One probability, a thousand markets

A prediction market runs on a small miracle of translation. A price is a probability, and a probability is the same everywhere — 62¢ means 62% in New York, in London, in Hanoi. Truth, unlike money, has no nationality. That universality is the whole promise: the most global financial instrument ever built, a single number the entire world could read off the same screen.

So here is the uncomfortable thing about the boom we're living through. The price is global. The market is not. Nearly every dollar of the roughly $24 billion a month now flowing through prediction markets¹ is trapped inside a silo — one license, one jurisdiction, one liquidity pool, one currency. We aren't building a global probability layer. We're building a thousand disconnected ones and calling them the same thing.

The category has, by now, won the argument. It's real money, it's institutional, a Federal Reserve study confirmed the forecasts are as good as the professionals', and CNN and CNBC put the odds on television. What it hasn't done is the unglamorous part — make the markets connect. This piece is about the two problems standing in the way, both of which are clearest from where I sit, in Vietnam.

$24B/mo

Traded on prediction markets — the argument is won¹

$22B

Kalshi's valuation, from roughly zero in 2021²

~$1T/yr

Projected category volume by 2030¹

Problem one: liquidity has a passport

Why is the market local when the price is global? Because a real-money prediction market is a licensed financial venue, and a license is granted one country at a time. Kalshi can operate because the U.S. regulator blessed it; it cannot take Vietnamese deposits, and a Vietnamese exchange — when one exists — won't take American ones. We've written before about how the whole game is a race across a map of licenses. The flip side of that map is this: every license is also a wall.

So a trader in Chicago and a trader in Hanoi cannot meet in the same order book — even on the exact same global event. "Will the Fed cut in September?" is a question that moves the price of money in Vietnam as surely as in Ohio. But the American trades it in a deep dollar pool on Kalshi, and the Vietnamese, if she can trade it at all, trades it in a separate, shallow pool somewhere else. Same event, same truth, two markets that never touch.

This is worse than it looks, because liquidity is the product, and liquidity is winner-take-most. A market with few traders is a noisy, manipulable, badly-priced market — the field's oldest problem. Split one healthy global pool into twenty national ones and you don't get twenty healthy markets; you get one deep pool in the dominant jurisdiction and nineteen thin, unreliable ones everywhere else. Fragmentation doesn't just divide the liquidity. It destroys most of it.

Problem two: the dollar's lucky accident

Now the subtler problem — the one I keep running into every time I try to explain this at home. The reason "62¢ = 62%" feels so natural is that the U.S. dollar is a lucky unit. A contract pays out $1 if it resolves yes. The price, quoted in cents, reads straight off as a percentage. Sixty cents, sixty percent. No arithmetic, no translation. A child could read the odds.

Try it in Vietnamese đồng. A dollar is about 26,000 đồng. A contract that pays one đồng is a joke — worth four-thousandths of a U.S. cent. So you'd pay out something like 26,000 đồng instead, and now the price of a 60% chance is ₫15,600. Quick: is ₫15,600 a 60% chance, or 59%, or 61%? You can't see it. The single feature that made prediction markets legible to ordinary people — the price is the probability — evaporates the moment you leave the dollar.

You can see it in one line. A binary contract that pays a unit \(U\) on a yes trades at price \(=p\cdot U\), so the implied probability is \(p = \text{price}/U\). The legibility only appears when the unit is one:

And the denomination is the smaller half of the currency problem. The bigger half is contamination. If the market you can reach settles in đồng but the thing you want to bet on is global, your position now carries an exchange-rate bet you never asked for: you wanted a view on the Fed and ended up also long or short the dollar against the đồng. It's the basis-risk problem again, wearing a different hat — the instrument isn't shaped like the view. Pile capital controls on top, and "just trade the deep U.S. market in dollars" isn't even an option for most people here. The clean dollar pool is right there on the screen, and legally out of reach.

