Prediction Markets Report August 2026: Volume, Open Interest and Platform Rankings

August 2026 edition. Sports now accounts for 70% of everything traded on Kalshi and Polymarket. Politics and elections take under 10% of that volume, yet hold 45% of all the money currently at risk. Two platforms run 95% of the market, and every challenger to them has failed.

This report tracks prediction market activity across 11 platforms using data sourced from Artemis. Rankings cover trading volume, open interest and trade counts separately, with the full dataset refreshed every month. Category and fee analysis covers Kalshi and Polymarket only, which together are 94.6% of event contract volume. Platform rankings and fee figures exclude Kalshi crypto perpetual futures and Nadex binary options, neither of which are event contracts, while category shares follow Artemis’s own classification and include them.

Key Findings

All figures below are sourced from Artemis unless otherwise noted. Category, fee and turnover figures cover Kalshi and Polymarket only, which together are 94.6% of event contract volume. Data covers 1 to 31 August 2026, with trailing twelve month figures running from 1 September 2025.

These are the headline numbers. The sections below break down the data behind each one: platform rankings and market share, the turnover gap between volume and open interest, what people actually trade, how prediction markets compare to betting exchanges and sportsbooks, platform fee economics, and who is winning and losing on these platforms.

Prediction Market Volume Rankings

The rankings below are based on trading volume as reported by Artemis. Volume is notional, meaning it counts both sides of every matched trade, and it includes market-maker activity, arbitrage and positions opened and closed before resolution. It is not a measure of money staked, and it is not comparable to sportsbook handle.

Every platform declined in August. The World Cup finished on 19 July, so August is the first full month without it and the figures below measure the absence of a single tournament rather than any structural change.

Open interest tells a different story. It closed August at $1.47bn, broadly where it sat through July. Traders did not take their money off these platforms, they stopped trading it in and out.

Kalshi took 70.4% of all event contract volume in August, and Polymarket a further 24.2%. Only Opinion reached 2%. Myriad and Probable reported no volume at all, and Nadex is excluded throughout as a binary options exchange rather than an event contract venue.

Kalshi held up better than Polymarket through the tournament’s absence, declining 27.1% against 33.6%. That gap is partly product mix: Kalshi runs US league sport year round alongside crypto and economics markets, while a larger share of Polymarket’s summer volume sat in World Cup contracts.

Rothera, the exchange behind Robinhood, fell 65% from $1.69B to $594M, the steepest fall of any platform with meaningful volume. It launched in May and its June and July figures were built almost entirely on World Cup contracts, so August is closer to its underlying run rate than either previous month.

Volume and Open Interest: The Turnover Gap

Volume and open interest measure different things, and the gap between them is the most useful signal in this dataset. Volume is the running total of everything traded. Open interest is the money sitting in unresolved positions right now.

In betting terms, volume is handle and open interest is the money tied up in bets that have not settled. Back a team at 2.0 and lay it at 1.8 to lock in a profit, and you have generated twice the matched volume on a single economic position. Prediction markets work the same way, and some categories do it far more than others.

The turnover ratio below divides median weekly volume by same-week open interest for each category. A reading of 9x means the category traded nine times its outstanding value in an average week.

Sports generates 69.6% of trading volume from 37.8% of the money at risk. Politics and elections do the opposite, producing 9.2% of volume while holding 44.9% of open interest. A political contract is bought and held for months or years because that is when it resolves. A sports contract resolves in three hours and the capital is recycled the same evening.

That is why volume alone is a poor measure of this market. It systematically overstates any category that settles quickly, and sports settles faster than anything else on these platforms.

What Drives Sports Turnover

Four mechanisms account for the gap, and none of them involve more people betting.

Economics sits at 0.54x, meaning the category trades roughly half its outstanding value in a week. Those are contracts on inflation prints and jobs numbers, bought and held to the data release. Science and technology behaves the same way at 0.73x.

This is why headline prediction market volume should not be compared to sportsbook handle without adjustment. A sportsbook counts a bet once when it is struck. A prediction market counts every trade in and out of the same position, and on sports that happens roughly nine times a week.

