Prediction markets vs sports betting
They look similar on the surface — pick an outcome, put money down, get paid if right. The mechanics underneath are different enough that one is suitable for systematic arbitrage and the other mostly isn’t.
1 · The pricing model
A sportsbooksets a line, charges a vig (the “juice” — usually 4-5% baked into the odds), and books your bet against the house. Your fill price is the line they offered when you clicked; you can’t see depth or post your own price.
A prediction market (Kalshi, Polymarket) runs an order book. YES and NO each have a bid and an ask; you can hit the existing book or post your own limit. Price is whatever clears between buyers and sellers, plus a small per-trade fee. You can see the full depth, partial-fill, cancel, scale in.
2 · The fee math
The sportsbook’s vig is invisible to the trader — it shows up as a worse price relative to true odds. The prediction market’s fee is explicit and itemized. Mathematically similar bottom line, but the visible/peer-to-peer structure is what makes arbitrage between two prediction markets feasible (you can compute the exact net edge) and arbitrage between two sportsbooks much harder (the lines are managed against the book, not against another sportsbook’s line).
3 · Regulation
US sports betting is regulated state-by-state under state gaming commissions. Where it’s legal at all, a sportsbook needs a state license; players have to be physically located in-state. Different state, different products, different rules.
Kalshi and Polymarket (post-Dec 2025 QCEX acquisition) are CFTC-regulated federallyas Designated Contract Markets, which preempts most state-level wagering rules. The contracts aren’t legally “wagers” — they’re event contracts. The exception is sports-related event contracts, which several states have actively challenged on the grounds that they look enough like sports betting to fall under state sports-wagering laws. See the legal-status guide for the current state of that fight.
4 · What this means for arbitrage
Sportsbook ↔ sportsbook arb is mostly hand-to-hand: you scan two books, find the same event with mispriced lines, hit one side, race to the other before the line moves. The vig is wide so the gross gap has to be wide; books also limit accounts that profile as arbitrageurs.
Prediction-market ↔ prediction-market arb(e.g. Kalshi × Polymarket) is what Tellmarket scans for. The order books are public, the fees are explicit, and there’s no account-limiting equivalent of a sportsbook ban — both exchanges welcome volume. The edges are usually small (1-3% net) and short-lived, but they’re real and computable.
Sportsbook ↔ prediction-market arb is sometimes called the bonus arbitrage flow — use a sportsbook free-bet promo on one side and a prediction-market hedge on the other. Real but promo-driven; not the kind of systematic edge a scanner surfaces.
5 · The honest summary
Both let you bet on outcomes. Prediction markets look more like stock exchanges: visible order books, peer-to-peer fills, per-trade fees instead of vig, federal regulation. Sportsbooks look more like casinos: invisible vig, house-as-counterparty, state-level licensing. Systematic arbitrage works on the prediction-market side because you can see exactly what you’ll pay; on the sportsbook side it works only on promos and very fresh line movements.