Polymarket Arbitrage: How To Find and Execute It

Buy YES at 47 cents and NO at 51 cents on the same Polymarket market and you've just spent 98 cents to guarantee a $1.00 payout — a locked 2-cent profit before fees, no matter which side resolves true. That gap is Polymarket arbitrage in its purest form, and it's the starting point for understanding why most of what gets marketed as "free money" on prediction markets isn't, once fees and slippage get applied. Here's how to actually find and execute it.

Step 1: Check the within-market YES + NO price

Every Polymarket market prices YES and NO as complementary shares between $0.00 and $1.00 that should sum to $1.00 at fair value. In practice, thin order books or fast-moving news can leave a temporary gap where the best ask on YES plus the best ask on NO adds up to less than a dollar. Buy both, and whichever side resolves true redeems for exactly $1.00 — you collect the difference between what you paid and that dollar regardless of the outcome.

This is the most mechanical arbitrage on the platform because it requires zero opinion on the underlying event. You're not betting on the election or the game; you're betting that basic arithmetic holds. That's also exactly why it's the hardest version to actually catch — see Step 3.

Step 2: Compare the same event across venues

Cross-venue arbitrage means finding the same real-world event priced differently on two platforms — most commonly Polymarket vs Kalshi, since both run deep CLOB order books on overlapping political, economic, and sports questions. If Kalshi prices a Fed rate decision at $0.62 for a cut and Polymarket prices the equivalent contract at $0.55, that 7-cent gap is a candidate — but only a candidate. Before treating it as tradeable, confirm the resolution criteria and settlement dates actually match. Prediction markets on the "same" headline event sometimes differ on exact wording, deadline, or the data source used to settle, and that difference alone can explain the entire price gap with no arbitrage present at all. Our Polymarket vs Kalshi comparison breaks down where the two venues structurally diverge on resolution and settlement.

Step 3: Subtract fees and slippage from the raw spread

This is where most theoretical arbitrage dies. Polymarket introduced category-based taker fees in 2026 — makers are never charged, but takers pay a fee that peaks at the 50-cent price point and tapers toward the extremes. The max taker fee per 100 shares is $0.75 for sports, $1.00 for politics/finance/tech, $1.25 for economics/culture/weather/other, and $1.80 for crypto markets. Geopolitical and world-events markets are fee-free. Full detail and the current schedule live at docs.polymarket.com/trading/fees — this changes, so verify before sizing a trade.

Kalshi's taker fee is calculated as round_up(0.07 × contracts × P × (1−P)), which simplifies to roughly 7 cents times contracts times P times (1−P). At the worst price point, 50 cents, that's $1.75 per 100 contracts — buying 100 YES at $0.50 costs a $1.75 fee. Near the extremes it drops sharply: buying at $0.90 costs about $0.63 per 100 contracts. See kalshi.com/fee-schedule for the live schedule.

Run the math on the within-market example above: buying YES + NO for 98 cents nets 2 cents of raw spread on 1 share. Scale to 100 shares and the raw edge is $2.00 — but if that market falls in Polymarket's politics category, the taker fee on each leg can run up to $1.00 per 100 shares, meaning two taker legs can consume the entire spread before it even settles. A cross-venue trade with a Kalshi leg near 50 cents adds another $1.75-per-100 fee on top. This is the arithmetic that separates a real opportunity from a spread that looks wide on a screenshot and evaporates the moment you account for what both fills actually cost.

Fee source	Rate	Worst case (per 100 shares/contracts)
Polymarket taker, crypto category	Symmetric around 50¢, peaks at midpoint	$1.80
Polymarket taker, politics/finance/tech	Symmetric around 50¢	$1.00
Polymarket taker, sports	Symmetric around 50¢	$0.75
Kalshi taker	7% × P × (1−P)	$1.75 at P=0.50

Step 4: Size the trade to available depth

A 7-cent cross-venue gap on a market with $50 of resting size on each side isn't a trade — it's a rounding error you can't fill at scale. Pull up the order book depth on both legs before committing capital. If you can only fill 40 of the 100 shares you wanted at the quoted price before the book moves against you, your realized spread is worse than your theoretical spread, sometimes enough to flip the trade negative after fees.

Step 5: Execute both legs as close to simultaneously as possible

Arbitrage only locks in profit if both legs fill near the prices you checked. A five-second gap between buying YES on Polymarket and shorting the equivalent exposure on Kalshi is five seconds where the market can move and turn a locked spread back into directional risk on one uncovered leg. This is precisely why manual arbitrage on liquid markets is a losing game against automated scanners — a human clicking through two separate platform UIs cannot execute two legs at the same instant the way a script watching both order books can.

PolyMarketMaker's predictive arbitrage scanner watches Polymarket US, Polymarket global, Kalshi, and PredictIt simultaneously, with per-market fee math built in so the spread you see already accounts for what Step 3 costs — not the raw, pre-fee number that looks tradeable until you fill it. PolyMarketMaker surfaces only the gaps that survive fees and current depth. Simulation is $149/mo, Live Trading is $299/mo.

Why most "free money" on prediction markets dies

Three things kill arbitrage that looks profitable on paper: fees eating the spread (Step 3), slippage from thin depth (Step 4), and execution lag between legs (Step 5). Add a fourth for cross-venue trades specifically — resolution mismatch, where two platforms are technically settling slightly different questions and the "arbitrage" was never real. Our broader guide to prediction market arbitrage and the arbitrage fee math breakdown both walk through worked examples of spreads that look wide until fees are applied — the same discipline applies whether you're comparing Polymarket to Kalshi or Kalshi to PredictIt. None of this means arbitrage is dead on these platforms; it means the surviving opportunities are narrower and faster-moving than the screenshots that circulate on social media suggest, and treating every visible price gap as free money is how new traders lose to fees they didn't model. For the broader playbook this fits into, see the Polymarket trading strategies pillar guide.

FAQ

What is Polymarket arbitrage?

Trading a price discrepancy that locks in a payout regardless of outcome — either within one market (buying YES and NO for a combined price under $1) or across two venues pricing the same real-world event differently.

Is Polymarket arbitrage still profitable in 2026?

Opportunities still appear, but Polymarket's 2026 category taker fees (up to $1.80 per 100 shares on crypto markets) and Kalshi's taker fee (up to $1.75 per 100 contracts at 50 cents) eat most spreads under roughly 2-3 cents once both legs are filled.

Why does within-market YES+NO arbitrage disappear so fast?

Because it's the most mechanical, lowest-risk form of arbitrage available, bots scan for it continuously. Any gap wide enough to survive fees gets filled within seconds on any market with meaningful volume.

Do I need to be fast to trade Polymarket arbitrage?

For within-market gaps, yes — they close in seconds. Cross-venue spreads driven by differing news reaction speed or liquidity depth can persist longer, sometimes minutes, giving manual traders a real window.

This article is for educational purposes only and is not financial advice. Trading involves risk of loss.