The Resolution Gap

01

Markets have arrived

Prediction markets stopped being a curiosity some time in 2025. Combined monthly volume on Kalshi and Polymarket rose from under $5 billion in September 2025 to about $24 billion in April 2026, by Pew Research Center's count of data from The Block ¹. Polymarket alone crossed $20 billion in January 2026, up from roughly $1.2 billion a month in early 2025, per TRM Labs' reading of on-chain contract logs ². Bernstein projects $1 trillion a year in volume by 2030, with industry revenue rising from about $500 million in 2025 to $10.8 billion ³.

Most of that money is sports. Pew puts sport at 80 per cent of Kalshi's volume and 39 per cent of Polymarket's from July 2024 to April 2026 ¹. Applying Kalshi's published fee formula to its public trade data, Yahoo Finance estimated that $234.6 million of Kalshi's $263.5 million in 2025 fees, or 89 per cent, came from sports contracts ⁴. Bernstein expects sport's share to fall to about a third by 2030 as institutions use event contracts to hedge economic, business and political outcomes ³.

Underneath the growth is machinery that gets less attention. Every contract ends with one question: did the thing happen. Someone has to answer it. Who that someone is, and what record they are allowed to point at, is the whole product.

02

How a market resolves today

On Polymarket the answer comes from UMA's optimistic oracle. Anyone can propose an outcome by posting a bond, typically $750, which opens a two-hour challenge window. If nobody disputes, the market settles on the proposal and the proposer takes a reward. If someone posts a matching counter-bond, the question escalates, and after a second dispute it goes to a vote of UMA token holders, which takes about 48 hours. The winning side keeps its bond and takes half of the loser's ⁵. UMA's documentation describes the voting layer as a staking system in which "stakers who did not vote, or who voted against the majority, are slashed, with the slashed amount redistributed to the majority voters" ⁶. The design assumes that a clear public source exists, that proposers can find it, and that voters can check it inside a short window.

On Kalshi the answer comes from the exchange. Kalshi has been a designated contract market under the Commodity Futures Trading Commission since November 2020 ⁷. Each market carries written rules naming an "outcome verification source", which Kalshi calls "an authoritative point of reference to confirm the legitimacy of the strike outcome"; the markets team decides when "resolution criteria has been met" ⁸. The rulebook Kalshi filed with the CFTC in 2025 defines a "Source Agency" as "the agency that publishes the Underlying and/or Expiration Value" for a contract, gives Kalshi "sole discretion to interpret a Contract's Terms and Conditions", and, where a Source Agency "is unavailable or undergoes significant modifications", lets Kalshi "designate a new Source Agency" after trading has begun. Ambiguous outcomes go to an Outcome Review Committee whose determinations "are final and not subject to review" ⁹.

Both systems work when the source is obvious, public, in English and on the wires within minutes.

03

Where the mechanisms stop working

Most of the world's verifiable events are not like that.

Take a Korean baseball result, a Bank of Japan rate decision, a national motocross championship in Mongolia, or a municipal election in an Italian province. Each has an official record-keeper that publishes a result. But the record sits on a local-language website, is rarely mirrored by English-language wires, and is sometimes a PDF or a scanned table rather than a feed. A UMA proposer must locate it and read it correctly within hours. A Kalshi analyst must have named it as the Source Agency in advance. Where neither is practical, the market is not listed, resolves slowly or resolves on whatever an anonymous proposer says the record says. Call that the resolution gap.

The skew shows in the listings. A June 2026 study of 6,047 Polymarket and Kalshi contracts about Africa and Latin America found African inventory "concentrated overwhelmingly in football while salient civic events produce little or no inventory", and concluded that "prediction-market inventories measure what platforms can settle as much as what traders believe" ¹⁰. Platforms list what they can resolve.

Even inside the well-covered slice, the source-of-truth layer has failed in public. In March 2025 a $7 million Polymarket market on whether Ukraine would agree to a minerals deal with the Trump administration before April resolved Yes, though no agreement had been signed. The Defiant traced the outcome to a single actor holding roughly five million UMA tokens across three accounts, about 25 per cent of votes cast. Polymarket said that "because this wasn't a market failure, we are not able to issue refunds" ¹¹. Three months later a market on whether Volodymyr Zelenskyy would wear a suit before July split traders over whether a black blazer with suit trousers, worn at the NATO summit on 24 June, counted ¹². It resolved Yes, was disputed for nine days, and was flipped to No, with $242 million traded by the end ¹³. In January 2026 Polymarket itself ruled that the US operation which captured Nicolás Maduro did not meet its contract's definition of an invasion, leaving more than $10.5 million unsettled ¹⁴.

None of these was a data problem. The blazer was on television and the raid was on every front page. The failure was in who got to say what the record meant, and on what authority. Move the same question to a Mongolian championship or an Italian ballot count, where the record is thinner and the proposers fewer, and it gets worse.

04

Provenance, not more proposers

The instinctive fix is to widen the funnel: more proposers, larger bonds, longer windows, deeper token pools. But UMA's slashing rules already push voters towards the majority rather than towards the evidence, which is the property a concentrated holder exploits ⁶. Adding participants to a process that rewards agreement produces more agreement.

The question a regulator will eventually ask of any of these platforms is who decided this resolved Yes, and what were they looking at. Kalshi's rulebook answers with a named Source Agency and a committee ⁹. UMA answers with a bonded proposer and a vote ⁵. Neither answer reaches back to the organisation that produced the result.

A market is a bet on an answer, and the answer is only as good as whoever is allowed to give it. The durable answer to "who decided" is the body that publishes the official record: the league, the statistics office, the electoral commission, the central bank. If that body signs its publication at release, with a timestamp and a key that belongs to it, a proposer, an exchange analyst or a token voter is no longer asserting what happened. They are pointing at a record whose origin anyone can check. Disputes shrink from "what does the source say" to "did the source sign this".

Contract wording will still be contested, as the blazer and the raid show. Signed records address the other failure, where the resolver cannot find or trust the record at all, and that is where the gap is widest.

05

What this means for verified records

LocalTruth is a modest attempt at that provenance layer. Organisations that publish results, statistics and other primary data anchor a hash of the file to Injective, signed from their own wallet, so anyone can confirm what was published, when, and by whom. The first live use is national motocross results in Mongolia, an event class well outside the slice today's oracles serve. It does not settle markets and it does not prevent disputes. It gives the data owner a timestamped, owner-signed record that a proposer, an exchange or a regulator can point to when someone asks who decided.