A Trader Says Kalshi's Crypto Volume Is Fake. Kalshi's Defence Is Better Than the Accusation — and Worse for the Industry

A quantitative analyst who trades under the name Beni, co-founder of a shop called Stealth Neolab, spent last week counting trades in Kalshi's ether perpetual futures book. On Sunday he published what he found: $539 million of volume over twenty-four hours, sitting on top of $3.1 million of open interest. A ratio of roughly 174 to one. Inside that volume, an implausible share of the tickets were for an identical size — $5,500, over and over.

A second analyst, Phin, pulled the tape independently and got numbers in the same place. Across 529,318 ether perp trades totalling $2.217 billion of notional, about $1.230 billion — 55.47 per cent — printed at exactly $5,500. The daily share ranged from 38.32 per cent on September 15 to 65.45 per cent on September 14.

Kalshi's crypto lead, who posts as IcoBeast, answered within hours. His answer is the interesting part of this story, and almost every write-up has treated it as damage control rather than reading it.

What Kalshi actually said

Three claims, and they are not the same claim.

The first is a correction: the chart that started the argument tracked prediction-market share, not perpetual futures, and the two are separate products with separate incentive programmes. That appears to be right.

The second is regulatory: Kalshi's market-maker incentives are filed publicly with the CFTC because they have to be. crypto.news puts the filing at September 2, certified September 16, running to December 31. The structure rebates taker fees down to 0.3 basis points and pays eligible makers a net 0.3 basis points. IcoBeast's framing was that this is more disclosure than an offshore venue offers, where the deals are "in the dark." Also right.

The third is the one that matters, and it is a concession dressed as a rebuttal:

"For example, if a trader buys 100,000 contracts priced at 30 cents, they spend only $30,000 in cash, but the system records $100,000 in volume."

That is Kalshi's crypto lead explaining that the exchange's published volume is the maximum potential payout of the contracts traded, not the money anybody put up. He is describing the convention accurately. Polymarket counts the same way. The convention is defensible — each contract settles at a dollar, so a dollar is the natural unit.

But look at what has just been conceded. The most-cited number in this industry is not a measure of money at risk, and the exchange says so, in public, when pressed.

The accusation is weaker than it looks

We are not going to print that a 174-to-one ratio proves wash trading, because it does not, and the outlets that implied it were reaching.

Perpetual futures are built for turnover. There is no expiry, funding is paid periodically, and the product's entire design encourages traders to open and close inside a session rather than carry. High volume against low open interest is the normal signature of a venue full of intraday traders — you can find similar ratios on venues nobody accuses of anything. Taken alone, the ratio is a description of what a perp is.

The clustering is a different matter. A book where more than half the notional arrives in identical $5,500 parcels is not describing a crowd of independent traders with independent views. It is describing a small number of participants executing a repeated programme. That is compatible with wash trading. It is equally compatible with a market maker mechanically working an incentive schedule — which is legal, filed, and exactly what a 0.3-basis-point maker rebate against a 0.3-basis-point taker fee is designed to produce.

Beni's sharper point is that when the round-trip cost of trading against yourself approaches zero, the return on manufacturing volume stops being negative. That is a structural observation about the fee schedule, not an allegation about a specific trader, and it is the version of his argument that survives scrutiny.

No CFTC enforcement action names Kalshi over any of this. As of today the agency's public filings contain nothing of the kind. Anyone telling you otherwise is inventing it.

The part nobody wants to say out loud

Nine days before this argument started, a different set of numbers pointed at the same structural fact from the opposite direction.

Over the opening football weekend, Kalshi printed back-to-back daily records — roughly $2.43 billion on Saturday September 12 and $2.46 billion on Sunday the 13th, about $4.89 billion combined. Multi-leg combination contracts made up 61 per cent of it, some $2.97 billion. The composition, from Next Event Horizon's tally: 2,159,748 combinations, 19,485,136 legs, a median of six legs per ticket, and 47 per cent of all parlay volume concentrated in tickets carrying eleven legs or more.

An eleven-leg combination is priced in pennies and counted in dollars. That is the mechanism. The same counting convention that turns a 30-cent contract into a dollar of volume turns a long-shot parlay into a very large number very efficiently, and the more legs you add, the more efficiently it works.

So: 61 per cent of the football volume is multi-leg notional, and the crypto book's volume is notional the exchange concedes is not cash. These are not two scandals. They are one accounting convention, showing up in two products, doing what it does.

To its credit, the Next Event Horizon analysis says this itself — volume "is not comparable to sports betting handle." The people closest to the data have been careful. The people quoting the data have not.

Including us

We have run these numbers. On September 16 we wrote about Kalshi's Week 1 vig using the combo figures. On the 19th we wrote about maker fees on parlays. Both pieces treated the volume totals as a real quantity, and both would have been better if they had said, in the first paragraph, that the denominator is notional and the cash figure is unknown and unpublished.

It matters because of what the number is for. Kalshi raised $1 billion at a $22 billion valuation in March and was reported in June to be seeking capital at roughly $40 billion. Aggregate volume across exchanges is the headline in every sector report — DeFi Rate's kickoff tally ran to $14.1 billion across venues. Volume is the metric that says prediction markets are beating sportsbooks. It is quoted in briefs to regulators, in fundraising decks and in our own copy, and the exchange's own crypto lead has now explained that it counts maximum payouts rather than money.

The strongest case against us

Here is the best version of the other side, and it is good.

Notional is the standard unit across derivatives. CME reports notional. Nobody accuses the futures industry of fraud for using the convention that its contracts are denominated in. Kalshi did not invent it, applies it consistently, discloses its incentive programmes to a federal regulator, and answered a public accusation within hours with a technical explanation that checks out. A company running a con does not volunteer the arithmetic that makes the con legible.

That is all true, and it is why this piece is not an accusation of fraud. The complaint is narrower and, we think, harder to dismiss: a unit that is correct inside a derivatives exchange becomes misleading the moment it is carried into a public argument about whether prediction markets are out-competing regulated sportsbooks, where the comparison figure — handle — means cash wagered. Two different quantities are being set against each other as though they were one.

What would settle it

Publish open interest and unique accounts next to volume. Publish cash deposited against contracts traded. Kalshi has the numbers; producing them is a design decision, not a disclosure burden.

And there is a natural experiment already scheduled. The perp incentive programme expires December 31. If the $5,500 clustering thins out when the rebates lapse, the sceptics were describing an incentive artefact and the volume was, in the ordinary sense, real trading that happened for a subsidised reason. If the clustering persists without the subsidy, the explanation everyone has accepted this week was wrong. We will check in January.

We do not know which way that goes. Neither, on the evidence published so far, does anyone else.

Editor's note: TrueEdge builds odds and pricing tools and earns affiliate commissions from licensed sportsbooks, which compete directly with the exchanges discussed here. This piece argues that a metric favouring those exchanges overstates them, which is a conclusion our commercial interest would incline us toward. Weigh it accordingly — and note that the same criticism lands on two of our own recent pieces.