On October 13, investment bank Macquarie released a new version of its prediction market estimate: full-year 2026 "taker volume" is projected at $187 billion. This measure counts only the side actively taking orders, excluding the market-making side. Last year's full-year figure was $22 billion, an 8.5x increase in one year. This new version is actually conservative—the September version gave $190 billion, and an earlier version gave $169 billion. The numbers have been raised all along, and the market has kept exceeding expectations.
Structural data is more interesting than the total. In September, Kalshi took a 63% share of sports event contracts, and DraftKings app downloads also reached a double-digit share. The pace is even more striking: the first week of September had no NFL schedule, yet the market still did $4.3 billion; on the Sunday when the NFL season opened, the single-day record was $802 million. Third-party figures put Q3 total contract volume at $188 billion, up nearly 70% quarter over quarter.
Analyst Chad Beynon's judgment is that the U.S. midterm election cycle will continue to supply volume for political event trading. Looking further ahead, Macquarie's July long-term estimate projected $1.5 trillion by 2030, of which the non-sports portion was $783 billion, overtaking sports derivatives at $705 billion—the main growth driver is gradually shifting from sports events to elections, economic data, and similar events.
For teams building package network platform solutions, the value of this report is that it turns "whether to add a prediction market category" from a gut call into an accounting question: category capacity, growth rate, and head concentration—three sets of numbers on the table—give the integration decision a basis. When selecting a solution, ask suppliers one more question: does the event contracts module already exist, and who provides the backstop for clearing and compliance.
There is also a half-truth: prediction markets are currently a special category under U.S. regulatory rules, and multilingual market rules differ. Whether to enter and which market to enter first must follow the license map; institutional data can only help halfway.