Polymarket End Date 2026: 88% Close Late
Polymarket end date timed on 16,924 traded markets: 88% stopped trading after it, a median 2h50m late, and all 813 tennis markets a week early.
Affiliate disclosure: contains affiliate links; we may earn a commission at no extra cost to you. We have no commercial relationship with Kalshi. 18+ only — see BeGambleAware.org. Last updated 27 October 2026.
Key takeaways
- A Polymarket end date is not a trading deadline: 88.30% of the time trading carries on past it. We timed 16,924 traded markets ending 21–25 October 2026, worth $595.6m. The median stopped 2h50m after its end date, a quarter more than four hours after.
- Only 48 of the 16,924 — 0.28% — closed within a minute of the date on the page, and one in five inside five minutes. The rest missed by hours, days, or in one case sixteen days.
- Tennis goes the other way: all 813 closed early, a median of 6d19h. Polymarket sets a tennis end date about a week after the match, so the countdown has nothing to do with when you can exit.
- Kalshi is the mirror image: across 62,394 settled markets, 90.92% stopped trading at or before the moment the answer was expected, a median of 5m00s before.
- The difference is structural, not sloppy. Kalshi publishes a trading deadline, a separate forecast of the answer, and a flag — set on 99.99% of markets — saying the deadline may move earlier. Polymarket publishes one date its docs define as resolution eligibility.
On this page
You are holding a position and the market page shows a date. The obvious reading is that the date is your deadline — trade until then, then wait for the money. It is not, and on the venue where most people learn the habit it is wrong by hours in the large majority of cases. We timed every traded market on both big prediction exchanges scheduled to end over four days in October 2026: 16,924 Polymarket markets worth $595.6m and 62,394 settled Kalshi markets. The Polymarket end date matched the actual close to within a minute on 48 of them. Here is what each date means, how far off it runs, and the field on each venue that answers the question properly.
What a Polymarket end date actually is
Start with the definition, because it is the whole problem. Polymarket's documentation defines the field in six words: the end date is "When the market is eligible for resolution." Eligible. Not closed, not settled, not paid. It is the moment the resolution machinery may begin, and nothing in that sentence commits the order book to shutting at the same instant.
The same page lists what happens after resolution, and the first item is "Trading stops." So trading stops at resolution, and resolution starts no earlier than the end date: the published date is a floor on the close, not the close itself. The docs never say how large the gap can be.
Kalshi splits the same information across separate fields, each with its own job — the reason the two venues behave so differently.
Field definitions quoted from each venue's own API documentation; the 99.99% is measured, not published.
Four fields against one. Kalshi's page can say trading ends at 7pm, that 7pm may move earlier, that the answer is expected at 10pm and that the outside limit is next March. Polymarket's single date stands in for all of it.
The end date is not when trading stops
We took every Polymarket market with an end date between 21 and 25 October 2026 that had traded at least $1,000 and compared that date with the closedTime the exchange records when the book shuts: 16,924 markets, 6,853 events, $595.6m. The result is not a near miss.
Polymarket: 16,924 traded markets with an end date 21–25 Oct 2026. Kalshi: 62,394 markets settled in the seven days to 27 Oct 2026, measured against expected_expiration_time. Read 27 October 2026.
The Polymarket column has its mass on the wrong side of zero. 88.30% stopped trading more than a minute after the date on the page, and the overshoot is not trivial: median 2h50m, 75th percentile 4h04m, 90th 5h05m. Three and a half percent ran more than a day past it, and the worst — a Counter-Strike match on the ESEA Advanced ladder — stayed open 6d14h beyond.
The early side is smaller but stranger. One market in sixteen shut more than a day before its end date, and the record — on whether the next Claude Haiku model would ship by a given October date — closed 16d07h early. Both tails have the same cause: the end date was never a statement about the order book. Kalshi's column is the other shape entirely.
How late the Polymarket end date runs, by market type
An average across a venue that lists tennis and Bitcoin in the same breath hides more than it shows. Tagging every market by its parent event — Polymarket publishes the tags itself — splits the 16,924 into groups that each behave consistently, and within a type the gap is reliable enough to plan around.
A negative median means the market closed before its published end date. The twelve named types with at least 40 markets; markets whose tags fall outside them are omitted. Volume is the exchange's own figure.
