Prediction markets are older than modern finance expects
Prediction markets are not a new invention. In 1503, Rome already had betting on who would become the next pope. By 1916, U.S. election wagering in New York alone reached about $211 million in 2012 dollars. On the busiest days, election-related trading volume was larger than the stock trading on Wall Street’s over-the-counter exchanges.
The core idea has always been the same: use money to price uncertainty. What took centuries to emerge was a durable, scalable market structure. That is the gap Polymarket and Kalshi eventually managed to close.
1988 set the intellectual base, but the market kept failing
Modern prediction markets trace their roots to 1988. Robin Hanson, widely regarded as the field’s founding theorist, published some of the earliest academic work on information markets and idea futures that year. The same year, three University of Iowa professors launched the Iowa Electronic Markets.
After five election cycles, the Iowa Electronic Markets produced probability forecasts that beat polls 74% of the time, reinforcing Friedrich Hayek’s 1945 view that markets are the most efficient way to aggregate crowd wisdom.
Even so, the 2000s and 2010s were filled with failed attempts. In July 2003, DARPA’s policy analysis market was shut down after one day, after two senators attacked it as a betting market on assassinations. The U.S. Congress blocked Hollywood Stock Exchange from becoming a true movie futures exchange. Intrade operated in Dublin for more than a decade before the CFTC sued it in 2012 for offering unregistered options to U.S. users; the platform shut down in March 2013.
Crypto was supposed to fix the problem. Ethereum’s mainnet gave developers programmable infrastructure, and decentralization seemed like a natural fit. But a new bottleneck emerged. Augur, launched in 2018, was weighed down by expensive Ethereum gas fees and poor user experience. At its peak, the platform had only 265 users, then fell to 37 within a month.
The real problem was structural, not just regulatory
By 2024, many projects had died. The usual explanation was harsh regulation, weak execution and clunky products. Nick Whitaker and J. Zachary Mazlish offered a deeper diagnosis in a widely read 2024 Works in Progress essay: a market that can sustain itself needs three core participant groups.
Those groups are savers, who want long-term returns; gamblers, who want excitement; and professional traders, who seek mispricings to arbitrage. Basic prediction markets appeal to none of them strongly enough. They are zero-sum, and after fees they become negative-sum. Savers stay away because they want positive-sum venues for wealth creation.
Most real-world events also resolve too slowly and cover topics that are too niche to attract gamblers, who usually prefer faster outcomes. Without savers and gamblers to provide counterparties, professional traders cannot find enough liquidity. What remains is professionals trading against one another. In practice, that becomes the real-world version of the no-trade theorem: if everyone is rational, nobody wants to be the other side.
Outside of a few categories such as sports and politics, most topics do not draw enough mainstream demand. Without trading volume, professionals have no reason to compete for tiny expected profits. Whitaker and Mazlish concluded that without external subsidies, an “everything market” for predictions cannot scale.
Polymarket solved the product problem first
Polymarket was founded by Shayne Coplan in 2020. A New York University dropout, he took part in the 2014 Ethereum ICO and wrote to Robin Hanson in 2019 about making prediction markets real. During the pandemic, he launched the product from his Lower East Side apartment in New York.
The platform avoided many early crypto pitfalls thanks to better infrastructure. It runs on Polygon, keeping gas fees down to a few cents. It settles in its own stablecoin, PUSD, so a $1 payout is really $1, with no price volatility during the life of the position. Trading uses a hybrid order book: off-chain matching for speed, on-chain settlement for trust. The result feels close to a centralized exchange while preserving non-custodial settlement.
Polymarket also chose a “launch first, fix regulation later” strategy. That gave it faster iteration and more freedom. During the 2020 U.S. presidential election, monthly trading volume reached about $26 million, then expanded further through pandemic- and culture-related markets.
Regulatory risk eventually caught up. In January 2022, the CFTC fined Polymarket $1.4 million and ordered it to block U.S. users. Compliance became the top priority. The platform geo-blocked the U.S., hired a former CFTC chair as an adviser, and continued operating for the rest of the world. In 2023, its trading volume was about $73 million, small by later standards but enough to survive the crypto winter.
The breakout moment came in the 2024 U.S. presidential election. Even with U.S. users blocked, Polymarket became the cultural face of the event. Election-related markets drew about $3.6 billion in total trading volume, and its Trump-win probabilities were more accurate than polls and expert commentary. That surge pushed prediction markets into the global spotlight.
A week after the election, the FBI searched Coplan’s apartment to investigate whether U.S. users had bypassed the 2022 ban through the international site. In July 2025, the DOJ and CFTC closed the probe without filing charges. Days later, Polymarket acquired CFTC-licensed exchange QCEX for $112 million. In October, Intercontinental Exchange, the parent of the New York Stock Exchange, agreed to invest up to $2 billion at an $8 billion pre-money valuation, and became Polymarket’s global distributor for event data. The QCEX deal allowed Polymarket to re-enter the U.S. market in December 2025.
By July 2026, Polymarket had processed 707.7 million trades and more than $111.9 billion in cumulative volume. It led the sector in 2024 before Kalshi overtook it.
Kalshi chose the opposite path: regulation first
Kalshi took a very different route from day one. Co-founders Tarek Mansour and Luana Lopes Lara both graduated from MIT. Lara was once a professional ballet dancer who performed in Swan Lake before moving into finance. The company launched in 2018 with a simple bet: staying compliant from the start mattered more than speed.
They waited nearly two years for approval. In November 2020, the CFTC approved Kalshi as a designated contract market, making it the first U.S. exchange with federal permission to list event-contract derivatives. That license became the legal backbone of the company, since federal law outranks state gambling rules. Kalshi went live in July 2021.