Political Markets: Where the Money Sits

The turnover table above contains the most counterintuitive number in this report. Politics and elections take 9.2% of trading volume and hold 44.9% of open interest.

Prediction markets are routinely described as having become sports betting platforms. On volume that is accurate. On money actually committed it is close to the opposite, because political contracts resolve in months or years rather than hours, and the capital stays locked in for the duration.

A sports contract is opened and closed the same evening. A contract on the 2028 presidential election has been open since 2024 and will not settle for another two years.

The Largest Political Markets

The table below covers the largest political markets on Polymarket by traded volume, with implied probabilities as at 31 August 2026. Polymarket publishes this data openly. Kalshi’s terms restrict commercial use of its market data, so Kalshi political markets are not included.

The 2028 cycle is the whole story. Two markets on a race more than two years away account for $1.69bn between them, roughly a hundred times the combined volume of every market on the current administration’s tenure.

That gap is worth sitting with. Markets on near-term political outcomes attract far more media attention than money. Markets on a distant election attract far more money than attention.

Ethiopia at $283M is the other surprise, and a reminder that Polymarket is a global venue rather than a US one. It is the third largest political market on the platform.

The Trump Exit Markets

A cluster of contracts covers whether Donald Trump leaves office before the end of his term. They are frequently cited and comparatively small.

The main contract has traded since November 2025 and its implied probability has fallen from roughly 20% to 6%, with the sharpest decline between May and July 2026.

What that price represents is worth stating plainly. It is the rate at which traders were willing to take the other side, not a forecast, and on a market of this size a relatively small number of participants set it. For comparison, the 2028 nominee market carries roughly ninety times the volume.

Why Political Markets Hold Capital

Three structural reasons explain the open interest concentration.

Political contracts have long resolution horizons, so money committed in 2026 to a 2028 outcome sits there for two years. Sports capital turns over the same week.

They also attract position holders rather than traders. Someone who believes a candidate is underpriced buys and waits, because there is no in-play equivalent and little to react to between now and polling day.

And they are the markets institutions and hedgers use. A contract on an election outcome has genuine hedging value for anyone whose business depends on the result, which is not true of a contract on a Tuesday night basketball game.

Prediction Markets Are Betting Exchanges With a Federal Licence

Most coverage treats prediction markets as something new. The trading model is not. Betfair launched peer-to-peer betting in 2000, and Smarkets, Matchbook and Betdaq followed over the next two decades.

What is new is the wrapper. Kalshi and Polymarket US operate as CFTC-designated contract markets, which places them under federal derivatives regulation rather than state gambling law. The second change is scope, because an exchange that once listed only sport now lists elections, inflation prints and Bitcoin prices alongside it.

The Real Break Is How Winners Are Treated

The structural difference between these platforms and the exchanges that came before them is not the technology. It is what happens when you win.

Betfair’s Exchange does not close winning accounts. It charges them. The Premium Charge, introduced in 2008, was replaced in January 2025 by the Expert Fee, which takes 20% of gross profit above £25,000 over a rolling year and 40% above £100,000, per Betfair’s published charges.

Kalshi, Polymarket and Novig all market on not limiting winning traders. That pitch is aimed at the incumbent exchange as much as at sportsbooks, and it is the clearest competitive difference between the two generations.

Three Different Ways to Charge for a Bet

The cost comparison between these models is regularly reported as a single percentage. It cannot be, because the three models charge on entirely different bases.

A sportsbook pricing a two-way market at -110 on both sides is quoting implied probabilities that sum to 104.76%. That 4.76% is the house edge, and it is charged whether the bet wins or loses because it is inside the price. US sportsbooks held 10.16% of handle across 2025, a figure inflated by parlays carrying far higher margin than straight bets.

A prediction market quotes YES and NO summing to roughly 100 cents and charges a visible fee on top. On a single position held to resolution, Kalshi costs roughly a quarter of a sportsbook’s two-way margin and an eighth of its blended hold.

