Polymarket and Kalshi Arbitrage: What Our Scanner Found in 36 Hours

Polymarket and Kalshi arbitrage exists, but rarely between the two venues. In 36 hours of OddsFantasy scanner data (17–19 October 2026), 597 pre-match arbs had a Polymarket or Kalshi leg, and only 26 of them paired Polymarket with Kalshi. Most paired a prediction market with a sportsbook, and the typical one was gone in about a minute.

Key numbers

• OddsFantasy's scanner found 2,589 distinct pre-match arbitrage opportunities across 783 matches between 17 October 21:00 and 19 October 09:00 UTC. 597 of them (23%) had a Polymarket or Kalshi leg.
• Only 26 were Polymarket against Kalshi (4% of the prediction-market arbs), with a median return of 0.45% after both venues' fees.
• Polymarket against a sportsbook was the richest pairing: 254 arbs with a median return of 1.36%, and 57.5% of them returned 1% or more.
• Prediction-market arbs closed fast. The median listing lasted 1.2 minutes, and 86% were gone within 5 minutes.
• 297 of 1,554 distinct value bets (19%) were on Polymarket or Kalshi, with a median EV of 2.7% after fees. They stayed listed for 4 to 6 minutes, against 13 minutes at sportsbooks.

So the useful question for a Polymarket and Kalshi arbitrage scanner is not only whether Polymarket is cheaper than Kalshi. It is whether either one is out of line with the rest of the market. The sections below break the numbers down by pairing, sport and speed, and show the fee math behind them.

Polymarket vs Kalshi arbitrage is the smallest slice

Polymarket and Kalshi often list the same games, and both are order books whose traders price off the same sportsbook lines, so when one moves the other follows. Fees then take the last few cents. Both venues charge takers rate × price × (1 − price) per contract: 0.07 on Kalshi's general fee schedule and 0.05 on Polymarket's sports markets (Kalshi fee schedule, Polymarket fees). At 50¢ that is 1.75¢ per Kalshi contract and 1.25¢ per Polymarket share, so in a 50/50 game the two asks must add up to less than 97¢ before a cross-venue arb pays anything.

Sportsbooks don't move in lockstep with prediction markets. Their lines change on their own schedule and soft books shade their prices, so gaps open between a prediction market and a bookmaker far more often than between the two prediction markets.

Which pairings pay the most

• Polymarket vs sportsbook: 254 arbs on 158 matches, median return 1.36%, 57.5% at 1% or more.
• Polymarket vs sharp book or exchange: 122 arbs on 91 matches, median 0.89%.
• Kalshi vs sportsbook: 96 arbs on 73 matches, median 0.84%.
• Kalshi vs sharp book or exchange: 99 arbs on 66 matches, median 0.83%.
• Polymarket vs Kalshi: 26 arbs on 24 matches, median 0.45%, 31% at 1% or more.
• Sportsbooks only, for comparison: 1,992 arbs on 689 matches, median 0.83%.

Polymarket's arbs ran larger than Kalshi's, partly because Polymarket lists more markets per game, including team totals, corners and set handicaps, where prices are set less tightly than on the main line. Kalshi's arbs came from the markets it lists most: moneylines, spreads and totals.

Where Polymarket and Kalshi arbs show up

• Football (soccer): 304 of the 597 prediction-market arbs, on 132 Polymarket matches and 54 Kalshi matches.
• Basketball: 128 arbs, on 31 Polymarket and 34 Kalshi matches.
• Tennis: 70 arbs, on 33 Polymarket and 17 Kalshi matches.
• American football: 52 arbs, on 10 Polymarket and 12 Kalshi games. NFL spreads were relisted again and again as the lines moved: 819 Kalshi NFL listings in 36 hours.
• Esports: 32 arbs, on 18 Polymarket and 6 Kalshi matches.

By market, Polymarket's arb listings came mostly from moneylines (573), team totals (439) and totals (276). Kalshi's came from spreads (1,123) and moneylines (769).

How fast Polymarket and Kalshi arbs disappear

Speed decides whether an arb is worth anything. Of 3,797 finished prediction-market arb listings, 44% were gone within a minute and 86% within 5 minutes; only 1.4% lasted longer than 15 minutes. Sportsbook-only arbs lasted a little longer (83% gone within 5 minutes, 4.6% past 15 minutes), because a sportsbook moves when its traders do, while an order book reprices whenever anyone trades.

A scanner that refreshes every few minutes misses most of these. OddsFantasy's feed lists an arb as soon as both prices line up and lists it again whenever they line up again, which is why one game can produce many listings.

