Kalshi Undercut the Books by 12 Basis Points in the One Market Almost Nobody Bets

Citizens JMP Securities pulled 28 prices on Friday, September 11 — moneylines and totals across NFL Week 1 — and found that Kalshi was charging less than either of the two biggest American sportsbooks. Blended implied vig: Kalshi 4.32%, FanDuel 4.44%, DraftKings 4.51%.

That is a real result and a genuine reversal. A year ago the same comparison ran the other way, with Kalshi at 4.84% against FanDuel's 4.42% and DraftKings' 4.48% across the NFL season. Analysts Jordan Bender and Isabelle Slavin credit rising volume, deeper participation and competition among liquidity providers, which is the textbook story of an exchange growing into its own spread. Nothing about it is spin.

It is also, as a guide to what a bettor pays, close to useless.

Break the 28 prices apart

The blend hides that the two halves of it point in opposite directions.

On moneylines — who wins the game, the single most common bet in American sports — FanDuel was cheaper, at 4.08% against Kalshi's 4.14%. On totals, Kalshi won comfortably: 4.50% against DraftKings' 4.71% and FanDuel's 4.80%.

So the headline advantage is not a general one. Kalshi is cheaper on over/unders by 20 to 30 basis points, more expensive on sides by six, and the average of those two comes out at a twelve-basis-point win because the totals gap is wider than the moneyline gap. If you bet sides, the study says to keep betting FanDuel.

Then there is the third market, which Citizens measured separately and which almost every write-up has treated as a footnote.

Kalshi's implied vig on a combined favourite-and-over position in Week 1 was 23.8%. DraftKings and FanDuel were both at 22.0%. Kalshi is 180 basis points worse on the multi-leg product — fifteen times the size of its advantage on the blend, in the other direction.

Where the money actually was

Kalshi set consecutive single-day volume records over opening weekend: $2.426 billion on Saturday, $2.433 billion on Sunday. Combo markets were more than half of both, at roughly $1.47 billion and $1.48 billion.

Straight football contracts were $328.7 million on Saturday and $409.9 million on Sunday.

Sit with those two numbers next to each other. On the biggest day in the exchange's history, single-game football positions — the market where the Citizens sample found Kalshi competitive — were about 17% of the volume. Combos, where Kalshi is the most expensive of the three venues tested, were more than half.

One honest caveat, because the published breakdown does not support the inference everyone is making: Kalshi's combo figure is not labelled as football-only, and sports were roughly 26% of the exchange's $12.98 billion week. We cannot tell from what has been released how much of that $1.48 billion was NFL. Nobody who has cited the number this week can either. What the disclosure does establish is that the multi-leg product is the centre of gravity of the business, and it is the product Kalshi prices worst.

This is not a Kalshi pathology. It is the same shape as the sportsbook P&L, which is why DraftKings and FanDuel hold 22% on a two-leg parlay and 4.1% on a side and have organised their apps around pushing you toward the former. The interesting fact is that the exchange, whose whole pitch is that peer-to-peer pricing beats a bookmaker's margin, has reproduced the bookmaker's margin structure and then set it slightly higher.

The fee

Citizens flagged this and then everyone dropped it: the combo comparison is "before transaction fees." The analysts put Kalshi's average charge at $1.62 per 100 contracts.

Kalshi's published general fee is 0.07 × C × P × (1 − P), rounded up, where C is contracts and P is the price in dollars. The function peaks at a fifty-cent contract, which is why the maximum is $1.75 per 100 and why the most expensive trade on the exchange is a coin flip. At a typical NFL side priced 60/40, it is $1.68 per 100 contracts.

Now put that beside twelve basis points.

A hundred contracts at 60 cents risks $60 to make $40. The $1.68 fee is 2.8% of the amount at risk, paid on entry, on top of whatever the spread already costs. The sportsbooks charge nothing separate; their margin is entirely inside the price Citizens measured. Whatever denominator you prefer — notional, stake, or expected win — the fee is an order of magnitude larger than the advantage the study found. It does not shave the twelve basis points. It swamps them.

That is the whole finding, stated plainly: Kalshi's price advantage exists, is confined to totals, reverses on sides, reverses badly on the product that generates most of its volume, and is measured before a fee that is roughly thirteen times its size.

The best argument against all of that

Two, and the second one is the real one.

First, the fee is a taker fee. An exchange lets you post a limit order and wait, and a patient maker pays the 0.0175 coefficient where maker fees apply, or nothing at all in markets without them. A disciplined trader working orders into the book genuinely does beat FanDuel's price, and no sportsbook offers that option at any price. I concede most of this. The caveat is that the flow arriving on a Sunday morning through a phone app is overwhelmingly taker flow — that is what a $2.4 billion day of retail football action is — so the honest version is that Kalshi is cheap for the trader who behaves like a market maker and expensive for the customer who behaves like a bettor.

Second, and much more important: price is not why anyone should be choosing an exchange. FanDuel's 4.08% moneyline is available to you until you start winning, at which point your maximum stake becomes $12 and the 4.08% is a number on a screen you cannot transact against. Kalshi will fill you at size because the counterparty is another trader rather than a house carrying your position. You can also exit before settlement at something near mid, which no sportsbook ticket allows. Neither of those shows up anywhere in a 28-price vig comparison, and both are worth vastly more than twelve basis points to the only customers who profit from this in the long run.

Citizens' own conclusion from the note points the same direction: they read the data as saying prediction-market cannibalisation of regulated sportsbooks is "not worsening and may be easing," with Kalshi drawing mainly from offshore rather than from DraftKings and FanDuel. That is consistent with everything above. The people moving to the exchange are not the ones comparing vig to three decimal places. They are the ones who could not get a bet down anywhere else.

Which is a much better argument for Kalshi than the one being made for it this week — and it is the argument the pricing study, read carefully, quietly supports.