That advantage narrows for anyone who trades in and out, because the fee applies to every matched trade rather than once per bet. At the sports turnover rates in the previous section, a position recycled repeatedly pays the fee repeatedly.

The Substitution Is Now Measurable

US legal sportsbook handle reached $166.94bn in 2025, up 11.0%, generating $16.96bn in revenue, according to the American Gaming Association. That is the benchmark this market is measured against, and it is starting to move.

H2 Gambling Capital estimated prediction markets accounted for roughly 27% of all legal US sports betting volume during the World Cup, up from about 9% at the start of 2026. The American Gaming Association puts diverted state tax revenue above $500m and reported that May 2026 handle fell year on year.

The AGA is not a neutral observer here, and its members lose when this volume moves. But the direction is corroborated by the handle data itself, and by the scale of what these platforms now process: $170.5bn of sports volume over the twelve months to August 2026, against $166.94bn of sportsbook handle across calendar 2025.

Those two figures cover different periods and are not equivalent measures. Notional volume counts both sides and every round trip, while handle counts a bet once. The comparison shows order of magnitude, not substitution one for one, and this report tracks the gap between them each month rather than treating them as the same number.

The convergence runs both ways. In 2026 Roobet became the first crypto casino to launch prediction markets, which is the same trade being offered from the opposite direction.

For where the rest of the gambling dollar is going, our Online Poker Traffic Report tracks 36 networks monthly, and our US Gambling Survey covers how Americans actually bet.

Kalshi and Polymarket Run Different Businesses

The two platforms look similar in aggregate and are not. They share a trading model and almost nothing else in what they list, who trades on it, or how they make money.

Politics is 93% Polymarket. Combination contracts are 100% Kalshi. Sports is the only category both platforms compete in seriously, and Kalshi holds roughly three quarters of it.

Kalshi’s Exotics Are Parlays

Kalshi’s largest single market is a combination contract, and the category did not exist a year ago. These are multi-leg products that pay out only if every leg resolves correctly, which is the definition of a parlay.

The structure is genuinely different from a sportsbook parlay. Kalshi cannot take the other side of a customer trade, because CFTC rules prohibit a designated contract market from acting as counterparty. Combinations are priced through a request-for-quote system in which market makers compete to fill the order.

The economics are less different. The counterparty is always a professional firm, with Susquehanna as the flagship market maker alongside Kalshi’s own affiliated trading arm. The app offers no way to take the other side of someone else’s combination, which reserves that position for firms trading through the API.

Combination contracts also explain a quirk in the turnover data. They trade at an average implied probability of around 9%, so a dollar of notional represents roughly nine cents of committed capital, which produces turnover ratios far above any other category.

Kalshi Is Becoming a Derivatives Exchange

On 3 June 2026 Kalshi listed the first CFTC-regulated perpetual futures in US history, starting with Bitcoin and expanding to thirteen tokens. Those contracts have taken 33.1% of Kalshi’s total volume since launch, and $15.34bn in August alone.

Perpetual futures are not event contracts. They are leveraged crypto derivatives with no resolution date, and they are excluded from the platform rankings and fee figures in this report for that reason.

The direction matters more than the exclusion. Kalshi is no longer only a prediction market, and its sports share is falling partly because a different product is growing underneath it.

Polymarket Runs Two Separate Businesses

Polymarket International settles on a public blockchain and is not CFTC-regulated. Polymarket US operates through a designated contract market acquired in 2025 and serves American traders under federal oversight.

The US entity has taken $15.48bn since February 2026. That is more than Robinhood’s exchange, Predict.fun, Limitless and Probable have taken in their entire existence combined, built in seven months from a standing start.

The two entities charge differently. Since February, US traders have paid 0.692% of notional against 0.335% internationally, so the regulated version costs roughly twice as much to trade on. In August the two converged, both landing near 0.75%.

Platform Fee Economics

Prediction markets charge an explicit fee rather than building margin into the price, which is the same model a poker room uses when it takes rake from the pot. The house is not betting against you, it is charging you to play.

The rates are public and the volumes are tracked, so unlike casino gross gaming revenue, this is one corner of gambling where operator economics can be calculated rather than estimated.

