Cross-Market Arbitrage on Polymarket: Bots vs Sportsbooks & Exchanges

Cross-market arbitrage compares a Polymarket contract with a position on another venue, such as an exchange or sportsbook. The central question is whether those positions cover the same outcomes under compatible settlement rules, after every fee and execution constraint.

Match the event before comparing prices

Read the exact event, deadline, resolution source and cancellation conditions on both venues. A sportsbook moneyline, an exchange contract and a Polymarket outcome can refer to similar events while treating overtime, postponement or a disputed result differently.

Build a payoff table for each possible event state, including voided or cancelled outcomes. If a state is missing or uses a different settlement rule, the position is not the fully matched hedge that the headline prices suggest.

Compare executable prices and available size

Use available order-book depth on Polymarket rather than a last trade. On the second venue, confirm the amount currently available, acceptance rules and any limits before calculating the combined position. A visible quote is not a completed fill.

Convert prices into a common payoff unit and include fees, spreads, currency exposure and settlement timing. Do not assume identical collateral or immediate access to proceeds. The current Polymarket settlement collateral is pUSD; another venue may use a different currency or contract structure.

Sportsbooks, exchanges and prediction venues

Kalshi and other prediction venues need the same contract-by-contract comparison. This guide does not assert that a particular current pair is interchangeable or profitable. Verify each venue’s own rules and your eligibility before using a pair.

A worked cross-venue hedge

Consider a hypothetical event with exactly two matched outcomes and identical settlement treatment on both venues. On Polymarket, buying 100 YES shares at an executable $0.40 costs $40 before fees and pays $100 if YES wins. On a second venue, suppose a $30 stake on NO at decimal odds of 3.3333 produces approximately $100 gross return if NO wins. The combined upfront cost is $70 before fees.

Under those deliberately simplified assumptions, the gross difference is approximately $30. This is arithmetic using invented prices, not a currently available opportunity. Fees, accepted stake size, contract wording and settlement timing can reduce or invalidate it. Decimal odds include return of the winning stake; counting that stake again would overstate the result.

Add a third state in which the external venue voids the bet and returns the $30 stake, while Polymarket resolves NO and the YES tokens pay nothing. The combined return is then $30 against the $70 cost. The apparent hedge loses $40 in that state before fees. This shows why a payoff table must include cancellation treatment, not just the two ordinary results.

For contracts with different units, first choose a common target payoff and calculate each leg's required quantity. Do not compare a one-share prediction quote directly with an exchange lot or a sportsbook stake. Save the conversion and all cost assumptions with the candidate so the calculation can be reproduced when prices change.

Automate comparison without assuming atomic execution

A scanner can normalize contracts and flag a candidate, but each venue has its own order interface. Submit only when the strategy can tolerate a failed or partial second leg. Track requested size, filled size and the remaining unmatched position independently.

Polymarket’s current SDK and order types do not make a trade on another venue atomic. A successful Polymarket acknowledgement is not proof that an external hedge filled. Use a bounded unwind rule rather than repeated orders aimed at recreating an expired spread.

Fees and resolution risk

Polymarket fees depend on the market’s category and price; the old universal flat-fee comparison is not applicable. Read the configured fee parameters and the second venue’s fees before calculating the net result. Resolution delays can also tie up capital even after both legs execute.

Prepare both venues before submitting either leg

Check balances, usable collateral, trading permissions and the maximum accepted size independently. Money on one venue does not automatically fund the other leg. A transfer made after the first fill may arrive too late to complete the hedge, leaving exposure during a changing event.

Choose a leg sequence and define what happens when acceptance differs from the expected fill. An external bet request may be rejected or accepted at a smaller amount; an exchange order may remain open. Record the confirmed amount on each side. A submission response that merely acknowledges receipt should not be counted as final acceptance.

Include currency conversion and withdrawal costs where relevant. The profit estimate should use the amount actually expected after all known charges, not a headline spread. If a required fee or acceptance rule is unknown, flag the pair for review rather than substituting zero.

Paper-test the complete cross-venue process

Record synchronized observations from both venues and model the delay between them. A pair selected from two unrelated historical timestamps may never have been available at the same time. Test the sequence of submission, confirmation and possible unwind, including the cost of closing only one leg.

Keep a candidate log even when the system rejects a pair. Reasons such as mismatched overtime rules, unavailable size or an unknown fee explain why a visible discrepancy was not traded. Review unresolved positions separately from settled results so the test does not quietly assume that pending proceeds are spendable.

The algorithmic VPS guide can help size a continuously running comparison process. A Polygon VPS concerns infrastructure for Polygon-related software, while each external venue still needs its own integration. Hosting improves process availability; it does not synchronize the venues' settlement or guarantee either leg.

Keep internal arbitrage separate

For YES/NO sets and related contracts entirely inside Polymarket, see Polymarket arbitrage bots. Those relationships belong to the internal arbitrage guide rather than a comparison with outside venues.

For strategy selection and paper trading, see automated trading on Polymarket. Paper-test cross-venue settlement differences as well as normal price changes.

Host the Polymarket component of a cross-market bot

A cross-market bot can need both venue readers and its permitted order process to stay available. QuantVPS's Dublin hosting for Polymarket publishes under 1ms to Polymarket's London servers. Polymarket's own terms decide who can trade; a VPS doesn't change that.

FAQs

Do matching market titles mean the outcomes are equivalent?

No. Check timing, resolution sources and void rules. Similar descriptions can still produce different payouts for the same real-world event.

What happens if only one venue fills?

The position is exposed until it is hedged or unwound. Define a residual-position limit and a stop condition before submitting the first leg.

How should sportsbook odds be compared with token prices?

Convert both positions to the same gross payoff and include the required upfront stake or token cost. Decimal odds include the winning stake in their gross return. Check void rules and accepted size before treating the calculated payoff as matched.

Can profits from one venue immediately fund another trade?

Not necessarily. Resolution, withdrawal and transfer timing differ between venues. Maintain balances for the intended legs in advance and include capital that remains tied up in unresolved positions.