"[T]hey set the expiration at 10am... Meanwhile polymarket was 11:59pm." That is a complaint that surfaced in our research on cross-platform settlement, and it points at the most expensive fact in prediction markets: two venues can list what looks like the same question and settle it as two different questions. Who gets paid is decided by the contract, and Kalshi and Polymarket write different contracts. During the government funding fights of the past year, the two platforms diverged on all three of the levers that decide settlement: the source they trust, the deadline they read it at, and the definition they apply to it. Each divergence moved real money. One of them turned a contract trading at 97 cents into a penny at the stroke of midnight. We will get to that one, because it is the cleanest settlement lesson either platform has ever produced.
The Quick Answer
Kalshi and Polymarket use different rulebooks, so the same real-world event can settle Yes on one platform and No on the other. The three differences that decide winners are the settlement source (Kalshi binds each contract to a named Source Agency; Polymarket's criteria often allow "a consensus of credible reporting"), the deadline (Kalshi's expirations frequently land in the morning ET, while Polymarket's typically run to 11:59 PM ET), and the definition of the event itself. The shutdown markets that proved all three, a midnight collapse from 97 cents, and a checklist for trading the same event on both platforms are below.
The Three Levers That Decide Settlement
Every event contract is a written proposition, and the price is an implied probability of that proposition, not of the headline above it. When we compared the two platforms head to head, the differences were mostly about access and fees. The settlement differences are sharper, because they can produce opposite payouts from one event. Here is how the two rulebooks split on the government shutdown question, with the rules text pulled from each platform's own published terms as of August 12, 2026.
| Lever | Kalshi (SHUTDOWN contract terms) | Polymarket (market resolution text) |
|---|---|---|
| Source | "The Underlying for this Contract is notices on the website of the U.S. Office of Personnel Management." | OPM's Operating Status page on some markets; on others, "official information from the United States government... however, a consensus of credible reporting may also be used." |
| Deadline | Morning ET expirations are common; the SHUTDOWN terms set the initial iteration at 11:00 AM ET, and Kalshi's live funding markets close at 10:00 a.m. ET | Typically "by [date], 11:59 PM ET," with some markets cut at 12:00 AM ET on the date itself |
| Definition | A notice that the government "is at least partially shut down on [date] due to a lapse in appropriations" | Varies by market: one live market requires agencies actually suspending operations, a sibling market pays on any lapse "regardless of whether it results in any operational impact" |
The row most traders never price is the deadline row, so start there. The complaint at the top of this article is a deadline complaint, and the platforms' own documents show it is describing something real.
The Deadline Gap: A Morning Clock Against A Midnight Clock
Kalshi's SHUTDOWN contract terms define the market by a snapshot: the Expiration Value is what the Office of Personnel Management's website says "on the Expiration Date at the Expiration time," and the terms set the initial expiration at 11:00 AM ET. Its live government-funding markets, which ask whether appropriations legislation "become[s] law" before a series of dates, close at 10:00 a.m. ET on those dates per Kalshi's own market data. Polymarket wrote the January 2026 version of the shutdown question the other way: "This market will resolve to 'Yes' if the U.S. Office of Personnel Management (OPM) announces another federal government shutdown due to a lapse in appropriations by January 31, 2026, 11:59 PM ET."
Same event, same source, and, going by the contract's standard initial-iteration terms, settlement windows that differ by roughly thirteen hours. Anything OPM posts in the afternoon lands inside Polymarket's window and outside a morning Kalshi snapshot for that same date. A trader who buys Yes on both platforms is not holding the same position twice; those are two different propositions that happen to share a headline, which is exactly the gap the complaint above is describing. Coverage of the January fight recorded the pricing side: as the deadline approached, Polymarket's contract ran from 40% to 88% and Kalshi's from 44% to 93% inside 24 hours. Read those two price paths against the platforms' rules texts and part of the gap between them is simply a gap between two clocks.
The direction of the risk is not one-sided, either. Kalshi's snapshot rule contains its own trap: "Should no data be available at the Source Agency on the Expiration Time at the Expiration Date, then the Contract would resolve to No," and "Revisions to the Underlying made after Expiration will not be accounted for." A late notice, a delayed page update, a correction posted an hour after the snapshot, none of it counts. On Polymarket, the equivalent trap is the midnight boundary, and in November 2025 it went off on a market with about $13 million in volume behind it.
The Midnight Lesson: 97 Cents To A Penny
Here is the payoff of the story promised in the opening. During the 43-day shutdown that ended in November 2025, Polymarket listed a ladder asking what day the shutdown would end. The resolution text was precise: the market resolves to "the calendar date (ET) of the first day" on which OPM "announces that the U.S. federal government is not shut down," and it would do so "even if OPM lists a later reopening date." Not the day the Senate voted. Not the day the president signed. The day the source announced.