Three patterns, none of them noise. Every ball sport closes late by three to four hours — soccer 3h58m, NBA 2h55m, NHL 3h00m, NFL 3h24m, college football 4h05m — closing late in 100% of cases in four of those five. That is a game plus the wait for an official result: the end date sits near kick-off, and the book shuts when the thing is over.
Crypto is the tightest clock on the venue, suspiciously so: the 25th, 50th and 75th percentiles are the same number, 1m20s. That is not a distribution, it is a setting — the book shuts a fixed eighty seconds after the price window ends. If you trade only Bitcoin hourlies, the end date is close to honest.
Weather is the worst of the late group — median 18h20m past the end date, 90th percentile 1d17h — because a day's high temperature cannot be known until the day is over and the station publishes. And then there is tennis.
Every tennis market closed a week early
Of 813 tennis markets across 585 matches, 813 closed before their published end date. Not most. All of them. The median shut 6d19h ahead of the date on its page — the tightest-clustered result in the census: 6d21h early at the 10th percentile, 6d13h at the 90th, an eight-hour spread across a near-seven-day offset.
The mechanism is mundane and the consequence is not. Polymarket stamps a tennis market's end date roughly seven days after the match — a backstop wide enough to cover a rain delay, a retirement or a rescheduled court. The match then finishes in two hours, the market resolves, and the book shuts. So the countdown on the page tells you about the contingency window, and the thing you actually needed — when you can no longer sell — came and went six days earlier.
This is the finding most likely to cost someone money, because it inverts the intuition. On a soccer market a late close is a few extra hours of liquidity you did not expect. On a tennis market the close arrives a week before the deadline you were watching, and anyone planning to exit "before the end date" finds the book shut while still inside their own plan.
The 23-market study that is currently the best page on this subject put tennis at 6.3 to 6.8 days early; our 813-market median of 6d19h sits inside their range, which is a good sign for both. What 23 markets could not show, and 16,924 can, is that it holds for every tennis market without exception while the rest of the venue runs the other way.
Kalshi publishes two clocks, not one
Nobody writing about end dates has looked at Kalshi, which is odd, because Kalshi answers the question in its docs. Its market lifecycle page defines close_time as "When trading stops. May be moved earlier if can_close_early is true." That is what a position holder wants: a deadline, plus a warning that it is provisional.
The second clock is separate: expected_expiration_time is "When the outcome is expected to be known" — a forecast, not a commitment, with latest_expiration_time as the outside bound. Kalshi's help centre is blunter: "The market's displayed close time may not equal determination time. Trading can end before or after an event for operational reasons."
So the two venues are not doing the same thing badly — they are doing different things. Kalshi publishes the deadline as its own field and keeps it true by moving it; the flag saying it can move was set on 99.99% of the markets we measured. Polymarket publishes the forecast and leaves the deadline unstated, which is why the field people read cannot answer them.
The asymmetry in one line. On Kalshi, the date next to the countdown is the trading deadline, and it gets revised rather than broken. On Polymarket, the date next to the countdown is when resolution becomes possible, and trading routinely continues past it — which means the two venues' countdowns are not comparable quantities, however identical they look on screen.
That leaves one measurable question on the Kalshi side: how far apart are its two clocks? If trading stops well before the answer, you hold a position you cannot exit while the result is unknown — the same problem from the other direction.
Kalshi's gap, measured on 62,394 markets
We pulled every market that settled in the seven days to 27 October 2026 across 177 named series spanning every category Kalshi lists, recording close_time against expected_expiration_time: 62,394 markets, 4.61bn contracts. 90.92% stopped trading at or before the moment the answer was expected, a median of 5m00s beforehand, and 8.88% closed late. The category table is also the clearest evidence that these offsets are designed rather than emergent.
A negative median means trading stopped before the answer was expected, which is the normal case on Kalshi. Categories under 100 markets are omitted. Contracts are the exchange's own volume figures.
Look at how little those medians vary within a category. Crypto's is a flat five minutes at every percentile we checked, and several other groups sit on equally exact numbers. These are parameters somebody typed, not the residue of a process, and they are published per market before you trade.
Sport is the exception, for the same reason as on Polymarket: a game does not end on a schedule. Kalshi keeps revising the forecast as the game runs, which is why its sports markets show a gap in both directions, bunched around the final whistle, rather than a fixed offset.
