In July 2026 the largest prediction market venues reported their best month on record. Combined volume across Polymarket, Polymarket US and Kalshi reached $50.59 billion, an all time high, according to The Block. In the same month, the money sitting in positions nobody had closed went the other way. Covers reported that Kalshi's open interest fell from a peak near $1.4 billion in early July to roughly $788 million by the end of the month, while the Polymarket platforms slipped from about $550 million to $422 million.
Both statements are accurate and they are not in conflict. They count different things. One counts activity that has already finished, the other counts exposure that is still live. A category that leads with the first number and almost never publishes the second is not lying to anyone. It is choosing which question its readers are equipped to ask.
July set a record and most of it was one tournament
The 2026 FIFA World Cup ran from 11 June to 19 July, and it sat across both of the record months. The Block put Kalshi's July volume at $37.7 billion, up 14 percent on June, while Polymarket's international venue fell 26 percent to $7.9 billion and Polymarket US rose 54 percent to $5 billion. Individual fixtures carried remarkable weight. The market on the Spain against Argentina final drew close to $1.9 billion on Kalshi, and Polymarket's tournament winner market took around $4 billion.
Concentration is not a new discovery about this category. The Pew Research Center, in a May 2026 analysis by Kaitlyn Radde, found that sports made up 80 percent of Kalshi's volume and 39 percent of Polymarket's across the period from July 2024 to April 2026. Sporting Crypto's quarterly review put Kalshi at 86 percent sports for the second quarter of 2026. We have written before about why sports is simultaneously the growth engine and the weakest legal flank. What the summer added was scale, and a demonstration that a single fixture can move more than many venues move in a quarter.
The same month has two volume numbers and both are honest
Ask what Kalshi traded in July 2026 and the answer is either $37.7 billion or $12.37 billion, depending on which series you happen to read. The Block reports notional volume. The Defiant, working from DefiLlama's prediction market category, reports cash volume, and puts Kalshi at $12.37 billion for July against $3.68 billion for Polymarket.
The gap is structural rather than an error, and Pew states the convention plainly in its methodology: in notional taker volume, each contract is counted at its $1 notional value rather than at the price somebody actually paid. A contract bought at 30 cents adds a full dollar to the tally. Notional measures the largest sum the contracts could ever pay out. Cash measures what was put at risk to get there. Divide one published July series by the other and Kalshi's ratio lands near three to one.
Combination contracts are where the two numbers come apart
Sporting Crypto's second quarter review, published on 10 August 2026, does the arithmetic that explains the spread. On ordinary single contracts the ratio of notional to handle sits close to two to one, because the average contract changes hands somewhere near the middle of the price range. Multi leg combinations behave very differently, because probabilities multiply as legs are added. The review puts Kalshi's combination activity at roughly $20 billion of notional out of $65.7 billion for the quarter, on $1.67 billion of handle, which it works out to an average combined probability of about 8.4 percent. Close to a third of the headline number came from contracts where something in the region of eight cents of stake booked a full dollar of notional.
Its author is careful about this and so are we. That is not fabricated volume. Somebody took the other side, the contracts are genuine and the risk is genuine. But a dollar of combination notional and a dollar of single contract notional say different things about how much capital, conviction and information passed through the venue, and summing them into one line erases the distinction. It is a close cousin of the point the CFTC made in its own vocabulary on 12 August 2026, when the Division of Market Oversight issued Staff Letter 26-23 on the self certification of incentive programs, which we covered in our piece on bought volume and demand.
Open interest answers the question volume cannot
Open interest counts contracts still outstanding at the close of the day, meaning positions that nobody has unwound. Volume counts what changed hands. The two move independently, and the distance between them is the distance between attention and commitment. A market can print enormous volume from traders passing the same exposure back and forth across an afternoon and finish with almost nothing open. A market can equally sit quiet with substantial open interest because participants formed a view and mean to hold it through to resolution.
Put the July figures side by side. Combined volume of $50.59 billion for the month, and combined open interest at the end of it somewhere near $1.2 billion on the Covers numbers. That is roughly forty dollars of reported trading for every dollar still committed when the month closed. The ratio is not damning by itself, since short dated contracts written on a tournament are supposed to resolve and clear. It is simply information the headline does not carry. Our own ten metric framework from March 2026 left open interest out, and having spent a summer watching the two series diverge, we think that was a gap in it.
