In June 2026, Polymarket ran a market on a question that sounded simple: would MicroStrategy, the company that’s spent years hoarding Bitcoin, sell any of it before May 31?
The company’s own SEC filing said yes. It sold 32 BTC between May 26 and May 31, squarely inside the window. It was documented and public, and anyone could check it.
The market resolved No.
More than $60 million had been traded on that question. People who read the filing correctly, who called reality exactly right, lost their money anyway. Nobody hacked the platform. Nobody broke a rule that existed when they traded. The system did precisely what it was built to do, and that’s the part that should worry you more than a hack would.
This is a piece about resolution, which is probably the least glamorous and most important part of any prediction market, and the part almost nobody checks before they trade.
The problem: markets can’t see the world
A prediction market can’t observe reality. All it can do is ask someone or something to tell it what happened. Whoever that is ends up controlling your payout, more than the price chart or the news ever will.
So before you trade anything, the real question is this: who decides what happened, and what happens when they get it wrong?
There are two answers in use right now, and they behave differently under pressure.
| A company decidesKalshi, Polymarket US | A vote decidesPolymarket international | |
|---|---|---|
| Who calls it | The exchange itself. | Holders of a cryptocurrency token called UMA. Not judges, not subject-matter experts. |
| Under what rules | Rules filed in advance with a federal regulator. | The market’s written terms, read by whoever turns out to vote. |
| When the question is genuinely ambiguous | The exchange interprets its own filed rules. | The outcome becomes whatever the largest coordinated bloc of token holders decides happened. |
| If you disagree | A formal complaints process, with a regulator sitting above the company if it fails. | No regulator above the vote. You know the result, not whose name goes on the complaint. |
Answer one: a company decides
On the regulated US platforms (Kalshi, Polymarket US), the exchange itself declares the outcome, under rules it filed in advance with a federal regulator. There’s a formal complaints process and a regulator sitting above the company if that process fails. It isn’t perfect, but at least it’s legible. You know exactly whose name goes on the complaint.
Answer two: a vote decides
The international version of Polymarket handles disputes another way entirely. Contested markets go to a vote of people holding a particular cryptocurrency token, called UMA. These aren’t judges or subject-matter experts. It’s whoever holds the most tokens and bothers to show up.
For most markets this works fine: the answer is obvious, nobody disputes it, and the vote is a formality. The trouble starts when a question is genuinely ambiguous. At that point the outcome stops being what happened and becomes whatever the largest coordinated bloc of token holders decides happened. Which brings us back to MicroStrategy.
How Yes became No
The market’s rule asked whether the company would sell Bitcoin “by May 31.” Nobody had pinned down what “by” meant, and it turned out there were two honest readings:
- The sale happened between May 26 and 31, inside the window by any normal reading of English.
- The public confirmation, the SEC filing, didn’t land until June 1, a day after the deadline.
Sixty million dollars traded on a sentence with two legitimate readings, and nobody clocked the gap until it mattered.
Here’s the part that should bother you more than the No itself. Traders allege Polymarket added new language after trading had already closed, specifying that a sale had to be publicly confirmed before the deadline rather than merely completed by it. That’s the thing that flipped a Yes into a No. It’s also, as of July 2026, the subject of a lawsuit: two traders sued Polymarket in New York in early July, arguing the platform rewrote its own rules retroactively to dodge paying out. Whatever a court decides, the lesson underneath doesn’t change. On some platforms, the rulebook isn’t fixed at the moment you trade. It can move under you.
This wasn’t a one-off
None of these are edge cases anymore.
A Wall Street Journal investigation into Polymarket’s dispute process found that in most contested markets, more than half the votes came from just ten wallets. Roughly one in five disputes involved a voter who had money riding on the very outcome they were deciding.
In 2025, a market on a Ukraine minerals deal resolved Yes after a single large token holder cast about a quarter of the vote, despite no deal being finalized. Traders on the other side lost around $7 million. Polymarket’s response was blunt: this wasn’t a platform failure, so no refunds were coming. Everything, technically, sat within the rules.
And disputes are climbing rather than shrinking. More than 1,150 Polymarket markets were disputed in the first five months of 2026 alone, already ahead of the full 2025 total.
None of this means prediction markets are rigged or that most trades go bad. Most markets resolve quickly and cleanly, exactly the way you’d expect. But the industry’s central sales pitch, that prices reflect the truth better than polls or pundits, rests entirely on a step most traders never think about. The trading part works. Deciding what happened is the part nobody has fully solved yet.
What you can actually do about it
- Read the rules, not the headline. If there’s a deadline, work out whether it refers to when the event happened or when it got confirmed. That one distinction has already decided a $60 million market.
- Check for added clarifications. Some platforms can attach notes to a market after it launches, and those notes become part of the rules you’re trading against. Trading on the original wording alone means trading against rules you’ve never read.
- Price in the ambiguity. If two reasonable people could read a rule two different ways, the sure thing you think you’re buying is worth less than the price suggests.
- Know which game you’re in. Once a market goes to dispute, you’ve stopped betting on the event. You’re now betting on the outcome of a vote among token holders you’ll never meet, and that’s a different market, so treat it that way.
You’re never betting on reality. You’re betting on how the rules describe reality, and on who gets the final word when that description turns out to be ambiguous. The gap between those two things is where money quietly disappears, through fine print nobody read until it cost them.
Nothing here is financial advice. Prediction markets involve real risk, so never trade money you can’t afford to lose.