Why it doesn't fix itself

In a normal market, fragmentation like this would heal on its own. If Kalshi priced the Fed cut at 62% and a Vietnamese venue at 57%, an arbitrageur would buy the cheap side, sell the dear side, and pocket the spread — and that very act would drag the two prices together until they met. The law of one price is just arbitrage, working. It's how a thousand quotes for the same stock collapse into one.

Here, it can't run. To close that 5-point gap you'd have to move money from the đồng pool into the dollar pool and back — across two licenses that don't recognize each other, two currencies, and a capital-control regime that forbids the flow. The trade that would unify the price is illegal, impossible, or both. So the pools never converge. Each becomes its own little island of "truth," drifting. The single global probability we started with quietly fractures into many.

What solving it actually looks like

The naive fix — one giant global exchange — is a non-starter, and not only because regulators won't allow it. They shouldn't have to: a country has every right to license and surveil the markets its citizens trade. The license being national isn't a bug to route around; it's the foundation. The answer has to keep the license local and connect everything else.

That points to a federation, not a monolith. Picture locally licensed venues — one per jurisdiction, each settling in its own currency and legible to its own people — sitting on top of a shared back-end. Two things get shared. First, a single resolution layer, so "did the Fed cut?" resolves identically everywhere and the islands at least agree on what happened. Second, a common unit of account and settlement rail, so liquidity can net across borders without any single venue breaking its own rules. Local front-ends; global plumbing.

Of those two shared layers, resolution is its own deep problem — getting "did the Fed cut?" to resolve identically everywhere is harder than it sounds, and worth a piece of its own. But the unit of account is the one that quietly decides who holds power, and it's where the rest of this piece goes. It sounds like plumbing — pick a token, wire it up — and it is. But the choice of what a thousand markets settle in turns out to be one of the most consequential decisions in the whole design.

The settlement half of the problem

So what fills that box? Three years ago this would have been a hand-wave. To let a market in Hanoi net against a market in New York, both sides have to hold and settle in the same thing — a unit each can hold, that moves across borders cheaply, and that nets cleanly. There was no good candidate; you could draw the federation but not finish it. Now there's a real answer, because the dollar went digital and got a law. And the thing the entire industry is quietly converging on is a US-dollar stablecoin.

For America that's just convenient. For an emerging market, choosing the dollar as your settlement rail is a monetary decision with a name — dollarization — and it's arriving through the back door, one app at a time. That is the second, subtler half of the currency problem: not just that the đồng is illegible on the screen, but that the cure for it may quietly move the country's savings into someone else's currency.

Why the answer is probably a stablecoin

The reason there's an answer now is regulatory, not just technical. The GENIUS Act, signed in July 2025, is the first US federal framework for payment stablecoins: it requires every coin to be backed one-for-one by dollars or short-dated Treasuries — exactly the property a settlement rail needs.⁶ The market followed. Total stablecoin capitalization grew about 49% in 2025, from roughly $205B to over $300B.⁷ A regulated, dollar-pegged, programmable token that moves worldwide in seconds is very nearly the perfect connective tissue for a global market — and, not incidentally, it restores the legibility we lost back in the đồng: a token worth ~$1 makes 62¢ mean 62% again, no division by 26,000. Settle the world's markets in it and they're all legible, all nettable, one click apart. Which is precisely the problem.

$306B

Stablecoin market cap, end-2025 — up ~49% on the year⁷

1:1

USD/Treasury backing the GENIUS Act now requires⁶

Jul 2025

First US federal stablecoin law signed⁶

A dollar stablecoin is dollarization

Here's the part that's obvious from Hanoi and invisible from San Francisco. If Vietnamese traders settle their markets in a US-dollar stablecoin, they are — at the margin, and then more than the margin — moving their savings and their risk into dollars. That is dollarization: the slow substitution of a foreign currency for the local one inside your own economy. It has happened before, to countries that didn't choose it, and it costs a central bank its most important levers. When enough of the economy runs on dollars, the State Bank's interest rate, its exchange-rate management, its lender-of-last-resort power — the very things prediction markets would help citizens forecast — all weaken. The feature that makes the dollar stablecoin a perfect global rail (it's the dollar) is exactly what a monetary sovereign has spent decades guarding against.

