Kalshi Set Two All-Time Records in Two Days. Most of the Money Was in Parlays, and Most of the Number Is Not Handle

Two all-time records in two days. $3.45 billion on Saturday, $3.68 billion on Sunday, and across the eleven venues Aldrin Research tracks, $9.43 billion for the weekend — more than 7 per cent above the one before it. Kalshi took 73.7 per cent of a Sunday that cleared $5 billion in notional across all platforms for the first time. It was the company's fifth $3 billion day and seventeenth $2 billion day since football started.

Now the sentence that should go alongside every one of those numbers. Combo contracts accounted for $3.1 billion of Sunday's trading. Single-event sports accounted for $1.46 billion.

The record was set by parlays.

What "notional" counts

A Kalshi contract settles at $1 or $0. Notional volume counts every contract traded at its $1 face value, regardless of what anybody actually paid for it. Buy a contract at 12 cents and the tape records a dollar. The gap between the figure and the cash is widest exactly where this market is growing fastest, because long-shot legs trade at low prices and multi-leg combos price lower still.

This is not a criticism of the metric. It is the correct metric for an exchange, where the thing being measured is contracts changing hands, and nobody at Kalshi is hiding the definition. It is a criticism of the comparison. When a headline sets $3.68 billion against a sportsbook's handle, it is putting face value next to money wagered, and those are not the same unit. A $100 parlay ticket at FanDuel is $100 of handle. A six-leg Kalshi combo bought for $4 that pays $100 is $100 of notional and $4 of risk.

Next Event Horizon's work on the September 12-13 weekend is the best public look at how much that matters. On a Kalshi weekend of $4.89 billion, combos were $2.97 billion — 61 per cent. Those combos were 2,159,748 separate combinations containing 19,485,136 legs between them. Forty-seven per cent of combo volume sat in what the analysis calls centipedes: eleven legs or longer. The median combo carried six.

Nineteen million legs is the number to hold onto. Each one contributes face value to the tape.

Then do the arithmetic the other way

Here is the comparison we think is actually informative, and it is less flattering to the sportsbooks than the handle version.

Kalshi's 2025 books, as reconstructed by Revenue Memo, show $22.88 to $23.8 billion in nominal volume producing $260 to $263.5 million in fee revenue. That is an effective take of roughly 1.14 to 1.22 per cent. The fee is quadratic — ⌈0.07 × C × P × (1 − P)⌉ — which peaks at even money and collapses at the extremes: a hundred contracts at 50 cents cost $1.75 in taker fees, the same hundred at 10 cents cost 63 cents. More than 89 per cent of that fee revenue came from sports.

Apply the 2025 take rate to Sunday and you get something in the region of $44 million in fees from one day. We should be explicit that this is an estimate carrying real error: the 2025 mix was less parlay-heavy than today's, long-shot legs sit in the cheap part of the fee curve, and we do not know how Kalshi assesses fees on a combo as opposed to its constituent legs. Nobody has published that, and we are not going to pretend otherwise. Sacra's own estimate — roughly $4 billion annualised as of July 2026, on about $30 billion of monthly volume — implies something close to the same take, which is the only independent check available.

Now the regulated comparison. Every licensed online sportsbook in New York — eight of them, in the largest legal market in the country — took $573.4 million in handle in the week ending September 27 and kept $48.6 million of it at an 8.5 per cent hold. FanDuel alone did $221.3 million and kept $20.1 million.

One Kalshi Sunday, estimated conservatively, is in the same range as a full week of gross revenue for every book in New York. And New York's first three weeks of the football season were down 28 per cent in revenue year over year.

Why the parlay share is the whole story

A sportsbook parlay is the most profitable product on the menu because the book prices each leg with a margin and then multiplies the margins together. Hold on a six-leg parlay at a major book is not 8.5 per cent; it runs several times that. It is also, by a distance, the product that draws the most regulatory attention, because the price is opaque to the customer and the appeal is behavioural.

Kalshi's combo is the same bet with a different cost structure. The customer buys into a market rather than against a house, and the venue takes roughly a percentage point rather than embedding a compounding margin in every leg. On the customer's side of the ledger, that is cheaper — materially so, and not as a matter of marketing. It is why combos are 61 per cent of Kalshi's weekend, over 50 per cent of DraftKings' own exchange and 83 per cent of Underdog's — and why, on the same record weekend in September, the one tracked venue without a parlay product did not set a record at all.

So the incumbent books face an unpleasant pair of facts. The competitor is taking share at the exact product that funds them, and it is doing so by charging less for it. DraftKings' response has been to join: Jason Robins has the company market-making on its own exchange, which he describes as unlocking "access to an additional layer of the value chain." DraftKings Predictions did $217.8 million on Sunday, an operator high, up 10.8 per cent in a week. That is a real business and it is also a fraction of the $3.68 billion next to it.

The strongest objection, and what we think of it

The sharpest response to everything above is that we have picked a metric that flatters the exchange. Notional double counts in a way handle does not: a contract bought and resold three times before settlement logs three times. Combo legs inflate it further. Market makers — including, now, DraftKings — generate volume on both sides without taking directional risk, and a venue can be enormous in notional terms while the actual pool of customer money cycling through it is a small fraction of the number. By that reading, $3.68 billion is a vanity figure and the fee estimate built on it inherits every one of its problems.

That objection is correct on the mechanics and wrong on the conclusion, for one reason: the fee revenue is not an estimate of customer money, it is an observation of exchange income, and it is large whichever way the volume was generated. If wash-adjacent market-making churn is producing the notional, Kalshi still collects on both sides of each fill. The take rate is the number that survives the criticism, and it is the number that reconciles with Sacra's independent estimate.

What the objection does defeat is the "Kalshi is bigger than DraftKings" framing, and that framing should stop. On money at risk, it is not close. On revenue, it is closer than the sportsbook industry would like, and the trend line is pointed the wrong way for them.

What would change our reading

If Kalshi publishes combo fee mechanics showing a per-combination rather than per-leg assessment, the $44 million estimate comes down and our arithmetic needs redoing in public. If the combo share of volume falls back toward the single-event share over the next month, then this weekend was a parlay-promotion artifact rather than the shape of the business. And if the Ohio and Nevada enforcement waves start pulling brokerage distribution out of states, the volume curve breaks for reasons that have nothing to do with product.

Until then, the honest summary of the weekend is narrower than the headline and more interesting. A federally regulated exchange set a record by selling eleven-leg parlays more cheaply than the sportsbooks do, and the number everyone quoted to describe it measures something other than what most readers assumed.

---

Editor's note: TrueEdge builds odds and pricing tools and earns affiliate commissions from licensed sportsbooks, which compete with the venues discussed here. No part of this analysis is a recommendation to trade anything.