Kalshi Just Had Its Biggest Week Ever. Nearly Two-Thirds of It Was Parlays, and the Headline Number Is 4.3 Times the Cash

Kalshi booked $15.27 billion of contract volume in the week of September 14 to 20. It was the exchange's largest week of 2026, it included the first $3 billion day in the company's history — $3.17 billion last Sunday, against $2.81 billion on the Saturday — and it put Kalshi at 92.9 per cent of the combined contract volume transacted by itself and Polymarket Global.

In the same report, from the same data, sits a second number: $3.53 billion. That is the execution-price figure — the dollars that actually changed hands.

The two numbers describe the same week. The first is 4.33 times the second. And the gap is not evenly spread across the exchange; it is concentrated almost entirely in the category that made the week a record.

Where the volume came from

The category split for the week runs: combos 62.32 per cent, sports 21.21 per cent, crypto 12.78 per cent, financials 1.33 per cent.

Combos are Kalshi's multi-leg product. A parlay, in the older vocabulary. Sixty-two per cent of $15.27 billion is roughly $9.5 billion, which means the single largest line item on the exchange's biggest week was not the presidential-odds market that made its reputation, not the Fed contracts, not even standalone NFL game winners at 21 per cent. It was combination tickets.

This is not an aberration of one week. Next Event Horizon's count of the opening football weekend — September 12 and 13, $4.89 billion across two days — put combos at 61 per cent, about $2.97 billion, against $1.43 billion of standalone sports trading. Inside that weekend were 2,159,748 separate combinations containing 19,485,136 legs. Just under half of those legs, 9,149,979, were football. The median ticket carried six legs. Forty-seven per cent of parlay volume, and 26 per cent of all combinations, came from tickets of eleven legs or more.

Eleven legs.

Why the denominator matters

Here is the mechanical point, and it is not a contested one. It comes from the same analysts who produced the volume counts, in a methodology note published August 26:

"Multi-leg parlays can trade for as little as a fraction of a cent while still carrying a notional value of $1, greatly exaggerating the amount of money changing hands in this category."

A Kalshi contract settles at $1 or $0. Volume is counted in contracts, at $1 each. So a bettor who pays half a cent for an eleven-leg longshot — a ticket that is, by construction, enormously unlikely to cash — has spent $0.005 and generated $1.00 of reported contract volume. A 200-fold multiple, on a single ticket, in the category that is 62 per cent of the total.

That is the whole explanation for the gap between $15.27 billion and $3.53 billion. It is not an accounting trick and nobody is hiding it: Kalshi publishes both figures, and the execution-price number sits in the same report as the headline. But the $15.27 billion is the number that travels. It is the one that appears in the aggregator write-ups, the one that gets compared to sportsbook handle, and the one that does work in a fundraising deck.

Sportsbook handle is not a perfect analogue either — a $10 parlay stake is $10 of handle regardless of how long the odds are — but it is at least money the customer put down. The $15.27 billion is not money anybody put down. The $3.53 billion is.

The strongest version of the other side

Notional is a defensible number, and the case for it is better than the dismissals suggest.

A clearinghouse does not care what you paid for a contract. It cares what it owes if the contract settles at $1. Every one of those eleven-leg tickets is a dollar of potential settlement obligation, and the exchange has to be able to cover all of them at once. Measured as exposure — as the thing that determines margin requirements and capital adequacy — notional is not merely defensible, it is the correct figure, and it is the figure futures markets have quoted for a century. Nobody reports crude oil turnover by summing the premiums.

There is a second, better argument. Turnover is what an exchange is for. High volume relative to money at rest is a sign of liquidity, not of fakery, and we made exactly this point yesterday about the ether perpetual book. An exchange with deep, fast, cheap markets should show enormous volume against modest open interest. That is the product working.

Both of those are right. Neither rescues the use the number is actually put to.

The $15.27 billion is not being quoted to describe settlement exposure or to demonstrate market depth. It is being quoted to answer the question "how big is this business, and is it taking the sportsbooks' customers." For that question the execution-price figure is the honest one, and it is 23 per cent of the headline. An industry that wanted to be understood would lead with $3.53 billion and footnote the notional. It leads with the notional.

We should be specific about our own record here: this publication has quoted Kalshi contract-volume figures without that caveat, including in pieces published last Saturday and yesterday. The criticism lands on us.

What the 62 per cent actually means

Set the denominator argument aside for a moment, because the composition finding survives it intact. Whichever number you use, combos are the majority of the exchange's activity.

Kalshi's founding pitch — the one that won the CFTC fight, the one in every profile — was that an exchange is structurally better for the customer than a sportsbook. No house. No vig in the traditional sense. Peer-to-peer pricing, with the exchange taking a small fee for matching. On a two-sided market in a liquid contract, that pitch is substantially true, and it is why we have argued against criminalising these venues.

The combo product is a different animal. We reported on September 19 that Kalshi's fee structure on combos runs at a markup over the single-leg equivalent, which is the same economics a sportsbook runs on a parlay and for the same reason: multi-leg tickets are where margin lives, because the edge compounds across legs and the customer is buying variance rather than price. An eleven-leg ticket is not a person expressing a calibrated view on eleven propositions. It is a lottery ticket.

So the trajectory is this. The exchange that argued it should exist because it is not a sportsbook now generates the majority of its volume from the product that most resembles a sportsbook's most profitable one.

DraftKings, which runs both a sportsbook and a prediction market, offers the comparison. Jason Robins told a Wells Fargo webinar on Tuesday that combo trades are running at roughly 30 per cent of NFL Sunday activity on DraftKings' prediction product, with more than a million customers on it. Thirty per cent at the sportsbook company. Sixty-two per cent at the exchange.

What would change our mind

If Kalshi's execution-price volume grows at the same rate as its contract volume over the rest of the football season, the notional critique loses most of its force — it would mean the combo mix is not being driven by penny longshots but by real money. That is a testable claim and the data to test it is published weekly.

The composition point is harder to walk back. The only thing that would undo it is the single-leg markets growing faster than combos, and right now they are going the other way: standalone sports fell from 29 per cent of the total on the opening football weekend to 21.21 per cent last week, while combos rose from 61 to 62.32.

We do not know how much of the combo volume comes from a small number of high-frequency accounts versus a broad retail base. Kalshi has the account-level data and has not published it, and no outside tracker can separate the two. Until somebody does, the most important question about this record week — whether it represents a million new customers or a concentrated few — is genuinely open, and anyone telling you otherwise is guessing.

Editor's note: TrueEdge builds odds and pricing tools and earns affiliate commissions from licensed sportsbooks, which compete directly with the exchanges discussed here. A piece arguing that the exchanges' headline growth metric overstates them is a piece our commercial interest would incline us toward writing, and readers should weigh it on that basis. Our position that federally licensed exchanges should not be criminalised has not changed. Nor has the fact, noted above, that the same methodological criticism applies to figures we ourselves published on September 19 and September 21.