Every September, a small group of glaciologists, climate modellers and satellite analysts wait for the same number. It is the minimum extent of Arctic sea ice, measured at the end of the melt season by the National Snow and Ice Data Center in Colorado, and it has become one of the most-watched indicators in climate science. The record is held by September 2012, when the ice shrank to around 3.39 million square kilometres. Nothing since has come close.
Now Polymarket is asking traders to price the same question the scientists are. The Arctic sea ice extent market on Polymarket resolves on the minimum daily extent reported by NSIDC between 1 August and 1 October 2026, with the resolution source measured to the thousand square kilometres. That is the contract. It is one question, sliced into seven bands of possible outcomes, and the resolution rule is unusually clean for a climate market.
What the contract actually measures
Arctic sea ice extent is not a vibe. It is a specific satellite-derived measurement of the area of ocean covered by at least 15% ice, averaged daily and published by NSIDC. The September minimum is the lowest single value reached during the melt season, usually somewhere between the second week of September and the start of October. The contract takes that number and asks which of seven bands it lands in: below 4 million square kilometres, between 4.0 and 4.2, between 4.2 and 4.4, and so on up to 5 million and above.
For context, the 2012 record sits well inside that lowest band. The 2007 minimum, the previous record before 2012, was a touch above 4.1 million. Most years since 2010 have come in somewhere between 4.1 and 4.9 million square kilometres. The bands on this contract are not arbitrary; they straddle the realistic range of modern melt seasons, with the bottom band reserved for a genuine new record.
That is what makes it analytically interesting. The contract is not asking whether the Arctic is warming. It is asking, in effect, where in the post-2012 distribution this particular summer will land, and giving traders a structured way to disagree about it. If you want a primer on the mechanics, our explainer on how prediction market odds work walks through how a probability distribution like this gets priced.
Why the lowest band is the focal point
The seven bands are not equal in their meaning. Six of them sit inside the recent historical range; one of them, the bottom band, would require breaking a 14-year-old record. That asymmetry is the editorial heart of the market. A trader pricing the lowest band is not just predicting a low ice year, they are pricing the probability that 2026 produces a new all-time low.
That is a heavier question than it sounds. The 2012 minimum was driven partly by an unusual August cyclone that shattered already-thin ice; matching it requires either a similar weather event or a base state of the ice that has degraded enough to do without one. As of 17 June 2026, the Polymarket contract had the bottom band, below 4 million square kilometres, trading around 58%, with the next band up (4.0 to 4.2 million) priced near 25%. The remaining five bands shared the rest. That is the snapshot; it will have moved by the time you read this, and the structure matters more than the level.
Worth flagging what the contract does not measure. It does not track ice volume, which is what most climate scientists actually worry about. It does not track multi-year ice, the thick, durable floes that have been disappearing fastest. And it does not track Antarctic ice, which behaves on a different rhythm entirely. The contract measures extent, because extent is what NSIDC publishes daily and what can be cleanly resolved on a fixed date. That is a feature, not a bug, but it is a narrow slice of the underlying physics.
What a climate contract can and cannot tell you
Prediction markets are good at aggregating diffuse information into a single number. They are less good at aggregating information that nobody actually has yet. A US election market in October is rich because thousands of traders are reading the same polls, watching the same rallies and forming overlapping views. A sea ice market in June is thinner; the melt season is a slow-moving physical process, and the people with genuine edge are a small population of polar scientists, modelling groups and a handful of weather desks.
That does not make the market useless. It makes it a different kind of instrument. Treat it less as a forecast and more as a barometer of where the informed-trader consensus sits relative to climatological priors, and the reasons prediction markets are accurate carry over with some caveats about thin participation. The market price is one input among several, alongside the SIPN Sea Ice Outlook published each summer and the operational forecasts from the various national ice services.
iPredicta tracks climate contracts of this kind alongside the more crowded political and crypto markets, partly because they tend to be under-discussed and partly because the resolution rules are usually cleaner. A satellite measurement published by a single named institution on a single named date is about as unambiguous as event-contract resolution gets.
Frequently asked questions
How does this Arctic sea ice contract actually resolve?
It resolves on the minimum daily Arctic sea ice extent reported by the National Snow and Ice Data Center between 1 August and 1 October 2026, measured to the thousand square kilometres. The market stays open until NSIDC publishes the 1 October figure, then settles immediately on whichever band the minimum fell into. Later revisions to the data are not considered.
Why does the contract use seven bands instead of a simple Yes/No record question?
Banding turns a continuous physical measurement into a discrete contract that can resolve cleanly. It also lets traders express more than a binary view: someone who thinks 2026 will be a low year but not a record can price the second band rather than being forced to bet on a record outright. The bottom band, below 4 million square kilometres, is the one that would require breaking the 2012 record.
