Volume Follows The License.
$50.6 billion in July, another record. Kalshi's share rose to 74.5%. Polymarket's regulated US arm grew 54% while its international platform fell 26%. IG Group paid up to $1.3 billion for a licensed venue rather than building one. Four independent signals, all pointing the same way.
July set another record. The share picture is what changed.
Kalshi, Polymarket and Polymarket US posted $50.59 billion in combined volume in July, an all time high and a 7.8% increase on June's revised $46.95 billion. The 2026 FIFA World Cup, which ran to the 19 July final, accounted for the majority of the increase.
The concentration is the more consequential movement. Kalshi recorded $37.7 billion, roughly 74.5% of the combined total, up 14% month on month and its highest monthly figure on record. In June its share of the equivalent total sat closer to two thirds. The category set a record and became materially less distributed in the same month.
One caveat operators should carry into any comparison: these figures measure taker notional volume, not revenue and not customer deposits. A single contract can be traded several times before settlement, which lifts recorded volume without an equivalent amount of new capital entering the venue. July's record demonstrates turnover and liquidity depth. It does not establish that $50.6 billion was deposited.
The category set a monthly record and became less distributed at the same time. Both facts matter, and they point in different directions for anyone deciding where to build.
The most revealing number in July was a migration
Polymarket operates two venues. Its offshore international platform, and Polymarket US, the regulated domestic arm operated by QCX LLC. In July those two moved in opposite directions.
Polymarket US grew 54% month on month to $5 billion, its strongest month since the iOS waitlist was removed in May. The international platform fell 26% to $7.9 billion. Combined, the two processed $12.9 billion, down from roughly $14 billion in June.
The most plausible reading is migration. Rutgers statistician Harry Crane estimated, on blockchain wallet analysis, that roughly 30% of the offshore platform's volume originated from US linked accounts over the twelve months to 30 April. When a licensed domestic venue became properly accessible, a meaningful share of that demand moved onto it.
Kalshi's July performance points the same direction. Its share rose to 74.5% behind a CFTC exchange license that no competitor in the US holds on equivalent terms.
For operators outside the US the read is direct. Users do not have a structural preference for offshore venues. They use them when nothing licensed and local is available, and they leave when something is.
People think offshore volume is loyalty. But it is actually the absence of a licensed local alternative, and users move the moment one exists.
The platforms that defined July
Three distinct strategies ran in parallel this month: acquire a licensed venue, build one, or wire the category into infrastructure that already exists. None of the major movers started from zero.
Robinhood
$156M quarterly revenue · routing to its own venuePrediction markets became the fastest growing line in the company, earning more than crypto and equities in Q2. Robinhood is also steadily routing flow to Rothera, the venue it launched with Susquehanna in June. Its orders now represent 17.5% of Kalshi's volume, down from close to 50% a year ago.
IG Group
Acquired Underdog · up to $1.3BA listed CFD broker bought into the category rather than building. Underdog is the third largest US venue by regulated notional volume, live across 30 states, having launched prediction markets only in September 2025. IG expects to more than double its US revenue on the back of it.
Kalshi
74.5% share · $37.7B in JulyExtended its lead materially, with a $256 million single day crypto volume record on 31 July. Also the most exposed to distributor concentration, given Robinhood is steadily moving flow to a venue it part owns.
Polymarket
US arm +54% · international −26%The clearest natural experiment the category has produced. Two venues, one product, one regulated and domestic, one offshore. Volume moved decisively toward the regulated one once access opened.
Hyperliquid
HIP-4 · permissionless deploymentShipped permissionless market creation. Stake HYPE, deploy a market, retain a share of the fees. The first credible attempt at treating market creation as an open primitive rather than an exchange function.
Cboe, Jump
Institutional plumbingCboe filed to list binary options on corporate earnings metrics. Jump Trading doubled its prediction markets team. Professional flow is being connected.
Every significant entrant this month either bought a license, built a venue it controls, or connected the category to distribution it already owned. The infrastructure question is settled. The remaining question is who holds the license and the customer.
Robinhood put a public number on the margin
Robinhood's Q2 earnings are the most consequential disclosure the category has produced. Prediction markets generated $156 million in revenue, against $100 million for crypto and $129 million for equities. Only options earned more.
This is the first quarter in which prediction markets out earned both crypto and stocks at a major retail brokerage. Revenue is up more than tenfold year on year. Event contract volume reached 13.6 billion in Q2, up 55% from 8.8 billion in Q1. The line now accounts for roughly 20% of Robinhood's total trading revenue, on a product that was a pilot eighteen months ago. Total Q2 revenue was $1.31 billion, up 32% year on year and ahead of consensus.