Kalshi has taken $2.04bn in fees over twelve months, more than six times Polymarket’s $311.4M across both entities. Its Q2 take rate of 1.318% on event contracts is three times Polymarket’s blended 0.445%.

The gap is wider than the headline suggests once you separate Polymarket’s two businesses. Since February, American traders have paid 0.692% while international traders paid 0.335%, so the regulated entity costs roughly twice as much to trade on. In August the two converged, both landing near 0.75%.

Kalshi’s fee formula explains why the rate moves. Its published fee schedule sets the taker fee at 0.07 multiplied by the number of contracts, the price, and one minus the price, which peaks at 1.75% when a contract trades at 50 cents and falls toward both extremes. A contract at 90 cents costs about 0.63%.

That structure charges most where uncertainty is highest. A coin-flip market is the most expensive thing to trade on Kalshi, and a near-certainty is among the cheapest.

What the Fees Buy

An independent analysis of Kalshi trade data estimated its 2025 take rate at just under 1.2%, with roughly 89% of fee revenue coming from sports. Our own calculation from Artemis data puts the trailing twelve month rate at 1.259%, which is close enough to treat both as sound.

That makes Kalshi one of the largest fee-earning gambling operators in the world, taking $2.04bn on $162bn of event contract volume without holding a gambling licence in any jurisdiction. Our crypto casino market report covers the same question for onchain operators, where gross gaming revenue cannot be observed directly at all.

Who Actually Wins on These Platforms

Prediction markets are peer-to-peer, so there is no house taking the other side. That does not mean there is nobody on the other side, and the data on who profits is now public.

Bloomberg analysed Kalshi’s trade records and found that retail traders lost a net $294 million on combination contracts during 2026, before fees. The same analysis found combinations carried an average implied probability of around 9%, against 43% for the platform’s other contracts.

Gambling Insider, benchmarking against New Jersey and Maryland regulatory filings, calculated that combination bettors lose roughly 8 to 9 cents per dollar staked. A sportsbook keeps about 19 cents per dollar on parlays and 6 cents on straight bets.

The Blended Figure Flatters It

That 8 to 9 cent figure covers everyone trading combinations, which includes the professional firms taking the other side. Strip those out and the picture changes.

Gambling Insider’s analysis argues the retail taker’s real loss rate runs two to four times the blended headline, and that over one measured fortnight it was worse than a sportsbook. The averaging works the same way a poker room’s overall rake figure would if you included the regulars in the player pool.

There is a structural reason for the imbalance. Kalshi’s app offers no way to take the other side of someone else’s combination, so that position is available only to firms trading through the API, with Susquehanna as flagship market maker alongside Kalshi’s own trading arm.

The Edge Moved, It Did Not Disappear

On a straight two-way market the exchange model is genuinely cheaper than a sportsbook, for the reasons set out earlier in this report. A 1.29% fee against a 4.76% overround is not a marginal difference.

On combinations, that advantage narrows or reverses. The platform is not the counterparty, but a trading firm is, and it prices for a living.

Prediction markets removed the house edge from the price and replaced it with an explicit fee. What they did not remove is the gap in skill between a retail trader and a market maker, and on multi-leg products that gap is where the money goes.

The exchange model is not new. Betfair has matched bettors against each other since 2000, and anyone who traded on it knows the appeal was never that it was cheap. It was that a winning account did not get closed.

What is new is the licence. A CFTC exchange cannot refuse a profitable trader the way a sportsbook can, and it cannot take the other side. That is a genuine structural improvement for anyone who wins.

It is not the same as removing the house edge. On a straight two-way market you keep more of your expectation than at any sportsbook. On a multi-leg combination you are trading against a firm that prices for a living, and the numbers say retail is losing that trade.

A Consolidating Duopoly

The headline story is growth. Underneath it, this market is narrowing fast.

Kalshi and Polymarket took 63.0% of event contract volume in December 2025. By August 2026 they took 94.6%. Every platform that has tried to hold third place has lost it inside a few months.

