The Truth Machine Got Rich the Year Its Story Fell Apart

Executive summary

Two years ago, prediction markets were a curiosity that data nerds and election obsessives checked on their phones. In 2026, they are a category that absorbed a billion-dollar capital round, an exchange-operator investment from the parent of the New York Stock Exchange, and a working federal regulatory posture in the United States.

While that's all real, the other part is that the industry sells itself as an information machine, a place where dispersed knowledge gets priced into a probability that beats pundits and polls, and the 2026 record keeps undercutting such claims in public. The venue that won, Kalshi, won by becoming a sportsbook wearing a derivatives license, with sports running around 80 to 90% of its listed contracts and revenue by its own account to a federal court. The price discovery that does happen is driven by a few percent of traders and a handful of whales. The markets that get the most media attention are the least calibrated. And the onchain platform that built its brand on transparent, auditable settlement spent the same period paying creators to film fake winning bets on copycat websites, which is now the subject of a federal investigation.

The mechanism is far from worthless. But it does mean that the value sits in a narrower part of the stack than the marketing claims suggest, and a serious reader should be able to tell the two apart. The most interesting story related to prediction markets is the distance between what the industry says it is, a truth machine for a confused world, and what its order flow rewards.

The investable and policy-relevant value is in the plumbing rather than the casino. Regulated clearing and surveillance, oracle integrity, institutional data and hedging tools, and the structured-products layer that is starting to wrap event contracts into things a treasurer can buy are where durable demand shows up. A bet on more retail velocity in sports and politics is a bet on the part of the business regulators are most likely to clip, and on the part of the story the evidence supports least. The rest of the report defends that view and stress-tests it.

Key findings

1. Kalshi took three quarters of a record month. In The Block's 3 August 2026 snapshot, Kalshi recorded $37.7B of July volume against $7.9B on Polymarket International and $5.0B on Polymarket US: 74.5% of a three-platform total of $50.59B, an all-time high.
2. The winner won on sports. Kalshi told a federal court that sports event contracts made up between 80% and 90% of both its listed contracts and its revenue as of February 2026.
3. The volume numbers may be worse than unstandardized. Dr. Rajiv Sethi and co-authors found periods in which wash trading accounted for as much as 60% of total Polymarket volume, concentrated in the era before the platform charged fees.
4. The forecasting value is real but small. A Federal Reserve working paper found Kalshi's macro markets well calibrated, with a perfect record on the day before FOMC rate decisions, but those liquid macro markets make up a tiny slice of the volume.
5. A thin layer sets the price. Roughly 3% of traders account for most of Polymarket's price discovery, and in a study of $67B of Polymarket volume the top 1% of profitable users captured 76.5% of all profits.
6. Regulation split in two directions at once. The CFTC under Chairman Michael Selig invoked emergency powers to keep Kalshi trading in New York, while federal district judges in New York and Connecticut held that the Commodity Exchange Act does not preempt state gambling law and 44 state attorneys general told the Commission it has no authority over sports contracts.

Key charts

Methodology

The report sorts its sources into three tiers:

Triangulation. Official and primary sources for platform mechanics, fees, custody, tokenomics, and legal status; regulator releases and court records for the jurisdictional analysis; central-bank and academic work for theory, microstructure, and accuracy. Public market metrics were cross-checked across reputable analytics compilations but not normalized to a single audited basis, so the report leans on directional comparisons with explicit caveats rather than false precision. Where a claim is load-bearing it is footnoted to the strongest available source for that specific claim, and no footnote is used as a thematic bundle covering several unrelated facts. Where the evidence is visible but not verified, the text says so. Volume charts are directional reconstructions.

Dates. Core academic datasets and most market-structure analysis reflect information available through Q2 2026 unless otherwise dated. A publication-date update incorporates July venue volumes, Robinhood's Q2 results, financing developments, and material regulatory and legal actions through 27 August 2026. Interviews were conducted on 3 August 2026. Sources were last checked on 27 August 2026. A further addendum dated 2 September 2026 incorporates The Block's first full post-World-Cup monthly print and its revised July series; no other section reflects information after 27 August 2026. Market volumes, prices, user counts, valuations, regulatory positions, and other time-sensitive figures may change after publication.

Interviewees. Dr. Robin Hanson, Dr. Rajiv Sethi and Wasim Fukase are cited solely for their attributed remarks. Their participation does not imply authorship of, endorsement of, or agreement with the report or its conclusions.

Cite this report

Hima, D., & Shala, A. (2026). Prediction Markets Report 2026: The Truth Machine Got Rich the Year Its Story Fell Apart. Hodl Up Research. https://hodluppod.com/reports/prediction-market-report-2026

Disclaimer: None of this is investment advice, and the report is not a licensed advisor. Its jurisdictional summaries are strategic guidance and not legal advice.