On November 12, the funding bill was signed and the shutdown was, in the day's news coverage, over. Traders piled into "ends November 12" Yes, which reportedly peaked at 97 cents, an implied 97%, while OPM's Operating Status page still showed a lapse in appropriations. Polymarket put a banner on the market telling traders to read the rules. By 11 PM ET the price had slid to 72 cents, still an implied 72% on a proposition the source had not yet satisfied. At midnight, November 12 could no longer be the announcement date, and the contract collapsed to about a penny. OPM updated its page on November 13, and the November 13 rung, which had traded far lighter, paid $1.00. Roughly $13.1 million changed hands on the November 12 market alone per Polymarket's market data, against about $6 million on November 13 and roughly $30 million across the ladder over four days.
Burned traders alleged the outcome defied reality, and that frustration is understandable: the government they were trading on really did get funded on the 12th. But the published criteria asked a narrower question, and the settlement followed the text. That is the recurring shape of prediction market settlement disputes on both platforms: nearly every one of these fights comes down to which document was asking the question, rather than what happened in the world.
Now notice the cross-platform version of the same night. A Kalshi shutdown-on-a-date contract reading its source at a morning snapshot and a Polymarket end-date ladder reading the announcement date at midnight can both settle correctly, off the same OPM page, and still pay opposite stories about when the shutdown ended. Both settlements are defensible from their own texts, because the two venues asked different questions.
Same Source On Paper, Different Sources In Practice
The OPM page keeps showing up in this article for a reason: it is the cleanest case study, because when both platforms name the same source, every remaining divergence has to come from deadline and definition. Most market pairs are not that clean. Kalshi's structure binds each contract family to a named Source Agency in its contract terms, and its rulebook gives the exchange "sole discretion to interpret a Contract's Terms and Conditions." Polymarket's criteria frequently name a primary source and then widen: its live appropriations market lists Congress.gov and official government information first, "however, a consensus of credible reporting may also be used."
Each approach fails differently, and I mean fails in the mechanical sense, not the accusatory one. A single named source is predictable but brittle: if the source is late, silent, or ambiguous at the reading time, the contract settles off whatever it shows, which is how a snapshot rule turns into a No. A credible-reporting standard is flexible but interpretive: someone has to decide when reporting has reached consensus, and traders on the wrong side of that judgment tend to allege the judgment was wrong. Kalshi's CFTC-filed rulebook at least publishes its escalation path in advance, and we will come back to it, because knowing who breaks ties is part of knowing what you own.
Worked Example: One October Deadline, Three Different Questions
Deadline and source divergences take a dramatic night to show up. Definition divergences are visible on any ordinary day, and as of August 12, 2026 you can watch one on Polymarket's own board, no second platform required. Two open markets point at the same October 1 funding deadline. "Government shutdown by October 1?" requires the real thing: a lapse in appropriations "that results in federal government agencies suspending non-excepted operations," and it says plainly that a lapse with no operational impact "will not qualify." Its sibling, "Federal Appropriations Lapse on October 1?", pays on the paperwork alone: any partial or full lapse qualifies "regardless of whether it results in any operational impact," measured at 12:00 AM ET on October 1.
The prices are where it gets instructive, because as of this writing they are arguably crossed against the wordings. The shutdown market's Yes traded around 16.5 cents, an implied 16.5%, while the lapse market sat near 14.5 cents, an implied 14.5%, small books both, with roughly $8,000 traded between them. Walk the logic: the shutdown market demands everything the lapse market demands plus actual suspended operations, so the stricter question should not be priced above the looser one, and yet there it is, two cents higher. On books this thin, the honest read is that the crowd has not put the two texts side by side, and that incoherence is the whole lesson in miniature. A reader who has done the comparison knows something the prices have not absorbed yet. Kalshi, meanwhile, frames its live coverage of the same fight in the opposite direction, asking whether funding legislation becomes law before September and October dates, with those 10:00 a.m. ET closes. Three venues' worth of questions about one deadline, and not one of them is "will there be a shutdown, you know, in the normal sense."
The lesson generalizes past politics. Weather markets define a high temperature by one station's reading. Sports markets define a comeback by lead size at a specific cut. Mention markets define "says" by a qualifying broadcast. Whatever the topic, the resolution criteria are the market, and two platforms writing criteria independently will diverge often enough that you should assume divergence until you have read both.