Third place has changed hands three times in ten months. Opinion held it from November through March, Predict.fun took it in April and May, Robinhood’s exchange took it in June and July, and Opinion holds it again in August on $740M against the $8.08B it did in January.

Probable went from $2.15B in February to nothing by May. It is the only platform in the tracked set to have stopped reporting entirely.

Robinhood Could Not Hold Third Place

Rothera is the clearest test of whether distribution and capital are enough to break into this market. It is a CFTC-regulated exchange and clearinghouse built with Susquehanna, and it routes event contract trades from one of the largest retail brokerages in the United States.

It launched in May, reached $2.09B in June, and fell to $594M by August. Eighty-two percent of its entire volume came from World Cup contracts.

That pattern is the important one. Robinhood did not fail to attract traders, it attracted them for one tournament and did not keep them afterwards. A platform whose volume evaporates when a single event ends has not built a market, it has rented an audience.

The tail keeps failing for a structural reason. Traders go where the spreads are tightest, spreads are tightest where the volume already is, and liquidity compounds toward whoever has most of it. Every challenger in the table above launched on an event-driven spike and then lost its traders to the two platforms with deeper books.

Biggest Event Benchmark: The 2026 World Cup

Each edition of this report tracks the largest single event in prediction market history. The 2026 World Cup currently holds that position, and it will do until a Super Bowl or a US election displaces it.

Three platforms listed World Cup contracts and took $29.8bn between them across six weeks. The peak week ending 5 July drew $6.55bn, and open interest on tournament contracts peaked at $726.3M in late June.

The last column is the one that matters. Kalshi and Polymarket both had a strong tournament, but neither depended on it. Rothera took 82% of its entire volume from a single competition, which is why it collapsed the moment the final whistle went.

For scale against the sector this site already tracks, the crypto casino market processed $46bn in deposits across 41 operators over twelve months. One football tournament, on three platforms, reached roughly two thirds of that in six weeks.

Why Tournament Volume Is Not Tournament Handle

World Cup contracts turned over roughly nine times their open interest in an average tournament week, in line with the sports figure across the rest of this report. The $29.8bn is therefore a trading figure, not a measure of money staked on the outcome.

Artemis also publishes a per-match table for the tournament, and it does not reconcile with the platform totals. Match-level contracts account for $4.96bn against the $29.8bn total, and coverage differs sharply by platform, capturing 11.7% of Kalshi’s tournament volume against 23.1% of Polymarket’s.

That gap means per-match rankings cannot be published from this dataset without misleading readers about which platform led on individual fixtures. Most tournament volume sat in outright winner, group and progression markets rather than single match outcomes.

Tournament contracts were 25.2% of all prediction market volume during July. A market that concentrates a quarter of its activity into one competition is still event-led rather than habit-led, and that is the distinction between a market growing and a market becoming durable.

The next benchmark event will test whether the traders acquired in June and July are still trading.

Regulation: What Would Move These Numbers

Every figure in this report depends on prediction markets remaining legal to operate at their current scale. That question is unsettled, and it is being fought in three places at once.

The CFTC Has Moved Toward Permission

Under chairman Michael Selig, confirmed in December 2025, the Commodity Futures Trading Commission withdrew the restrictive rule proposed in 2024. A new proposed rule published in June 2026 sets out a test under which broad-outcome sports contracts, covering final scores, point differentials and tournament progression, are largely permitted.

The scale of the shift is visible in the listings. Event contract filings went from roughly five a year between 2006 and 2020 to about 1,600 in 2025.

In April 2026 the CFTC went further and sued Arizona, Connecticut and Illinois to block state regulation of prediction markets, asserting exclusive federal jurisdiction.

The Courts Are Split

The most important ruling so far went Kalshi’s way. In April 2026 the Third Circuit affirmed an injunction barring New Jersey from enforcing its gambling laws against the platform, holding that sports event contracts are swaps under the Commodity Exchange Act and therefore subject to exclusive CFTC jurisdiction.