So the unit-of-account question isn't a technical detail to be optimized away. It's a genuine trilemma, and you can have at most two of the three things you want.

A local-currency stablecoin keeps sovereignty but reintroduces the đồng's legibility problem and strands you in a thin, local pool — straight back to the fragmentation we started with. A central-bank digital currency keeps sovereignty and can be programmable, but a CBDC that settles cleanly across borders is years away and carries its own surveillance baggage. The dollar stablecoin wins every box but the last — and the last is the one a finance ministry will fight over.

Why this is a Vietnam problem first

Every part of this is sharpest here. The đồng makes the legibility problem acute; capital controls make the fragmentation total; a volatile exchange rate makes the contamination real; and there's no domestic license yet, so the deep pools are all offshore and out of reach. Vietnam is, in a sense, the hardest version of the problem.

Which is exactly why it's the one worth solving. We argued in "The forecast Asia is missing" that the value of a good market-based forecast is highest where the existing tools are thinnest — and that's emerging Asia, not Manhattan. The same logic applies to the plumbing: whoever solves local-currency, locally-licensed markets that still plug into global truth unlocks the markets that need them most.

What makes this concrete rather than theoretical is that Vietnam just built the runway. After years in a grey zone, the National Assembly passed a Law on Digital Technology Industry in June 2025 that recognizes digital assets under the Civil Code, effective January 2026; a September 2025 resolution then opened a five-year pilot for a regulated crypto-asset market.⁸ A country with one of the highest crypto-adoption rates on earth has, for the first time, a controlled sandbox in which exactly the settlement question can be answered: can you give people legible, well-settled markets without handing the dollar the keys to your monetary system?

The most plausible answer is a hybrid the pilot could actually test: settle and net in a regulated stablecoin for cross-border legibility, but denominate and report to users in đồng, with the central bank in the loop on reserves and flows — sovereignty preserved at the layer that matters, legibility borrowed where it helps. Nobody has shipped that. It's unsolved, it's contested, and it's the kind of thing that gets decided by whoever builds the first credible version under a real license.

What's hard

The honest caveats are large, and they run through both halves of the problem. On liquidity: federating local venues onto shared rails is an enormous coordination and engineering lift, and it only pays off once enough venues plug in — a cold-start that has defeated better-funded attempts. On the unit: a stablecoin is only as good as its reserves, and a depeg turns "62¢ = 62%" back into chaos — the GENIUS Act reduces that risk without erasing it. Capital controls and the pilot's own guardrails may simply forbid the cleanest design. And the deepest tension doesn't resolve: the more useful the shared dollar rail is, the more it erodes the sovereignty it runs through. There is no settle-everywhere-in-a-readable-unit option that doesn't, somewhere, cost a central bank some control.

None of this is a reason for pessimism; it's the nature of an open problem — a map of real choices with real costs, not a list of bugs to patch. The category spent five years proving that prediction markets function. The next five are about something quieter and more important: making them connect. A probability is supposed to be one number the world can agree on. Right now it's a thousand numbers, in a thousand currencies, behind a thousand licenses — and the markets we keep promising the world don't connect until they share a unit, while the unit they're reaching for is the dollar. Closing that gap so that 62¢ in New York and ₫16,000 in Hanoi mean the same 62% and trade as if they were one — without quietly exporting a country's monetary sovereignty in the process — is the unglamorous plumbing under every grand vision of this space. Whether emerging markets get to keep a hand on the wheel while they plug in is, I think, one of the more consequential questions this whole category raises, and it'll be answered first in places like Vietnam, not in Washington. It's the layer we've chosen to build toward at Seeker, with Vietnam as the launchpad: local license, local legibility, global truth, sovereignty intact. The demo is live; the license is the goal; the answer isn't written yet. We don't claim to have solved it. We've decided it's the one worth solving.