The figure that matters for operators is revenue per unit of volume. Analyst Sheel Mohnot estimates that per $1,000 traded, Robinhood earns roughly $0.13 on equities, $1.47 on options, $2.50 on crypto and $23.00 on prediction markets. Those are modelled estimates rather than reported figures, but Robinhood's own revenue split is consistent with the direction.
A product line that did not exist in this form two years ago now earns more than the equities business it was built alongside.
The post-World Cup retention test
Prediction markets accounted for 27% of all US legal sports betting volume during the tournament, up from 9% in January. Kalshi's market on the Spain versus Argentina final drew roughly $1.9 billion on a single match. Polymarket's World Cup winner contract attracted around $4 billion. Chainalysis recorded $20 billion in on chain prediction market volume tied to the tournament since January, with close to 400,000 wallets participating and under 1% showing illicit ties.
Then it stopped. Open interest across the three venues fell roughly 40%, from around $2 billion at the start of July to $1.2 billion by the end. Kalshi's own open interest went from a $1.4 billion peak to roughly $788 million. That is contract settlement rather than collapse, but it draws a clean line between a tournament and a run rate.
The second half of the year is a retention question. The venues that convert World Cup traders into recurring users across Fed decisions, earnings prints and the NFL season will separate from those that captured the spike and lost it.
For operators, the relevant read is that a major sporting event is one of the cheapest customer acquisition mechanisms available, and the product has to be live before it starts.
Build or buy now has a public price
IG Group paid up to $1.3 billion for Underdog: roughly $1.1 billion upfront enterprise value plus a $200 million earnout tied to 2026 performance. Underdog posted net revenue of around $466 million for the twelve months to 30 June, up 21% year on year, with Q2 EBITDA of roughly $46 million.
What IG bought was a vertically integrated license stack. Futures commission merchant, designated contract market and derivatives clearing organization, plus live distribution across 30 states. That stack is the asset. IG acquired the full trade lifecycle under CFTC oversight rather than assembling it, and expects the deal to more than double its US revenue and grow US monthly active customers more than tenfold.
For operators outside the US the equation differs in one respect that matters. The regulatory perimeter that made Underdog worth $1.3 billion is a US perimeter. In most other markets the equivalent local licenses, payment rails, currency support and customer trust are already held by incumbent brokers and exchanges. The most expensive component is the one they already own.
What this means if you run a trading platform
June answered whether the demand exists. July answered two harder questions: whether it monetizes, and where the volume settles once users have a genuine choice.
On monetization, Robinhood removed the ambiguity. Prediction markets out earned both equities and crypto in a single quarter, on materially thinner volume. IG Group then put a $1.3 billion valuation on a licensed venue with 30 state coverage and a year of operating history.
On where volume settles, Polymarket ran the experiment in public. One product, two venues, one regulated and domestic and one offshore. Volume moved 54% toward the licensed venue and 26% away from the offshore one in a single month. Kalshi's rise to 74.5% share behind a CFTC license points the same way.
The components that determine that outcome are local licenses, local payment rails, local currency support and existing customer trust. Those are the slowest and most expensive parts to build, and outside the US they are already held by incumbent brokers and exchanges rather than by the platforms currently taking the volume.
Your users already speculate on macro events, sports outcomes and price movements. Prediction markets formalize that behavior into a product with materially better revenue per unit of volume than the instruments currently offered to them, and without a new acquisition budget. The technology is the deployable part. The license is not, and you already have it.
Key takeaways
Add prediction markets to your platform.
Shift provides the operator infrastructure. Your brand, your users, your revenue. Live in weeks.
July volume, share and open interest data from The Block's volume dashboard, published 3 August 2026. June baseline revised upward from $44.8 billion to $46.95 billion as trading data completed.
Robinhood figures from the Q2 2026 earnings release. Take rate figures per $1,000 traded are modelled estimates from analyst Sheel Mohnot, not reported by Robinhood.
IG Group and Underdog figures from IG Group's announcement.
World Cup on chain volume, wallet participation and illicit activity share from Chainalysis, 31 July 2026. US linked share of offshore Polymarket volume estimated by Harry Crane, Rutgers University, twelve months to 30 April 2026.
Kalshi cumulative trade count reported by KalshiData, a third party tracker, and not confirmed by Kalshi directly. Not used as a headline figure in this edition.
All volume figures measure taker notional volume rather than platform revenue or customer deposits.