Key figures: Polymarket valuation (reported): ~$15bn · Kalshi valuation (Bloomberg): ~$22bn · Polymarket March 2026 volume (self-reported): $10.6bn · Bernstein full-year 2026 projection: $240bn

Last week, Polymarket entered talks to raise $400 million at a reported $15 billion valuation, according to Bloomberg, barely 18 months after the platform was a niche crypto experiment used mainly to bet on US elections.1 Its rival Kalshi raised over $1 billion in March at a reported $22 billion valuation, led by Coatue Management, according to Bloomberg.2 Polymarket says it processed $10.6 billion in trading volume in March 2026 alone, its first month above $10 billion, while blockchain analytics firm TRM Labs estimates that prediction markets as a whole exceeded $20 billion in monthly volume by early 2026.3 Bernstein projects total prediction market volume could reach $240 billion for the full year, roughly quadrupling the $51 billion traded in 2025.4

These are extraordinary numbers, and they should be read with appropriate caution. Platform-reported volumes are not independently audited in the way that traditional exchange data is. But even discounting for inflation in self-reported figures, the directional trend is clear: prediction markets have grown by roughly an order of magnitude in under two years, attracting serious institutional interest from the likes of ICE, the parent company of the New York Stock Exchange, which has reportedly committed up to $2 billion in Polymarket across multiple tranches.1

The more interesting question is theoretical. Prediction markets rest on a specific claim: that when people put money behind their beliefs, the resulting prices aggregate information more efficiently than any poll, pundit, or model. If that holds, these platforms are not gambling. They are the purest form of price discovery in existence.

How the Mechanism Works

A prediction market is deceptively simple. A platform lists a yes/no question, say "Will the Fed cut rates in June?" Traders buy shares in "yes" or "no" at prices between $0.01 and $0.99. If the event occurs, "yes" shares pay out $1. If it does not, they pay nothing. The current share price therefore represents the market's implied probability that the event will happen.

Event contract: A financial instrument that pays a fixed amount if a specified event occurs and nothing if it does not. The market price of the contract reflects the crowd's implied probability of that outcome.

The theoretical case rests on two pillars. The first is the efficient market hypothesis: prices in liquid markets reflect all available information because profit-seeking participants trade on any edge they possess, driving the price toward the true probability. The second is the wisdom of crowds: the median judgement of a large, diverse group of people acting independently tends to be more accurate than any single expert.5

The mechanism is the same one that operates in equity or commodity markets: participants with superior information buy underpriced contracts and sell overpriced ones, and in doing so they move prices toward the correct value. Unlike a poll, where respondents face no penalty for being wrong, prediction market participants have skin in the game.

The Evidence So Far

Academic evidence predating the current boom is genuinely impressive. Research using Iowa Electronic Markets data found that prediction markets outperformed polls in US presidential elections 74 percent of the time, with the advantage growing sharply for forecasts made more than 100 days before election day.6

The 2024 presidential election was the breakout moment for modern platforms. While major polling averages had the race as essentially a coin flip, Polymarket had Trump at 58 percent on the Monday before election day, according to CNN, a lead that proved far closer to the actual result.7 The platform attracted over $3.2 billion in total wagers on the presidential race, according to Fortune.7 When Biden withdrew on July 21, a Harris win contract on Polymarket jumped from 18 to 29 cents within hours, a shift that traditional polling took days to register.8

Polymarket + Polymarket US monthly taker-notional trading volume ($bn). Source: The Block; BitKE.3
3

Data caveat: Prediction market volumes are self-reported by platforms and derived from on-chain data. They are not subject to the same independent auditing standards as regulated exchange volumes. Figures should be treated as indicative rather than precise.

However, accuracy claims deserve scrutiny. Polymarket advertises a 94 percent accuracy rate, but independent analysis by data scientist Alex McCullough found this figure applies four hours before resolution, when outcomes are often already clear. One month before resolution, accuracy is closer to 90 percent. A peer-reviewed Vanderbilt study of 2,500 political markets during the 2024 election found platform-wide accuracy of 67 percent on Polymarket, 78 percent on Kalshi, and 93 percent on PredictIt. Thinly traded contracts are a recognised weakness: as Bloomberg's Matt Levine has observed, markets with only a handful of active participants can be moved meaningfully by a single order, undermining the information-aggregation that is supposed to make prediction markets work.9

"Financial markets are generally pretty efficient, and the evidence suggests that the same is true of prediction markets."

– Eric Zitzewitz, professor of economics, Dartmouth College.7

Where the Theory Breaks Down

Efficient markets require conditions that real markets approximate imperfectly: large numbers of rational participants, low transaction costs, freely available information, and no barriers to arbitrage. Three issues stand out.

Thin markets are noisy markets. The impressive accuracy figures come disproportionately from high-liquidity, high-profile events like US elections. Many individual contracts on Polymarket have only a few thousand dollars of liquidity, where a single large trade can move the implied probability by several percentage points. In these markets, the "wisdom of crowds" degrades into the opinions of a handful of participants.

