Will China Invade Taiwan? The Odds, and What They Cover

Published 21 September 2026 · Prices frozen 21 September 2026, 16:10 UTC

Every page ranking for Taiwan invasion odds prints the same 4% and stops. One of the aggregators on the first page of results was still serving a market that expired in June; another has a number three weeks old sitting under a “live” label. That matters because $51.4m has traded across Polymarket’s four China-Taiwan conflict contracts, and the single number everyone quotes is the one that tells you least — it is one rung of a three-date ladder, and it resolves Yes on events a long way short of what readers picture. We read both exchanges in the same pass, pulled the full price history on every contract, swept all 11,189 open Kalshi events to see what the other venue lists, and did the conditional arithmetic between the rungs. The short version: the market is calm, it is calmer about 2027 than about this quarter, and the most interesting thing in it is a 1.30-point gap almost nobody prices.

Where the China-Taiwan invasion odds stand

Here is the whole complex in one place: three invasion horizons, the clash contract that shares a deadline with one of them, and the China-Japan contract written to the same template. Prices are raw last trades with the live quote beside them, because on the thinner rungs the spread is a material part of the story.

Two things jump out of that table before any analysis. The first is concentration: the end-of-2026 contract took 85.4% of the complex’s last-day turnover and carries 82.8% of its lifetime volume. When a page quotes “the Taiwan market”, that is the one it means, and the other rungs are priced by far fewer people. The second is that a China-Japan military clash is priced above a China-Taiwan one — 5.50¢ against 5.35¢ — on contracts with word-for-word identical definitions. Whatever the market thinks is most likely to go bang first, it is not the strait.

Three dates, three prices, one curve

Three contracts asking the same question with three deadlines is a term structure, and it carries more information than any single rung. Subtract each rung from the next and divide by what is left, and you get the market’s implied probability for each window on its own.

The right-hand column is the finding. The market prices the next three months at roughly 1.8 times the per-month invasion risk of any month in 2027. Geopolitical risk curves usually slope the other way — the further out you look, the more room there is for the world to change — so an inverted hazard curve is a specific claim that this quarter is unusually loaded.

Two readings compete, and honesty requires both. One is that the elevated near-term rate brackets a real event: Taiwan votes on 28 November, inside that window. The other is that the 2027 contract is simply the thinner book — $193,492 of depth against $578,013, and a full penny of spread against a tenth of one — and thin books drift low on far-dated tail risk because nobody wants capital parked for fifteen months. We lean towards the second explanation carrying most of the weight, but the first is not nothing, and the way to tell them apart is to watch whether the 2027 rung moves after 28 November.

The clause that decides this market

Every aggregator quotes the first sentence of the resolution criteria. Almost none quotes the second, which is where the money is.

Read that carefully and the contract is not what the headline suggests. Taipei administers Kinmen and Matsu, inhabited island groups that sit a few kilometres off the Chinese mainland and more than a hundred from Taiwan itself. An offensive intended to take either would resolve this market Yes. So would an offensive against any other inhabited islet under Taipei’s administration. The 4.05¢ is the price of the cheapest qualifying military action, not the amphibious assault on the main island that the phrase “invade Taiwan” puts in a reader’s head.

That cuts both ways, and it is worth being precise about which. It makes 4.05¢ look high if you were pricing a full invasion, because the contract pays out on scenarios far below that threshold. It makes the same 4.05¢ look low as a measure of the outright war risk readers actually care about, because the true probability of a main-island landing must be some fraction of it. Either way, a number quoted without this clause is being quoted wrongly — which is the same failure we found in the Russian Duma markets, where one verb in the rules was worth twenty points. Our methodology page sets out how we read resolution criteria before printing a price.

A clash is barely likelier than an invasion

Polymarket runs a separate contract on whether there will be a China-Taiwan military clash before 2027. It shares a deadline with the end-of-2026 invasion contract — both settle on 31 December 2026 — and an invasion would satisfy its definition too. So the difference between them is a clean price for one specific thing: violence that stops short of an offensive.

The spread has compressed from around 4.5 points in the spring to 1.30 points today, the tightest it has been since the clash contract’s opening week. Stated as a conditional: the market now says that if Chinese and Taiwanese forces exchange fire before 2027, there is a 75.7% chance it is part of an offensive to take territory rather than an isolated incident.

That is a strange thing to believe, and it is worth saying plainly that we do not think it reflects a considered view about escalation dynamics. The last twelve days did most of the compressing — the clash contract fell from 8.05¢ on 9 September to a 30-day low of 5.35¢ while the invasion contract ticked up to a 30-day high of 4.70¢ on 18 September. Two thinly traded lines moved in opposite directions on $3,089 of daily volume between them. The more useful explanation is in the next two sections: the clash contract is written so tightly that almost nothing currently happening in the strait can resolve it.