The decay arrived on schedule
What happened after 19 July is the part worth sitting with. The Defiant, again on DefiLlama figures, reported that through 17 August 2026 Kalshi had done $5.73 billion in cash volume, a pace of about $10.45 billion for the full month, against $12.37 billion in July. Polymarket had done $1.18 billion, pacing near $2.14 billion, against $4.29 billion in June. Worldwide Google Trends interest in prediction markets stood at 17 in the week of 9 to 15 August, measured against an index value of 100 in the week the tournament opened, a fall of 83 percent.
None of that is a crisis, and we would not write it as one. A venue whose product is event contracts should expect its activity to track events, and a quiet August after a World Cup is the system working as designed. The risk is in what gets built on top of the peak. Valuations, hiring plans, listing ambitions and arguments to regulators about market significance all tend to lean on the largest available figure, and that figure was assembled out of a tournament with an end date. We have argued that the category built its speed layer faster than its trust layer. How a venue reports itself belongs to the trust layer.
The number nobody quotes is already required by statute
Here is the part that tends to surprise people. For a designated contract market, publishing open interest is not a courtesy extended to curious readers. It is a statutory duty. Core Principle 8 of the Commodity Exchange Act, at 7 U.S.C. § 7(d)(8), provides that a board of trade "shall make public daily information on settlement prices, volume, open interest, and opening and closing ranges for actively traded contracts on the contract market". The Commission's implementing rule at 17 CFR § 38.450 repeats that sentence, and Core Principle 7 at 17 CFR § 38.401 sits alongside it requiring accurate general information about contract terms and trading mechanics.
That duty binds Kalshi. Since the CFTC approved an amended order of designation for QCX LLC, operating as Polymarket US, it binds that venue as well. It does not reach Polymarket's international platform, which is one more reason the two halves of that business do not belong in a single undifferentiated volume line. The category, in other words, already computes and publishes contract by contract the number that would put its monthly headline into proportion. What it does not do is lead with it, or carry it into the monthly summary where the headline lives.
Publishing the second number costs almost nothing
None of this asks anyone to stop reporting volume, and nothing here treats a large number as evidence of bad behaviour. The ask is small and largely clerical. When a venue or a data provider states a monthly figure, four things can travel with it at negligible cost.
- Which measure produced it. Notional or cash. Pew names its measure in the methodology note, and anyone publishing a figure can do the same in one line.
- How much came from combinations. If close to a third of a quarter's notional arrives from multi leg contracts priced in single digit cents, a reader needs that before comparing the figure to anything else.
- Open interest at the close of the period. Already published daily by the regulated venues under Core Principle 8. Carrying the month end value into the monthly summary is a copy and paste.
- Concentration. The share of the period that came from the single largest event, and from the largest category. July would have looked very different described that way.
Directive 02 of our manifest asks that fees, spreads, settlement sources, custody arrangements and conflicts of interest be stated in plain language on the surface where the trade happens. A monthly volume release is a different surface from an order ticket, but the principle carries over without strain. A number is only transparent when the reader can tell what it counted.
Reading a headline figure without being misled by it
For anyone weighing whether a venue deserves their attention, the useful habit is to stop treating a single number as a summary of anything. Ask which measure produced it. Compare a venue against its own history rather than against a rival quoted on a different methodology, because the ratio of notional to cash moves with product mix and a venue carrying more combinations will look larger on notional alone. Read open interest next to turnover, because a venue that empties out completely after every event is a structurally different business from one that carries positions between them.
That is not advice about what to trade or where, which is not something we offer. It is closer to what we set out about the gap between headline liquidity and the liquidity you can actually execute against. A large figure on a landing page and a market you can exit at a fair price are two separate claims, and as of August 2026 only one of them is routinely being made.
The number you lead with says what you think matters
A category that wants to be treated as market infrastructure will be judged in part on how it describes itself. As of August 2026 the description that travels is a single monthly figure, produced on a convention that inflates it relative to cash at risk, dominated by one sport, and unaccompanied by the measure that would show how much of it stayed. Nobody is deceiving anyone on purpose. Notional has a long and legitimate history in derivatives markets, the convention was inherited rather than invented, and the data providers describe their methods openly when asked.
The inherited convention is simply doing work it was never designed for. In a market whose contracts are short, whose prices often sit a long way from the middle of the range, and whose activity clusters on tournaments, notional volume describes the health of a venue less well than almost any other figure the venue already computes. The venues that come through the next quiet month looking solid will be the ones that were publishing the harder numbers during the loud one. If your firm wants to hold itself to that in public, the commitment is open to sign.