← All articles|The Strategy Toolbox

Sports Arbitrage with Prediction Markets: Real Math, Real Catches

Two venues, one game, different prices: sometimes you can take both sides and get paid whichever way it ends. The math is real and this article does it in full. So is the fine print, and this article does that in full too.

Arbitrage is the finance word for a free lunch: a set of positions that profits no matter the outcome. In sports it has existed as long as bookmakers have disagreed with each other. What is new is that prediction markets created a second kind of venue, one that moves at a different speed and answers to different economics than a sportsbook, and where two pricing systems meet, gaps open.

Before the math, the honest headline: real arbitrage opportunities are small, brief, and rationed by the sportsbook's tolerance for winners. This is a technique worth understanding deeply and deploying selectively, not a lifestyle.

The condition: implied probabilities that sum under 100%

Every price is an implied probability. A sportsbook's +120 underdog is priced at 45.5% (100 divided by 220, stake divided by total payout). A prediction market contract at 50¢ is 50%.

An arbitrage exists when you can buy opposing outcomes at implied probabilities summing to less than 100%. Book says Team A wins 45.5% of the time; exchange says Team B (the only other outcome) wins 50%. Total: 95.5%. The missing 4.5% is yours if you size both sides correctly.

The worked example

Tonight's game, no draw possible. Your sportsbook is slow to move and still shows Team A at +120. On the exchange, Team A trades at 50¢, meaning Team B's side costs 50¢.

The play: bet $100 on Team A at the book, and buy 220 shares of Team B on the exchange at 50¢ ($110).

Team A wins: the book pays $220 total. Your exchange shares die. You put up $210 all-in. Net: +$10.
Team B wins: your 220 shares settle at $1 each, $220. The book ticket dies. Net: +$10.

Ten dollars, guaranteed, on $210 deployed for a few hours: about 4.8% risk-free. The share count that balances the two sides is simply the book ticket's total payout ($220) bought as shares on the other side. Payout-matching, the same arithmetic as the hedging article's share-matching, pointed at two venues instead of one position.

General recipe: take the book ticket's total return, buy that many shares of the opposing contract, and check that the total cost of both legs is less than that return. If it is, the difference is locked.

Where gaps actually come from

Gaps have causes, and knowing the causes tells you where to look.

Speed. Prediction markets reprice live, continuously, on open order books. Sportsbook lines, especially live lines, update on a delay and sometimes suspend entirely. In the minutes around scoring plays, injuries, and momentum swings, the exchange is the present and a slow book is the recent past. This is where a live chart earns its keep: watching the market price move while a book line sits still is watching a gap open in real time, and capper.app's live charts are, functionally, that radar.

Crowd pressure. Books shade lines toward the public's favorite teams and inflate prices on longshots. Exchanges clear at whatever two-sided flow agrees on. Systematic shading creates systematic gaps at the edges.

Event chaos. In the seconds after a shock, both venues are wrong in different directions (the overshoot problem). Arbing into chaos is expert-mode: spreads gape exactly when gaps look biggest (why that costs you).

The honest catches, all of them

Books limit winners. The catch that dominates all others. Arbitrage requires the sportsbook leg, and sportsbooks shrink the limits of accounts that consistently beat their prices. Every arb you take spends some of a finite resource: your book account's remaining tolerance. The exchange side has no such problem; exchanges are indifferent to winners.

Fees and spreads eat small gaps. Our $10 example ignored the exchange fee (about 1.5¢ to 1.75¢ per 50¢ contract as a taker) and the spread you cross to fill 220 shares. Call it $4 to $5 of friction on this trade: the "4.8%" is really closer to 2.5%. Gaps under about 2% of deployed capital are usually mirages once friction is honest. Do the net math every time.

The window closes fast. Stale lines get refreshed, suspended, or bet into shape within minutes. Both legs must fill; an arb with one leg filled is not an arb, it is a naked position you did not plan (liquidity check first).

Capital lives in two places. Both accounts must be funded in advance. Money sitting idle waiting for gaps has a cost too.

Settlement fine print. Overtime rules, pushes, and void policies differ between venues. A "both sides covered" position where the venues define the outcome differently is a trap. Read both rulebooks for the specific market before sizing up.

The realistic version of this strategy

For most traders, the durable value here is not grinding arbs; it is that the exchange price is the best free fact-check on every sportsbook number you ever see. The cash-out offer, the live line, the boosted special: hold each against the live market price and you know instantly whether the book is being fair. Sometimes that comparison says "decline the cash-out and hedge on the exchange instead" (the worked version). Occasionally, it says "this gap is big enough to take both sides." And every time, it converts you from someone who accepts prices into someone who compares them.

That habit, unlike pure arbitrage, does not get your account limited; you are just the informed party at the counter.

Educational information, not financial advice. Availability of both venue types varies by state (current map). Prediction markets involve risk of loss, and their legal status varies by location and changes over time.