"Trading continues right up to the end date" is wrong
One widely-ranking Polymarket guide states it plainly: "Trading continues right up to the end date." It is the natural reading of the field name, published with no number attached. Our census says it is wrong in both directions at once — not an overstatement, but a claim with no stable relationship to what happens.
Trading continued past the end date on 88.30% of 16,924 markets, by a median of 2h50m — so "right up to" understates it. And trading stopped before the end date on 11.41%, including every one of 813 tennis markets by nearly a week — so for those it overstates it badly. The sentence is accurate for the 0.28% that closed within a minute.
Being fair to the field: it is not documented as a trading deadline, and Polymarket does not claim it is. The official definition — "eligible for resolution" — is honest, narrow and easy to miss. The error is downstream, in every guide that translates it into a deadline because a deadline is what readers came for. The end date says roughly when the question is settled, give or take a day; for when the book shuts there is a different field, and checking it is two lines of work.
What the gap costs a position holder
A mistimed close decides whether you can cut a losing position, whether capital is free for the next trade, and whether the price you can get is worth having. Three real cases, each at the top of its type by volume:
The pattern worth internalising: a late close costs you nothing and occasionally helps, while an early close removes a choice you thought you had. The early tail sits in exactly the markets where the end date looks most generous — tennis, esports, and anything phrased "will X happen by date Y", where the answer can arrive the moment X happens. On Kalshi the risk is reversed and smaller: trading stops shortly before the answer, so you cannot trade the result, but you are rarely surprised about when.
Reading the real deadline in ten seconds
Neither venue hides this. Both answer in one unauthenticated request; the fields to read differ.
# Polymarket: is this book still open, and what is the end date really?
curl -s "https://gamma-api.polymarket.com/markets?slug=YOUR-MARKET-SLUG" | python3 -c "import sys,json
m=json.load(sys.stdin)[0]
print('trading now :', m['acceptingOrders'], '| closed:', m['closed'])
print('endDate :', m['endDate'])
print('gameStart :', m.get('gameStartTime'))
print('rules :', m['description'][:300])"
# Kalshi: the deadline itself, and whether it can move
curl -s "https://api.elections.kalshi.com/trade-api/v2/markets/YOUR_TICKER" | python3 -c "import sys,json
m=json.load(sys.stdin)['market']
print('trading until :', m['close_time'], '| can move earlier:', m['can_close_early'])
print('answer expected:', m['expected_expiration_time'])
print('outside limit :', m['latest_expiration_time'])"Three rules follow. Do not trust active on Polymarket — read closed and acceptingOrders. Read the rules text, because for "by date Y" questions the governing deadline is written there in prose and nowhere in a date field. And treat a sports end date as the start, not the finish: add three to four hours for a ball game, subtract a week for tennis.
On Kalshi there is one rule: the close time is the answer, and can_close_early says it may arrive sooner. How long after that your money comes back is a separate field, measured here across 15,382 markets.
Check the close before you size the trade
The date on the market page is the one number most traders never verify, and on one of these venues it is wrong 88% of the time. Two lines of curl tell you when the book actually shuts — before you need to get out, not after.
Visit Polymarket.com →Affiliate link to Polymarket; we have no commercial relationship with Kalshi. 18+ only. Polymarket restricts access from the United States, Canada and other listed jurisdictions. Prediction markets carry a substantial risk of loss. Nothing here is financial advice.
My verdict on the Polymarket end date
Treat it as a label on the question, not a deadline on your position. Across 16,924 traded markets it matched the actual close within a minute 48 times, and the two ways it misses are not symmetrical. A late close is free; an early close takes away an exit, and it happens most in the markets whose end dates look furthest away.
The comparison with Kalshi is not a scoreboard, because the venues publish different objects. But the design difference has a clear winner for anyone holding a position: four fields that each answer one question beat one field that answers none. Kalshi's close time is revised rather than broken, can_close_early warns you in advance, and 90.92% of the time trading stopped at or before the expected answer.
Polymarket's fix is not even a product change: publish the close, or stop letting a resolution-eligibility date sit where a deadline belongs. Until it does, the end date is a hint, the rules text is the contract, and acceptingOrders is the truth.
Related guides
- Polymarket Payout Time 2026: 37,659 Timed — the clock that starts where this one stops.
- Kalshi Payout Time 2026: 15,382 Timed — the same measurement on the exchange: a 2m52s median wait.
