Article

Polymarket Arbitrage: Why a Price Gap Is Not Guaranteed Profit

Two platforms, opposite bets, an apparent profit. See how fees, partial fills and mismatched rules can undo a Polymarket–Kalshi arbitrage trade.

By VARVAR · Published Nov 8, 2026

Hands hold YES and NO puzzle pieces above a desk with a calculator and receipts, in a textured paper collage.

Cross-platform arbitrage involving Polymarket combines opposite outcomes on different platforms for less than their combined payout. Profit requires matching settlement rules, completed purchases in equal quantities, and a price gap that covers all costs. Two different prices on a screen are only the starting point.

Imagine two browser tabs and one hockey game. A team's victory looks cheaper on one platform than on the other. A pleasant thought appears: perhaps you can put together a trade that earns the difference whichever way the game ends. You would not even have to pick a team. Finally, sport without the stress.

Then come the fees, the different rules and the order that has somehow failed to fill. The stress returns before anyone has reached the ice.

The appeal of Polymarket arbitrage is easy to see. To understand whether a comparison with Kalshi can turn into a profitable trade, we need to follow the money through both purchases and the final payout. Some fictional numbers will help.

How prediction market arbitrage works

In an ordinary binary market, YES means the condition is met and NO means it is not. After final settlement, a winning share pays $1 and a losing share pays $0. A matching pair of YES and NO therefore pays $1 in total, provided both contracts answer the same question and settle consistently.

Here is a hypothetical example, not a set of current quotes. Platform A offers YES at 44 cents. Platform B offers NO at 51 cents. Assume the rules are equivalent and both purchases can be completed in the required quantity.

Buying 100 YES shares costs $44. Buying 100 NO shares costs $51. You have spent $95. If both platforms settle the event consistently as YES or NO, one position pays $100 and the other pays nothing. The difference is $5 before costs.

The opportunity comes from the combination. A cheap YES on its own is still a position that can lose. Adding a suitably matched NO is what changes the calculation.

Hypothetical example: 100 YES shares cost $44 and 100 NO shares cost $51, giving a $95 purchase cost and a $100 combined payout before costs under matching binary settlement.
Both purchases fill completely in this example, and the contracts settle under matching binary rules.

The same match can hide different questions

That $5 calculation depends on YES and NO answering exactly the same question. A matching team name, game date and pair of similar titles do not prove that they do.

Consider another fictional situation. On one site, you buy YES on a team winning in regulation, meaning before overtime. On the other, you buy NO on the team winning including overtime and shootouts. At a glance, you have backed the team and opposed it. All possibilities covered. Or so it seems.

The game is tied at the end of regulation, then the team wins in overtime. The first YES loses because there was no regulation win. The second NO loses because the team did win when overtime counts. Both positions pay zero.

This is not a description of a particular Polymarket–Kalshi pair. It shows how two similar questions can fail to cover each other. In an actual comparison, deadlines, the named source of the result, and postponement or cancellation rules also matter.

Both platforms have contract-specific rules. On Kalshi, the short rules summary is only a summary; the full conditions contain more detail. Reading those conditions is part of the arbitrage calculation. Otherwise, a difference in what you are buying can look like a discount on the same thing.

The screen price may not cover your whole order

The next obstacle is less dramatic but just as useful at ruining a tidy spreadsheet. The screen says 44 cents. How many shares can you actually buy for that price?

A market card's displayed price is not necessarily an available purchase price. Even the best actual offer may cover only a small quantity. To buy more immediately, you might have to accept higher prices from other sellers.

To check the Polymarket side of the trade, look at order-book depth in VARVAR: how many shares are offered at each price. For our intended purchase of 100 YES shares, the question is how many are available at 44 cents and what the rest would cost. The other platform needs its own check.

That can produce slippage: your average purchase price ends up worse than the price you planned around. Multiplying an attractive quote by the number of shares you want is easy. Finding all those shares at that quote is a separate problem.

A limit order lets you cap the price you will pay. It does not oblige anyone to sell. The order may wait, fill only partly or never fill at all.

Suppose all 100 YES shares in our intended combination have been bought, but only 20 of the 100 NO shares have filled. That leaves 80 YES shares without an opposite position. Part of the trade now depends on the match result after all. Two submitted orders are not the same as two completed purchases.

Where the five dollars go

Even if the rules and execution line up, you still have expenses to subtract before calling the difference profit.

Polymarket charges taker fees on certain markets. A taker accepts an offer already available in the order book; a maker leaves an order for someone else to trade against. Polymarket's current documentation says makers are not charged, but a limit order can still act as a taker when it immediately matches an existing offer.

Kalshi also charges trading fees, with differences between markets and maker fees in some cases. There is no universal rate that safely covers every trade on both platforms. Check the fees for the orders you would actually place.

Any applicable transfer, intermediary or conversion costs belong in the calculation too. Use the costs of your own funding route. A spreadsheet in which money appears in the right account by magic is missing a fairly important column.

Back to our $5. Suppose the combined expenses are $4. You have $1 left. The $4 is an invented amount for this example, not a Polymarket or Kalshi tariff. It illustrates how a promising first line can become a much smaller final line.

Be careful not to count the same expense twice. If your purchase cost already uses actual execution prices, the effect of slippage is already included. Subtracting it again would make the calculation wrong in the other direction.

Illustrative calculation: $5 before costs minus $4 in total expenses leaves $1.
Illustrative amounts, not platform fee rates: $5 before costs becomes $1 after $4 of expenses.

Two platforms mean separate balances and settlement

When comparing Polymarket and Kalshi, it is easy to imagine your two tabs as one large account. They are separate balances, separate positions and separate settlement processes. Money on one platform is not automatically available to buy a contract on the other.

Access to both platforms, funding methods and withdrawal timing therefore matter. A price gap is not an executable trade if one half of it exists only on your screen.

There is also the wait. An event can end before the result is confirmed and the payout becomes available. Contract rules or a disputed outcome may delay settlement. A dollar promised by your calculation and a dollar available to spend are different things.

While funds remain tied up in a position, you cannot use the same money for another purchase. The same gain has different practical value depending on how long you must wait to receive it.

Automation can send orders faster. It cannot make incompatible rules equivalent or guarantee that both purchases fill completely. Carrying out the wrong calculation at impressive speed is still the wrong calculation.

Compare the complete trade, not just two prices

A more useful question than “how far apart are these prices?” is: what remains after both purchases execute, all expenses are counted and the contracts settle under matching rules?

That brings the event conditions, available quantity, fees and payout timing into the same calculation. If one of those inputs is still unknown, the final profit is unknown too.

Arbitrage is appealing because it can reduce how much your result depends on who wins the game. For that to work, the opposite positions must actually cover each other. A price gap starts the investigation. Profit depends on the rest of the trade adding up.

Your privacy choices

Your settings and saved markets stay available without analytics. With your permission, we use optional analytics to understand how the site is used.

Settings could not be loaded. Analytics remains off in this page.

Remembers your privacy choice and supports the features you request.

Measures public page visits and feature use. When enabled, a visitor cookie can link visits for up to 90 days. No advertising tracking.

You can change your choice here at any time. Declining does not limit access.