In the first half of 2026, prediction markets stopped being a curiosity and became a fixture of American financial life. Polymarket and Kalshi together cleared record volume - by some estimates around $23.6 billion in March alone - and Polymarket logged a single-day record near $425 million in late February. Weekly turnover that sat in the low hundreds of millions a year ago now runs into the billions, a jump of roughly 2,000 percent year over year. The numbers are no longer the interesting part. What is interesting is that the political and regulatory fight these platforms spent years bracing for has, for now, broken in their favor - and that this victory may have distracted everyone from the problem that actually decides whether prediction markets become real financial infrastructure or just a faster way to gamble.

What actually changed in Washington

The clearest shift is at the Commodity Futures Trading Commission. Under its new chair Michael Selig, sworn in at the end of 2025, the agency withdrew an earlier proposal to ban political and sports event contracts and instead opened a rulemaking process in March 2026, asserting that event contracts fall under its exclusive federal authority. That is a profound change of posture. A year earlier the operating assumption was that Washington might try to shut these markets down. Now the federal regulator is effectively claiming them as its own turf.

The politics reinforced the law. President Trump publicly backed the CFTC's authority over event contracts, and Donald Trump Jr. advises both Polymarket and Kalshi through 1789 Capital. Polymarket, which had exited the United States after an earlier settlement, secured a path back through a registered intermediary with CFTC approval in December 2025. For platforms that spent years in legal limbo, this is as close to a green light as the system gives.

But a federal green light is not the same as a settled question. Gary Gensler, among others, argues that the CFTC lacks the authority under Dodd-Frank to treat these contracts as ordinary derivatives. The states are not waiting for that debate to resolve. A federal judge in Ohio called it absurd to treat these products as anything but gambling, an injunction landed in Massachusetts in January, Kalshi is fighting on the order of nineteen federal lawsuits, more than a dozen states have bills in motion, and the New York attorney general has issued warnings. The federal preemption claim is strong, but it is being tested in real time, and a single adverse ruling at the right level could reopen everything.

The fight everyone is having is the wrong one

Almost all of the public argument is about a single question: are these event contracts, or is this gambling? It is an emotionally satisfying fight, and it matters for licensing and taxes. Sports now make up the overwhelming majority of the volume - close to 87 percent of Kalshi's roughly $39.7 billion in annual turnover, and a large share of Polymarket's as well - so the line between a financial hedge and a parlay bet has never been blurrier.

But the gambling-versus-contracts framing is a distraction. Whether you call a market a contract or a wager, it only has value if its outcome can be settled honestly. That is the part of the system that nobody won in Washington, because Washington was never the place where it could be won.

The problem that actually decides this

Every prediction market rests on a resolution layer - the mechanism that decides, after the fact, what actually happened and who gets paid. On regulated venues this is handled by the operator. On the crypto-native side it runs through oracles, most visibly UMA on Polymarket. And this layer has been failing in ways that should worry anyone who treats these markets as truth machines.

The examples are not edge cases. Polymarket has seen disputes over a market on a so-called clavicular pregnancy that paid out around $16.46 million, and over whether a US and Iran peace deal had occurred, where the real question was not about the world but about how the oracle would read it. Analysts have flagged that a single large holder controlled roughly a quarter of the tokens used to resolve disputes on UMA. When the entity that decides the truth can also be the entity with the largest position, the market is no longer aggregating information. It is auctioning a verdict.

This is the argument prediction markets have not won. A market can be perfectly legal, fully regulated, deeply liquid, and still worthless as a signal if its resolution can be captured. The CFTC can bless the wrapper, but it cannot manufacture an oracle that cannot be bought. No volume record fixes this, because the weakness lives in the settlement, not the trading.

Integrity is the second-order risk

The resolution problem has a twin: who is allowed to trade, and on what. Both platforms adopted market-integrity rules in late March 2026 after a string of embarrassments. Kalshi suspended three congressional candidates who were betting on their own races. Polymarket faced scrutiny over trades tied to Venezuela and Maduro, prompting a Senate letter from Catherine Cortez Masto in January. These are not abstract governance concerns. A market where insiders can bet on outcomes they influence is not a forecasting tool, it is a tip line with a payout.

The uncomfortable truth is that the more useful a prediction market becomes - the more people treat its odds as a real probability - the more incentive there is to manipulate both the trading and the resolution. Success makes the integrity problem worse, not better, which is exactly the opposite of how a maturing market is supposed to behave.

Institutions are circling anyway

None of this has scared off serious money. Reuters reported institutional interest building through the spring, Chainlink is positioning a trust layer for market resolution, and the platforms are building their own rails - Polymarket moving toward its own chain and a central-limit order book, Hyperliquid advancing its HIP-4 framework. There is also a large offshore shadow market: US users are estimated to account for somewhere between 7 and 21 percent of global offshore volume, which itself runs to roughly $159 billion.

That institutional pull is the strongest argument that prediction markets are here to stay. It is also the strongest reason to fix the resolution layer quickly, because institutions will tolerate regulatory ambiguity far longer than they will tolerate a settlement they cannot trust.

What to watch

The headlines through the rest of 2026 will be about court rulings, state bans, and volume records. Those are worth tracking, but they answer a question that has mostly already been decided in the platforms' favor. The question that is still open is quieter and more technical: can these markets build a resolution and oracle layer that is economically impossible to capture, and can they keep insiders out of the trades they can move?

If they can, prediction markets graduate into genuine financial infrastructure - a real-time, hard-to-fake probability feed that the rest of finance can build on. If they cannot, they will have won Washington and lost the only argument that mattered, and the record volumes will turn out to have been the easy part.

This article is based on public reporting and data from 2026, including CNBC, Forbes, Reuters, NPR, Congress.gov, Galaxy Research and other public sources.