- Polymarket Resolution 2026: When UMA Decides — what happens between close and payout, and who can challenge it.
- Polymarket vs Kalshi 2026 — eligibility, deposits, fees and who settles a result.
- Polymarket Slippage 2026: 897 Books Measured — what the book looks like in those extra hours.
- Kalshi Fees by Market 2026: 195 Exceptions — the other per-market field nobody reads.
Polymarket end date FAQ
What does the end date mean on Polymarket?
Polymarket's documentation defines it as "when the market is eligible for resolution" — not a trading deadline, and not the settlement time. Across our 16,924 traded markets, trading stopped more than a minute after the end date 88.30% of the time, by a median of 2h50m.
Can you trade a Polymarket market after its end date?
Usually yes: 88.30% of markets ending 21–25 October 2026 were still open past their end date, 64.41% by more than an hour. The authoritative check is acceptingOrders on the Gamma API — not the date, and not active, which can read true on a closed market.
Can a Polymarket market close before its end date?
Yes, and on some types it always does. 11.41% of our sample closed early, 6.14% by more than a day. All 813 tennis markets closed early, a median of 6d19h, because Polymarket sets a tennis end date about a week after the match. Esports closed early on 90.0% of markets.
How accurate is the Polymarket countdown?
As a prediction of when trading stops, barely: 48 of 16,924 markets — 0.28% — closed within a minute of their end date, 20.38% within five minutes. Short-dated crypto is the exception, closing a fixed 1m20s after the end date at the 25th, 50th and 75th percentiles alike.
Why do Polymarket sports markets stay open after the game?
Because the end date sits near kick-off, not the final whistle, and the book shuts once the result is confirmed. Soccer ran a median 3h58m past its end date, college football 4h05m, NFL 3h24m, NBA 2h55m and NHL 3h00m, closing late in 100% of cases in four of those five.
Does Kalshi close markets early too?
It reserves the right to on almost everything: can_close_early was true on 99.99% of the 62,394 markets we measured. Trading stopped at or before the expected answer 90.92% of the time, a median of 5m00s early.
What is the difference between a Polymarket end date and the resolution date?
The end date makes the market eligible for resolution; resolution is when the outcome is proposed to the UMA oracle and, after an undisputed two-hour challenge window, becomes final. Trading stops at resolution, not at the end date — which is why the two drift apart.
How do I find out when a Polymarket market really closes?
Read three things rather than one: acceptingOrders and closed for the live state, gameStartTime for sports, and the market's description, where any contractual deadline is written in prose. For "by date Y" questions the description is the only place it appears.
Method and sources
Read on 27 October 2026. Polymarket: we queried the public Gamma API for every closed market with an endDate between 21 and 25 October 2026 and at least $1,000 of volume, walking the window in 30-minute slices and subdividing any that hit the API's 2,100-row paging ceiling — 16,924 markets, 6,853 events, $595.6m, no slice truncated. The gap is closedTime minus endDate. Market types come from the tags Polymarket publishes on each parent event, fetched for all 6,853. The $1,000 floor matters: without it the population is dominated by machine-generated books that never traded. Kalshi: all 15,423 series arrive in one request, from which we took 177 named series covering every category and queried each one's markets settled in the seven days to 27 October — 62,394 markets, 4.61bn contracts; the gap is close_time minus expected_expiration_time.
Three caveats. The two measurements compare each venue's actual close against the date it publishes, and those dates are defined differently, so the contrast is between what each venue tells you rather than between identical quantities. A market still open when we read it cannot appear, understating the latest closers. And the Kalshi series set is a named sample, not the whole exchange. Nothing here is financial advice.
- Polymarket docs: Resolution — "End date — When the market is eligible for resolution", "Trading stops" listed under what happens after resolution, and the two-hour UMA challenge window.
- Kalshi docs: market lifecycle — the four time fields quoted above, and
expiration_timemarked deprecated. - Kalshi help centre: market FAQs — "displayed close time may not equal determination time", and the instruction to check each market's Rules section.
- OrcaLayer: "Polymarket Gamma API endDate Is Not the Resolution Deadline" (24 Sep 2026) — the only prior study we found, on 23 hand-picked markets. Its tennis range, its esports range and its advice to read
closedandacceptingOrdersrather thanactiveall hold at our scale. - StartPolymarket: how markets resolve — the source of the "Trading continues right up to the end date" claim this guide corrects.