Polymarket and Kalshi arbitrage: the gap, and what survives the fees

Matching two contracts is the hard part, not comparing two prices

A cross-venue comparison is only meaningful when both venues are asking the same question and settling on the same evidence. Two markets can carry near-identical titles and still resolve differently: a different source of record, a different cut-off time, a different treatment of an outcome that half-happens. Comparing their prices produces a number that looks like an edge and is really a difference of definition.

Vultax works from a reviewed match registry rather than from title similarity. Each pair carries a grade — an exact reviewed match, a related but not equivalent pair, or a pair excluded by review — together with the resolution source, a note on the resolution rule, and whether the two close times are compatible. Only reviewed pairs reach the comparison, and the review date travels with them.

Then the fees, from each venue's own schedule

The two venues charge differently, and neither charges a flat percentage of notional the way a spot exchange does. Vultax models each side from the venue's published schedule — Kalshi's fee schedule for the Kalshi leg, Polymarket's own terms for the other — and applies a slippage allowance and an operational allowance on top before anything is called an edge at all.

The comparison table on this page shows the assumption it used in the note beneath it. Your own fees will differ: they depend on your account, whether you are taking or making, and what the venue's schedule says on the day. The point of publishing the assumption is that you can substitute your own and see what happens to the number.

What Vultax will not tell you

There is no execution here. Vultax does not route orders, hold funds, or connect to a venue on your behalf, and the cross-venue posture in the product is watch-only by construction: the calculation carries an explicit flag saying it is not a recommendation, and no plan changes that.

The parts that decide whether a remaining gap is real are also the parts Vultax cannot see: how much depth sits behind the quote at the moment you act, whether you hold cleared funds on both venues, how long a transfer takes, and whether both contracts are available to you at all. Kalshi is a US-regulated exchange with its own eligibility rules. Treat a surviving gap as a question to investigate, and expect most of them to fail on one of those four.

Working through one row

Take a row where one venue quotes an outcome at 39 and the other at 43. The headline is four points, and four points on a contract that settles at 100 sounds like a large number. Now subtract what it costs to be on both sides: a taker fee at each venue, charged per contract rather than as a slice of notional, plus the fact that you are lifting an offer rather than resting a bid. The after-fee column does that subtraction and usually returns nothing.

The second subtraction is the one no table can do for you. To capture the remainder you have to buy the cheaper side and sell — or buy the opposing outcome on — the dearer one, in the same size, before either quote moves. That means cleared funds sitting on both venues in advance, because a transfer between them is measured in hours and the quote is measured in seconds. Money parked on two exchanges to catch a fractional edge has a cost of its own, and it is rarely counted.

What is left after both subtractions is the real number, and on the contracts Vultax tracks it is usually zero or negative. That is not a defect in the market; it is what a functioning market looks like. The value of the table is that it shows you which of the visible gaps are the exception rather than making you work it out four times a day.

• —Raw gap: what the two screens say.
• —After venue fees: what the table computes, using the assumption printed under it.
• —After your fees: substitute your own schedule; the export exists for this.
• —After capital and timing: yours to judge, and usually where the remainder goes.

Why the gaps do not simply close

If a difference persists it is usually structural rather than an oversight. The two venues have different participants: one is reachable with a card and a US address, the other with a crypto wallet from most of the world, and there is no single pool of arbitrageurs able to trade both at scale. Capital that could close the gap is often not permitted to be on both sides.

Contract wording is the second reason. A pair that looks identical can settle on a different source, at a different hour, with a different treatment of an ambiguous outcome. Where that is true, the price difference is a correct assessment of two different bets and closing it would be the mistake. Vultax's reviewed match registry exists precisely so that this class of difference is filtered out before you see it, and pairs that fail review are excluded rather than shown with a caveat.

The third is time. A contract months from resolution ties up capital for months; the annualised return on a two-point edge held for ninety days is not what a two-point number suggests. The rate-path rows on the Kalshi page make this visible: the far meetings are where the widest gaps and the thinnest books both live.