Arbitrage Between Sportsbooks and Prediction Markets

Arbitrage between sportsbooks and prediction markets means backing one outcome at a bookmaker and the opposite outcome as an event contract, at prices that guarantee a profit either way. It works when the two implied probabilities add up to less than 100%. The gaps are real — the two platform types price events with completely different machinery — but they are smaller than a quick scan suggests. A gap that looks like 4% on screen is usually 2-3% once contract fees, the bid-ask spread and thin order-book depth are paid for, and it is only profit at all if both platforms settle the event the same.

Why sportsbook prices and prediction market prices disagree

A sportsbook posts a number. Traders on a prediction market negotiate one.

Books build a line from their own model, then shade it for liability and add a margin — that's why a standard two-way market sits near 1.91 on both sides instead of even money. They employ people whose whole job is reacting to a lineup scratch within seconds.

Event contracts work like an order book. The price is whatever the last buyer and seller agreed on, so it reflects whoever is awake and holding capital in that market. There's no margin baked into the quote, but there are trading fees, and on quiet markets the spread between best bid and best offer does the same job as vig.

That difference produces predictable places for gaps:

• Right after news breaks. Books repriced injuries and weather within seconds; a thin contract book can sit stale for minutes.

• On low-liquidity markets. Niche leagues, non-headline games, and anything outside the main slate.

• Between two prediction markets. Kalshi and Polymarket frequently disagree with each other on the same question.

• At the extremes. Longshot pricing behaves differently in a cents-based ladder than it does in a book's futures column.

Neither venue is reliably "smarter" than the other. Books tend to be faster on breaking information; contract markets tend to be cheaper to trade on the margin. The edge you are harvesting is a timing gap, not superior wisdom on one side.

When arbitrage between sportsbooks and prediction markets actually exists

Convert both legs to implied probability and add them. Below 100%, an arb exists. At or above, you're paying for the privilege.

Contract prices convert for free: a Yes contract at 55¢ pays $1 if it hits, so it implies a 55% chance. Sportsbook odds need one step — a decimal price converts as 1 ÷ decimal (our odds explainer covers the American and fractional versions).

Take an NBA game. Your book has the underdog at +145; a prediction market has the favourite's Yes contract at 55¢.

Decimal equivalent of +145 is 2.45; 1 ÷ 2.45 = 40.8%. Figures rounded to one decimal place.

That 4.2% gap is the gross edge. Split $1,000 in proportion to each leg's implied probability — $426 on the moneyline, $574 on the contracts — and either result returns about $1,043.

One caution before you get excited: this only works when the two legs cover every outcome exactly once. Two-way sports are clean. A soccer match with a draw needs all three results covered, or you have a bet, not an arb.

Fees, spread and depth eat most of the screen edge

Now price the same trade properly.

Kalshi's published fee schedule charges trading fees off a curve in the shape 0.07 × contracts × price × (1 − price), which peaks around the 50¢ mark and shrinks toward both extremes. On the $574 leg above — roughly 1,043 contracts at 55¢ — that's about $18. Polymarket's terms have differed by market and have changed over time, so read the current schedule rather than assuming zero.

Total outlay becomes $1,018 for a $1,043 return. Profit: about $25, or 2.5% — down from the 4.2% on screen, and that's before three more costs:

• Spread. The 55¢ quote is the price to buy. Selling might be 53¢. Your real entry is the offer, not the midpoint.

• Depth. Top-of-book size is often a few hundred dollars. Filling $574 may walk you two or three ticks up the ladder.

• Withdrawal and conversion. Moving USDC, card fees, or a currency hop can quietly delete a 1% margin.

For a fuller cost breakdown across venues, see our prediction market fee comparison.

Settlement rules are the leg risk nobody prices

The failure that actually hurts is not a price moving. It's both legs losing because the two platforms defined the event differently.

Check these before you commit capital:

• Overtime and extra time. Does the contract resolve on regulation or final score? Books vary too.

• Postponements. A book may void a bet on a game moved beyond 24 hours; a contract may resolve No, or roll to the new date.

• Player participation. Prop-style contracts and sportsbook props often use different minimum-appearance rules.

• Resolution source. A contract resolves on a named data source. If that source and your book's official feed disagree, you own the difference.

• Cancellation policy. A voided sportsbook leg returns your stake and leaves the contract leg naked.

Read both rulebooks in full the first time you trade a market type. After that it's a one-line check.

Will a sportsbook limit you for arbing?

Yes, routinely. Books detect the pattern without much effort: odd stake sizes, bets landing seconds after a line moves, consistent action on stale prices, and the same accounts appearing across a shared risk network. The usual response is a stake limit rather than a ban — you keep the account, but your maximum drops to pocket change.

Prediction markets are structured differently. They earn on volume, so an arber is a customer, not a liability. What limits you there is capital, order-book depth and speed. In the US, access also depends on which venues are CFTC-regulated — our list of regulated event-contract exchanges covers who is currently operating.

Sizing, execution and the leg you fire first

Arbs die in seconds. Preparation does most of the work:

• Pre-fund every venue. You cannot move money mid-trade. Idle capital on three platforms is the cost of doing this at all.

• Use limit orders on the contract leg. A market order on a thin book gives away the edge you came for.

• Size to real depth, not the top quote. If the ladder only supports $200 at 55¢, that's the trade size.

• Cap each opportunity. A fixed ceiling per arb keeps a single settlement dispute from mattering.

• Log everything gross. Both legs, both fees. Winnings are taxable income in most jurisdictions regardless of the net.

Practitioners split on which leg goes first. One camp fires the sportsbook leg first, because books can reject, delay or requote a bet, and a rejected leg leaves you with a manageable open position on the exchange. The other fires the thin contract leg first, on the logic that depth is the binding constraint and a partially filled order book is the harder problem to solve. Both are defensible; pick one, write it down, and don't improvise mid-trade.

Compare sportsbook lines and prediction market contract prices side by side at /compare.