Other courts have gone the other way. A Massachusetts state court ruled in January 2026 that Kalshi’s sports contracts are subject to state gaming law, calling the preemption argument overly broad. Nevada secured injunctive relief at district level, and the Ninth Circuit heard consolidated argument in April 2026 with a decision still pending.

Arizona filed 20 misdemeanour criminal charges, the first criminal action against Kalshi anywhere. New York’s attorney general and governor sued over unlicensed sports betting, seeking penalties tied to sports revenue.

Tribal Operators Have Their Own Case

Three California tribes sued Kalshi and Robinhood in July 2025 under the Indian Gaming Regulatory Act. A federal court denied a preliminary injunction, holding that IGRA does not reach third-party platforms, and the Ninth Circuit heard the appeal in July 2026.

More than 60 tribes filed amicus briefs and the Indian Gaming Association launched a litigation fund. Wisconsin tribes have filed separately.

Why the Fight Is About Sports Specifically

Nothing in the litigation concerns inflation contracts or election markets in any serious way. The dispute is about sport, because sport is 69.6% of trading volume and because it is the category that overlaps directly with a licensed, taxed industry.

The American Gaming Association’s position is that these contracts are sports betting under a different name, and that states are losing tax revenue they would otherwise collect. Kalshi’s position is that a federally regulated exchange is not a sportsbook, and that it never takes the other side of a customer trade.

Both are describing the same activity accurately. The disagreement is about which regulatory category it belongs in, and that is a question courts will answer rather than data.

Frequently Asked Questions

Methodology and Data Sources

This report combines primary data from Artemis with secondary figures from industry bodies, regulatory filings and independent analysts. The sections below explain what each source measures, how it is used, and where the limits are.

Primary Data: Artemis

All volume, open interest, trade count, category and fee figures come from Artemis, which tracks 11 prediction market platforms. Data is reported weekly, dated to the Sunday ending each week, and Kalshi and Polymarket were 94.6% of event contract volume in August 2026, so a two-platform analysis describes almost the entire market.

Two Reporting Tiers

This report works at two levels, and the distinction matters when reading any figure.

Platform rankings, market share and open interest cover all 11 tracked platforms. Category analysis, turnover ratios and fee economics cover Kalshi and Polymarket only, because Artemis does not publish category or fee data for the remaining nine.

That is not a significant limitation at present. Kalshi and Polymarket were 94.6% of event contract volume in August 2026, so a two-platform analysis describes almost the entire market. It would become a limitation if the tail recovered.

Derived Metrics

Two figures in this report are calculated by VIP-Grinders rather than reported by Artemis.

Turnover ratio is median weekly trading volume divided by same-week open interest, per category, over the trailing twelve months. The median is used rather than the mean because open interest grew roughly threefold over the period, and dividing a full year of volume by a single end-point would understate the result.

Substitution comparison sets prediction market sports volume against US sportsbook handle as published by the American Gaming Association. The two are not equivalent measures and the report says so wherever the comparison appears.

What We Exclude and Why

Platform rankings and fee figures exclude two products that Artemis tracks but that are not event contracts.

Kalshi crypto perpetual futures, listed from 3 June 2026, are leveraged derivatives with no resolution date. They reached 30.8% of Kalshi’s volume within eleven weeks and would materially distort both the rankings and the take rate.

Nadex is a binary options exchange trading currency pairs, indices and commodities rather than event outcomes.

Category shares are the exception. Those follow Artemis’s own classification, which includes perpetual futures inside Kalshi’s Exotics category, and cannot be recalculated without them.

Secondary Sources

Limitations

Editorial Independence

VIP-Grinders has no commercial relationship with Artemis and no affiliate relationship with any prediction market platform covered in this report. This page carries no affiliate links. No platform pays for inclusion, ranking position, or how it is described.

Update Schedule

This report is part of the VIP-Grinders research programme and is updated monthly. Each edition is published on the first of the month and covers the whole of the previous calendar month.

The next edition covers September 2026 and will be published on 1 October 2026. Our testing and verification standards are set out in how we test.

If you are a journalist, researcher or analyst using data from this report, contact João Mourato at VIP-Grinders for verification, methodology detail, or access to the underlying tables.