Why Cross-Platform "Arbitrage" Needs A Rules Check First
This is where the trap gets expensive, because price gaps between the platforms look like the cleanest trade in the business. Buy Yes at 60 cents on one venue, buy No at 35 cents on the other, collect a dollar at settlement, and the 5-cent gap is yours. We are fans of the real version of this: as we broke down on our show Arbitrage Betting Explained, a negative hold means the prices themselves can lock in the margin regardless of which side wins, and it is the same math that powers prediction market arbitrage within a single venue.
Across venues, that math has an unstated assumption: both contracts settle on the same proposition. Every divergence in this article breaks the assumption. A morning snapshot against a midnight deadline means an afternoon OPM notice settles your Kalshi leg No and your Polymarket leg Yes, and on the wrong pairing both legs can lose. A definition gap does the same thing without any timing drama at all. The paired positions were never opposite sides of one question, so the "locked" outcome was never locked. Before any cross-platform position, put the two rules texts side by side, the way we do when comparing prices across exchanges, and run the three-lever check:
- Source: the same named source on both texts, and the same fallback if that source goes silent or posts late.
- Deadline: the same date AND the same reading time. A morning snapshot against a midnight window is a thirteen-hour hole in your hedge.
- Definition: the same qualifying event, with the same exclusions, at the same threshold. This is the event-contract version of a habit sharp sports bettors already have. Shopping the number only works because every book grades the same final score, which is exactly what makes the live odds screen comparison honest: shop the number across every major book, and the thing you are comparing is actually the same bet. On prediction markets, you have to verify that premise yourself.
Who Decides When Settlement Goes Sideways
The last difference worth knowing before you trade both platforms is who breaks ties, because eventually a market you hold will land in the gray zone. Kalshi's rulebook publishes the machinery. The exchange can send an outcome to its Outcome Review Committee before settlement, the committee "shall review all relevant evidence and determine a final Market Outcome within a 24 hour period," and its determinations "are final." If an outcome truly cannot be determined and the terms do not cover the contingency, the rulebook's fallback can pay both sides off the last traded price; the rulebook's own example is a contract that last traded at $0.10 paying longs 10 cents and shorts 90 cents, a 10%/90% split of the settlement dollar. Kalshi also reserves the right, under Rule 7.2, to designate a new Source Agency if the original becomes unreliable, announced on its website. What that machinery does after expiration is its own subject, and what happens at settlement walks through it clause by clause.
Polymarket's November response shows the other model in action: warn loudly, then settle on the text. The banner told traders to read the rules while the rules did exactly what they said they would. Neither model prevents disputes, and users on both platforms have alleged bad settlements after landing on the wrong side of one. What the documents buy you is predictability. If you know the source, the reading time, the definition, and the referee before you click buy, normal settlement mechanics stop feeling like ambushes and start being priceable.
Kalshi Vs Polymarket Rules FAQ
Do Kalshi and Polymarket use the same settlement rules for the same event? No. Each platform writes its own resolution criteria, and they routinely differ on the settlement source, the deadline, and the definition of the event. The January 2026 shutdown markets shared a source, OPM, and still differed on the clock: a morning ET expiration structure on Kalshi against an 11:59 PM ET cutoff on Polymarket.
Do prediction market contracts ever settle against the obvious real-world outcome? Yes, when the contract's question is narrower than the headline's. Polymarket's shutdown-end ladder resolved to November 13, the OPM announcement date, even though the funding bill was signed November 12, because the text keyed on the announcement. A contract trading at 97 cents, an implied 97%, went to about a penny at midnight on exactly that distinction.
Do I need to read the rules on both platforms before trading the same event on each? Yes, and side by side. Confirm the three levers match: source, deadline, definition. If any of the three differ, you hold two different propositions, and any hedge or arbitrage built on them can lose both legs.
The Event Is Not The Trade. The Documents Are.
Back to the complaint this article opened with, because it deserves a straight answer. The morning-against-midnight gap it describes is real and printed: Kalshi's shutdown terms and its live funding markets carry morning ET cutoffs in their published rules, against Polymarket's 11:59 PM ET windows, and neither venue hides those documents. That is the honest, slightly uncomfortable resolution to the whole subject: nearly every cross-platform settlement "scandal" we researched was published in advance. Two venues, two documents, two questions, and the trader who reads both holds an edge over the trader who reads neither, because rule gaps are mispricings you can spot before the event instead of bad beats you absorb after it. It is the same settlement-first discipline our own published calls are graded against on our public scoreboard, and it travels to conventional sports markets, with one welcome simplification: there, every book grades the same final score, so the rules check shrinks to house rules and the work shifts to price. Our free expert picks are built on exactly that division of labor. Wherever you trade, read every ticket the way a settlement committee will read it. Terms first, price second, headline last.