Kalshi does not list this market at all

The natural next move on this site is to put the Kalshi price beside the Polymarket one, as we did for the Venezuela leader market. Here that is impossible, and the reason is itself the finding. We paginated Kalshi’s open-events feed to cursor exhaustion — 11,189 events — and found exactly five that mention Taiwan.

Not one is a conflict market. Kalshi’s Taiwan shelf prices elections, a travel advisory, diplomatic recognition and — in the one contract that touches cross-strait relations at all — the date on which Xi Jinping will physically visit Taiwan. That is the peaceful end of the distribution, written as a question about a state visit rather than a landing.

And none of it trades. Sixteen contracts, zero volume, zero open interest, no resting quotes. The contrast is the point: the same question that has moved $51.4m on one venue has not moved a single contract on the other, and the venue that took the $51.4m is the one that wrote a conflict contract. A US-regulated exchange listing Taiwan elections but no Taiwan war, and finding no takers even for the elections, tells you something about where this audience is — and it means every “Polymarket vs Kalshi” comparison you see on this topic is comparing a price to nothing.

Fourteen months of steady de-escalation

None of today’s calm is new, and the shape of the decline matters more than its endpoint. The end-of-2026 contract has been falling almost without interruption since the day it listed.

A 39.45-point fall, 91% of the opening price, with the steepest leg in the first six weeks. The other three contracts tell the same story on their own timelines: the 2027 rung from 29.50¢ to 11.50¢ since March, the clash contract from 28.50¢ to 5.35¢ since November, and the 30 September contract from 12.50¢ to 0.45¢ with nine days left on the clock. Four independent books, one direction.

The analytical case behind the drift is not hard to find. Joe Keary, writing for ASPI’s The Strategist in January 2026, argued that the anti-corruption purges through the PLA’s senior ranks have degraded command confidence and readiness for complex joint operations, that China’s economic and demographic headwinds raise the cost of a war estimated in the trillions, and that Beijing will therefore keep to coercion rather than invasion. That is a reasonable description of what the tape has been doing for fourteen months.

What the grey zone does not count as

Here is the tension a reader should hold onto. Pressure in the strait has not stopped — coast guard patrols, air and sea incursions and blockade rehearsals continue — and yet the clash contract keeps falling. Both things are true, because of how the contract is written.

That exclusion list is, almost item for item, a description of routine cross-strait activity. The contract is not a grey-zone tracker. It is a bet on an exchange of fire between the two militaries, which is a much rarer event and a much higher bar. Once you see that, the 5.35¢ stops looking like complacency and starts looking like an accurate price for a narrowly defined thing — and the 1.30-point spread over the invasion contract becomes less a claim about escalation than an artefact of two definitions that overlap far more than their names suggest.

This is the general lesson of the page. On a market like the China-Japan clash contract, written from the identical template, the same exclusions apply and the same caution is owed. Daily moves across every market we track land on our 24-hour movers page.

What to watch: 30 September, then 28 November

Two dated events sit between now and the end of this complex, and one of them arrives in nine days.

30 September. The nearest rung expires at 0.45¢ and will almost certainly settle No. The useful part is not the settlement but what the other rungs do around it: if the end-of-2026 contract does not fall as that rung drops out, the market is quietly re-weighting its remaining risk into the final quarter, which is exactly the inverted hazard curve above expressing itself.

28 November. Taiwan holds its nine-in-one local elections, with 19.65 million eligible voters choosing 11,051 officials. It is the one scheduled political event inside the elevated window, and it is also the cleanest cross-exchange comparison this topic offers.

Polymarket has the Kuomintang at 87.5¢ to win the most mayoral and magistrate races, a near-settled view; you can follow it on the local elections market page. Whether a KMT sweep changes anything in the conflict contracts is the question worth holding — a more Beijing-friendly local map is the kind of thing that should compress invasion risk further, and if it does not, that tells you the invasion price was never about Taiwanese politics.

Related on GeoOdds

• Will China invade Taiwan by end of 2026 — live odds, chart and resolution criteria
• Will China invade Taiwan by 31 December 2027 — the long rung of the ladder
• China x Taiwan military clash before 2027 — live odds on the narrower contract
• 2026 Taiwanese local elections — live party-winner board before 28 November
• Polymarket vs Kalshi: Venezuela leader odds — the same cross-exchange method, where both books exist
• Nobel Peace Prize odds 2026 — why a book that does not sum to 100 has to be read raw

Sources: the Polymarket Gamma and CLOB public APIs (events 34044, 281143, 281145, 79230, 83798 and 96786) and the Kalshi public events and markets APIs, both read at 21 September 2026, 16:10 UTC; Joe Keary for ASPI’s The Strategist on the PLA purges and Beijing’s coercion toolkit; the Taipei Times and Focus Taiwan on the 28 November 2026 local elections. Resolution-criteria quotations are from the market rules as published on Gamma. Prices captured 21 September 2026, 16:10 UTC and not updated since. Nothing here is investment advice — see our disclaimer.