 AM ET, to allow the resolution source to publish a correction. If a corrected reading is issued by then, it will qualify; if not, the erroneous 19°C reading will still be excluded and markets will resolve on the basis of remaining available data. VIP user hans32312 called it a good clarification.

User abhiz69 suggested that in such situations Polymarket should clear the book and clarify immediately rather than waiting, noting that two markets had already settled during the liveness period and the rest could have settled too.

⚠️ Bugs & Issues

pastorlul flagged a potential API bug: for esports markets, the line field in game/map handicap API responses appears to be inverted relative to other sports. A -1.5 line for a home-spread in soccer means home handicap -1.5, but in esports it appears to mean home +1.5. No official response was received.

Several developers discussed Cloudflare bypassing on sports markets. .alexzor and blunkoed speculated that 1–3 large firms may have been whitelisted, though this is unverified. jungle_jo raised frustration that achieving Platinum taker tier on a single account has become extremely difficult, while certain accounts continue to hold Obsidian tier. satatardy argued the taker tier system should be removed entirely in favor of reduced fees for all users.

If Polymarket had immediately cleared the books and clarified instead of waiting, would the outcome have been fairer to traders who got caught in the settlement window?

[Digest Polymarket & UMA Discord @predictdigest]

UMA DIGEST — 19 OCTOBER 2026

⚡ Disputes

Mass dispute: Dynamo Kyiv vs. Polissia Zhytomyr (~98–109 proposals)

A proposer known as "Flame" submitted roughly 100+ proposals for the Ukrainian Premier League match between FK Polissia and FK Dynamo Kyiv while the match was suspended mid-game due to an air raid alert (the alert triggered at ~90th minute, with ~2 minutes of stoppage time remaining). im.the.wizard noted he had previously disputed Flame at least 4 times for similar behavior but Flame continued proposing. The match was confirmed suspended via live stream (https://t.co/nbWCywkUFj), official UPL sources, Dynamo Kyiv's Telegram channel, and Ukrainian sports sites. The game eventually resumed and concluded, making all full-time and in-play proposals P4. im.the.wizard confirmed he obtained video of the match still in progress during the proposal window.

Disputed count: ~98 full-time proposals at 500 bond each (~$49,000 lost by proposer). user_v123 put the total at 109 disputes / $54,500. Multiple disputors split the rewards relatively evenly (largest single share was 19 disputes).

First-half O/U proposals: A handful of first-half proposals were also disputed by some participants, though abhiz69 and im.the.wizard noted these were not P4 per an existing clarification (halftime proposals are valid once the first half is complete). im.the.wizard called the halftime clarification "haphazardly made" and expects it to be revisited. exquisite_wolf_79651 noted a precedent where first-half disputes were previously overturned via clarification.

Cape Town max temperature market — erroneous data dispute

A weather reading of 19°C was recorded at Cape Town International Airport (FACT) at 05:28 UTC on October 18, a jump of ~6°C above surrounding readings and inconsistent with all nearby METAR/SPECI data. hans32312 and modali1 provided evidence that the 19°C SPECI report assigned to FACT was actually a duplicate of a reading from FABL (Bram Fischer Airport, ~900 km away), including identical wind, temperature, and thunderstorm conditions — statistically near-impossible for two separate airports. alexchunus argued all brackets should be held open under the market's erroneous-data rule (up to 7-day wait), while some brackets had already settled before action was taken.

user_v123 posted the official resolution clarification:
Per the Erroneous Data provision, the 19°C reading for Cape Town International (FACT) at 07:28 SAST on October 18, 2026 will not qualify toward this market's resolution. The market will remain open for 7 calendar days until October 25, 2026, 1:28 AM ET, awaiting a correction. If no correction is published, the market resolves on available data excluding the 19°C reading.

The 19°C "Yes" bracket is effectively dead. If no corrected reading is issued, 16°C is considered the likely true daily high. alexchunus noted some brackets had already misresolved before the clarification was issued.

📋 Clarifications

In-play soccer proposals (general reminder): Proposing soccer markets while a match is still in play has never been permitted. laurleo asked if this was a new rule; abhiz69 and jarmax confirmed it has always applied to soccer specifically.