The license also brought friction. New event listings moved slowly, and strict KYC requirements made growth harder. The biggest blow came with the 2024 U.S. presidential election, when Kalshi’s request to launch election markets was rejected by the CFTC. It missed the industry’s biggest catalyst.
Kalshi sued and won. In September 2024, a federal judge ruled that the CFTC had overstepped its authority and that election-related contracts were neither illegal nor gambling. With only 32 days left before voting, Kalshi reopened election trading. By then, the late start and strict KYC had already cost it international users and market share to Polymarket. Kalshi’s election-related volume was about $500 million, versus Polymarket’s $3.6 billion.
After the election, compliance began to pay off. Robinhood partnered with Kalshi to launch its first prediction market product, and Bloomberg Terminal directly integrated Kalshi’s data. Both deals were built on Kalshi’s regulatory status.
In 2025 and 2026, Kalshi expanded aggressively beyond politics. Sports became its largest category, helped by viral marketing. A street interview clip featuring the phrase “Knicks in four” during the NBA Finals drew tens of millions of views. The team also ran World Cup ads featuring Timothée Chalamet, Lionel Messi and Luka Dončić. During the World Cup, 3 million users generated $27 billion in trading volume.
By July 2026, Kalshi had processed 982.6 million trades and $155.7 billion in cumulative volume, overtaking Polymarket to become the sector’s new leader.
Brokers, exchanges and fintechs piled in
The rise of the two leaders pulled in exchanges, brokers and fintech firms. Many chose to move fast by integrating existing prediction markets into their own apps first, then testing demand.
Coinbase and Interactive Brokers went the aggregation route, pulling liquidity and markets from multiple platforms. CME Group built its own product from scratch. Robinhood took a more structured path: it routed orders to Kalshi first, validated demand, then built its own stack and launched the CFTC-regulated exchange Rothera in June 2026.
The bet paid off quickly. Robinhood’s event-contract business generated $156 million in revenue in the second quarter of 2026, more than 10 times year over year, and already exceeded the $100 million it made from crypto trading.
Gamblers and professionals have been served, savers still have not
Polymarket and Kalshi fixed many execution problems: clearer regulatory strategies, lower fees and a better consumer product. But whether they solved the deeper demand problem identified by Whitaker and Mazlish is still unresolved.
The gambler side has been partially addressed, but it is far from an “everything market.” Sports fit prediction markets naturally, and the main question is whether they can take share from traditional sports betting. So far, they have. Most volume this year has come from sports. Same-game parlays are a key growth driver: they made up just 3% of Kalshi’s total volume when launched in September 2025, but rose to 38% by July 2026.
Crypto price markets come next. Politics is no longer limited to elections: military and geopolitical conflict markets generated $2.76 billion in volume, slightly above the $2.73 billion U.S. election market. Add overseas elections, and the broader political category keeps growing. Culture markets, including music, film and celebrities, are also expanding. Federal Reserve rate decisions and inflation have volume too. Several categories are rising, even if the sector is still far from a true universal market.
Professional traders now have more reason to participate. They need both enough volume and counterparties beyond other professionals. Sports markets at $120 billion and crypto markets at $22 billion are already large enough to support that. Susquehanna joined Kalshi as a market maker in 2024 and later formed a prediction-market joint venture with Robinhood. Jump Trading invested in both platforms in exchange for liquidity provision, and Citadel is also said to be evaluating entry.
Savers remain the missing group. Prediction markets are still zero-sum, so capital allocated here gives up the yield it could have earned in Treasurys or elsewhere.
The next phase could be a global information pricing layer
The article argues that prediction markets may next evolve from niche platforms into infrastructure for pricing information worldwide. That shift could bring entirely new market designs.
One path is customized hedging. Companies could hedge risks that traditional finance and insurance do not cover. An ice cream shop, for example, could hedge against a cool summer. Traditional insurers often will not underwrite such niche exposure because it is hard to do profitably.
Another path is moving beyond the 0-to-1 binary format. Perpetual markets could allow continuous trading on any event. On inflation, for instance, binary markets can only ask whether inflation will exceed 3.1%. A perpetual market could let traders go long or short inflation itself. Basket markets and idea governance go further by pricing relationships between events, such as “if Elon Musk resigns, what happens to Tesla’s stock?” or “if the U.S. launches an invasion, where does oil go?” Those markets could help value assets more precisely and even support policy decisions.
AI agents may become round-the-clock truth finders. They could scan Telegram groups and social media for evidence faster than human commentators, trade against mispricings, and improve market accuracy.
On the media side, prediction markets could become more deeply integrated with journalism. Partnerships involving CNN and CNBC suggest a future where reporting does not only cover what has already happened, but also what may happen next. Long-tail topics could expand too, from tech layoffs to album performance and local community events. But demand will require a habit change: people would not just read the news, they would also trade on outcomes.
Yield-bearing collateral may help close the saver gap
The hardest problem remains the absence of savers. One possible fix is to stop using idle USDC as margin and instead allow yield-bearing assets such as sUSDe or tokenized Treasuries. Ethena’s sUSDe alone has generated annualized yields of 4% to 30% over the past two years by capturing perpetual funding rates, and it is already accepted as collateral on platforms such as Aave, Pendle and Morpho. If prediction markets supported similar collateral, saver capital could keep earning while serving as position backing, instead of sitting idle until settlement.
Another path is modular finance. Structured products could be built on top of prediction markets, and event contracts could themselves be used as collateral for lending. Once event contracts can be packaged with yield-bearing assets or pledged for borrowing, zero-sum betting would no longer be the only return source for that capital.