Polymarket and Kalshi value bets

A value bet is a price better than the fair odds, here Pinnacle's no-vig line. It needs one good price instead of two, which suits prediction markets, where getting both legs of an arb filled is the hard part. In the same 36 hours:

• Polymarket: 218 distinct value bets on 182 matches, median EV 2.68% after the taker fee, median listing 4.3 minutes.
• Kalshi: 79 value bets on 63 matches, median EV 2.66% after the taker fee, median listing 6.0 minutes.
• Sportsbooks: 731 value bets on 422 matches, median EV 2.67%, median listing 13.1 minutes.
• Sharp books and exchanges were as quick as the prediction markets, at 3.4 to 3.9 minutes.

Fees decide how many Polymarket value bets survive. In a snapshot of our feed on 18 October 2026 at 04:55 UTC, 718 of the 1,801 games Pinnacle priced for the next seven days had a Polymarket market, giving 8,006 shared markets. After the taker fee, only 1.9% of those markets had a side priced above Pinnacle's fair odds, and 9 bets passed every check. With the fee ignored, 66 would have passed. A tool that skips the fee shows about seven times as many Polymarket value bets as you can actually take.

Value also grew with the odds. On Polymarket, value bets at odds under 1.50 had a median EV of 1.49%, against 3.68% at 1.50–2.19 and 6.22% at 2.20–3.49. Kalshi followed the same pattern, from 1.75% under 1.50 to 3.18% at 2.20–3.49.

What a Polymarket and Kalshi arbitrage scanner has to get right

1. Net both fees before comparing. Each prediction-market price becomes odds of 1 ÷ (price + fee), using that venue's fee formula. Exchange odds lose their commission, and sportsbook odds stay as they are.
2. Price from the live ask, with money behind it. Kalshi prices are the best asks with a minimum dollar amount available, and Polymarket prices are rechecked against the live order book before a value bet is listed.
3. Match identical markets: the same line, the same period (regulation time or including overtime) and the same settlement. A Kalshi NFL spread and a sportsbook spread only form an arb if both settle on the same final score.
4. Check start times. Some Kalshi listings carry placeholder times; ITF tennis events, for example, were stamped late in the evening. OddsFantasy uses the earlier of Kalshi's time and the sharp book's kickoff, and skips any pair whose start times are more than two hours apart.
5. Compare against sportsbooks and sharp books too. 96% of the prediction-market arbs in our data involved a sportsbook, sharp book or exchange, so a scanner limited to Polymarket and Kalshi misses most of them.
6. Treat big edges with suspicion. The largest listings in our sample came from brand-new markets seeded near 50¢ and from thin order books. Check the depth and the rules before you stake.

Polymarket and Kalshi arbitrage software options

There are three kinds of tools. Cross-venue scanners compare prediction markets with each other: Oddpool's arb dashboard, for example, covers Kalshi, Polymarket and Opinion and updates every 30 seconds (Oddpool). Sportsbook odds screens compare bookmakers and add prediction markets as one more book. Developer APIs and open-source bots give you code to run yourself. OddsFantasy combines the first two: one feed that compares Polymarket and Kalshi with 50+ bookmakers across the US, Europe, Asia and Africa, including sharp books and betting exchanges, with every fee taken off first.

• Arbitrage and value bets in one place, with Polymarket and Kalshi priced after fees: the arbitrage finder and the positive EV finder.
• Paper trading first. Demo mode simulates the Polymarket value-bet automation against the live order book, fees included, before you commit money.
• Automation on Polymarket. Qualifying value bets can be placed automatically with your EV range, odds range, sports and stake. Kalshi and sportsbook bets are listed for you to place yourself.
• Bookmakers added on request, and one-time 30-day plans with no subscription (pricing).

For a full comparison of arbitrage and value betting tools and their prices, see best arbitrage betting software and best value betting software.

How OddsFantasy measured this

Every arb and value bet that enters the OddsFantasy feed is logged with when it was first seen and when it disappeared, plus its sport, market, venues and edge. This study covers pre-match listings first seen between 17 October 2026 21:00 UTC and 19 October 2026 09:00 UTC (36 hours), after the start-time and matching checks added on 17 October. A listing is one continuous stretch in the feed; a distinct arb is one match, market, set of selections and pair of venues, however often it was relisted. Returns are on total stake, after fees and commission. Value bets are measured against Pinnacle's no-vig odds and listed only when at least two other bookmakers price the same outcome at or below fair and none above it. These are listings, not results: they show what was available, not what was filled.

Before you trade Polymarket and Kalshi arbs

Check that you can legally use each venue where you live: Kalshi is a US-regulated exchange, and Polymarket restricts some countries. Read both venues' rules for the market, since settlement sources, overtime and postponements can differ. Place the leg most likely to move first, size your stake to the depth on the order book, and use the prediction market arbitrage calculator or our arbitrage calculator to split stakes. Trade responsibly: stake only what you can afford to lose.

To see today's Polymarket and Kalshi arbs and value bets as they appear, open the OddsFantasy arbitrage finder. For gaps inside Polymarket itself, read Polymarket arbitrage.