China GDP market: aterex163 flagged that the China Q3 GDP figure released on October 19 uses the phrase "preliminary estimates" rather than the exact title "Preliminary Accounting Results of GDP" specified in the market rules (scheduled for October 20). armanh01 and jamestalaricof argued these are the same release and China always uses that language. No dispute was filed.

Iwaki FC vs. Ventforet Kōfu (March 8 market): dyanreel raised that some props resolved 50/50 and the moneyline was proposed "Draw," but the first match played after March 8 was actually on March 14 (0–1 Iwaki). No resolution reached in the channel.

With over 100 proposals from a single actor during a live air raid, does this change how you think UMA should handle real-time geopolitical disruptions to sports markets?

[Digest Polymarket & UMA Discord @predictdigest]

POLYMARKET DIGEST — 18 OCTOBER 2026

📋 Clarifications

All 2026 Chicago Marathon winner markets (both Men's and Women's) have had their resolution rules updated. Due to missing official data from the Chicago Marathon results website, Polymarket confirmed that all related markets will resolve based on a consensus of credible reporting rather than the official results page. This affects markets for all listed runners including Brigid Kosgei, Hawi Feysa, Jacob Kiplimo, Amos Kipruto, and many others.

⚠️ Bugs & Issues

WebSocket order book sync issue: clericode reported a potentially significant bug where the WS order book snapshot on subscribe is sometimes stale by ~1 second on average, and occasionally by several minutes. In some cases, subsequent price_change messages fail to cover the gap, leaving the book in a permanently incorrect state. REST /book appears consistently fresh. Open questions include the correct resync strategy, whether a sequence number or gap signal exists, the exact hash formula, and what the level field (1/2/3) does in subscriptions.

CLOB latency under load: _valentyn_ raised a concern that spamming the CLOB during peak hours could push maker cancel latency beyond the taker delay, potentially breaking order priority. blunkoed noted the matching is "mostly FIFO with some quirks" and that past exploits around taker delay bypass have existed, though no known active one remains. llambre and others confirmed they have observed post/cancel latency spikes correlating with volatility.

Outdated API docs: blunkoed flagged that the official docs still incorrectly state a 250ms taker delay for crypto markets, which is no longer accurate.

Dublin API geo-block myth: blunkoed clarified that the Polymarket API was never geo-blocked for Dublin — only the frontend was. cryptowalkure added that moving a VPS from Dublin to London (AWS) yields no significant advantage since traffic routes through Cloudflare regardless.

Archived market refund: jenia9419 reported not receiving a refund two weeks after their market (will-yashar-win-the-second-most-seats-israel-2026) was archived.

How much should traders factor in the potential for stale order book data when building strategies on Polymarket's WebSocket feed?

[Digest Polymarket & UMA Discord @predictdigest]

UMA DIGEST — 18 OCTOBER 2026

���� Clarifications

A Polymarket market asking "Will the White House call a full lid by 6:30 PM on October 17?" was proposed early and disputed to P4. The consensus is that such markets must wait until one of the resolution sources (Roll Call or https://t.co/P88OFGhb3I) actually publishes a lid time — even if the answer is ultimately No. 7h0nk3r clarified: "Should just wait until the first lid for the day is published (or the next day's lid for the rare instances in which no lid is published for a day)." sam46 noted this matches prior precedent, and the market was successfully disputed by nefffooor.

⚡ Disputes

rex416 raised a question about an unresolved case, asking infinity_d (Polymarket) to explain how a disputed market differed from the Andy Burnham precedent, stating they lost the dispute but couldn't understand the distinction. infinity_d noted the resolution source was clear in the Polymarket UI — the player in question had no assists — and that the dispute thread had been empty, prompting them to post their opinion after the vote rolled. infinity_d also clarified they do not vote and hold no UMA staked, though nefffooor noted that posting a Polymarket opinion carries significant weight regardless.

Separately, laurleo flagged what appear to be several questionable proposals on sports markets (an O/U 4.5 dispute and a spread P4 for a game), commenting these are generating easy dispute profits and telling the proposer to "cut it out."

How should UMA handle disputes on markets where the resolution source is technically clear but the precedent for applying it seems inconsistent with past cases?