Insider trading is a structural concern. In traditional financial markets, insider trading is illegal because it undermines confidence in price fairness. In prediction markets, better-informed participants theoretically improve price accuracy. But the line blurs when those participants are government officials or campaign operatives betting on events they can influence. TRM Labs' analysis of on-chain trading patterns identified clusters of potentially coordinated activity coinciding with major geopolitical events, including US airstrikes against Iran, though the patterns are consistent with, rather than conclusive of, informed trading. Both Kalshi and Polymarket announced new insider trading rules on March 23, 2026, though enforcement mechanisms remain unclear given the CFTC's limited resources.10

The gambling question is now before Congress. Senators Adam Schiff and John Curtis introduced the Prediction Markets Are Gambling Act (S.4160) on March 23, 2026, which would ban CFTC-registered platforms from listing contracts resembling sports bets or casino games.11 A separate, broader bill from Senator Jeff Merkley would ban contracts on elections, government actions, and military operations entirely.11 If prediction markets are classified as gambling rather than information markets, they face a fundamentally different legal framework, one that could restrict participation, cap position sizes, and exclude institutional capital.

Prediction market annual trading volume ($bn), 2022 to 2026 (projected). Source: Bernstein; TRM Labs.4
4

2026 is a projection: The $240bn figure is Bernstein's forecast extrapolated from year-to-date trends. Actual full-year volume may differ significantly.

Competition and Market Structure

The industry is consolidating around two dominant platforms with distinct competitive advantages. Kalshi operates as a CFTC-regulated exchange, has been serving US customers since 2021, and has moved aggressively into sports contracts, which accounted for roughly 86 to 91 percent of its volume through the 2025-26 sports season. Polymarket is built on blockchain, only recently re-entered the US market after a regulatory settlement, and only began charging fees in early 2026.3 The valuation gap between them reflects Kalshi's head start in US monetisation.

The entry of ICE, which has reportedly committed up to $2 billion in Polymarket across multiple investment tranches, is reshaping the landscape.1 Robinhood partnered with Kalshi in early 2025 to offer prediction markets through its brokerage platform.10 CME and Cboe have explored event contracts. These institutions bring liquidity, regulatory credibility, and distribution. But their arrival also raises the barrier to entry for smaller competitors. The prediction market space may be heading toward the same concentration that characterises traditional exchanges: a small number of high-liquidity platforms capturing most of the volume, connected to a long tail of thin, illiquid contracts where prices are unreliable.

Prediction markets are a compelling test of a foundational idea: that decentralised, incentive-driven exchange produces better information than centralised authority. The evidence so far suggests they often do, particularly in deep, liquid, high-profile markets where the conditions for efficient pricing are closest to being met. Extend the mechanism to thinner markets, add the distortions of insider information, and layer on regulatory uncertainty, and the picture becomes considerably less clean.

The scale of growth is real, even if the precise numbers deserve healthy scepticism. And like all markets, prediction markets will ultimately be shaped by the same forces that have always mattered: competition, information asymmetry, incentive design, and the rules of the game.

Footnotes

  1. Bloomberg (opens in a new tab), "Polymarket Seeks $400 Million in New Funding at $15 Billion Valuation," 20 April 2026. ↩ ↩2 ↩3

  2. Bloomberg (opens in a new tab), "Kalshi Gets $1 Billion in New Funding at $22 Billion Valuation," 19 March 2026; Fortune (opens in a new tab), "Kalshi locks in $22 billion valuation," 20 March 2026. ↩

  3. BitKE (opens in a new tab), "Polymarket Tops $10 Billion Monthly Volume for First Time in March 2026," April 2026; TRM Labs (opens in a new tab), "How Prediction Markets Scaled to $21B in Monthly Volume in 2026," 27 March 2026. ↩ ↩2 ↩3

  4. Bernstein, prediction market industry report, April 2026. ↩ ↩2

  5. Wolfers, J. and Zitzewitz, E., "Prediction Markets," Journal of Economic Perspectives, 2004; The Conversation (opens in a new tab), "How polling failures paved the way for the explosive rise of prediction markets," 31 March 2026. ↩

  6. Berg, J., Nelson, F. and Rietz, T., "Prediction Market Accuracy in the Long Run," (opens in a new tab) International Journal of Forecasting, 2008. ↩

  7. CNN Business (opens in a new tab), "How prediction markets saw something the polls and pundits didn't", 8 November 2024; Fortune (opens in a new tab), "Polymarket users have wagered $3.2 billion on the outcome of the election—and the majority of bets are on Trump winning ", 5 November 2024. ↩ ↩2 ↩3

  8. Fortune (opens in a new tab), "Prediction markets post-Biden," July 2024. ↩

  9. The Defiant (opens in a new tab), "Polymarket is Up to 94% Accurate In Predicting Outcomes: Analysis," March 2025; Clinton, J.D. and Huang, T., "Prediction Markets? The Accuracy and Efficiency of $2.4 Billion in the 2024 Presidential Election" (opens in a new tab), Vanderbilt University, 2025; Matt Levine, "Prediction Market Making Is Hard" (opens in a new tab), Bloomberg, 13 April 2026. ↩

  10. TRM Labs (opens in a new tab), 27 March 2026; Sacra (opens in a new tab), Polymarket company profile. ↩ ↩2

  11. Congress.gov (opens in a new tab), S.4160, "Prediction Markets Are Gambling Act"; Axios (opens in a new tab), "New Senate bill would ban prediction markets on sports, politics and military," 26 March 2026. ↩ ↩2