Compare n' Bet™
Compare n' Bet™
The math of betting both sides between books to lock in profit, the realistic ROI on a serious arb strategy, the difference between arbing and middling, and why the practical risks usually eat more of your edge than the theoretical math suggests.
Arbitrage and middling are mathematical strategies that exploit price differences between sportsbooks. In most jurisdictions, the activity itself is legal. Sportsbook terms of service, however, frequently prohibit or restrict arbitrage activity, and detection can result in account limits, account closure, and balance forfeiture as outlined in operator terms.
This guide describes the math and observed mechanics. It does not advocate for, instruct, or encourage any of the following:
Customers are responsible for reading and complying with each operator's terms before placing bets. Compare n' Bet does not endorse activity that violates applicable law or operator terms. Examples in this guide are illustrative and use legitimate venues (sharp-friendly books, exchanges) where arbing is generally welcomed by the operator's stated business model.
Arbitrage exists when the prices on two opposite outcomes of an event, taken at different books, imply combined probabilities below 100%. Bet both sides in the right ratio and your total payout exceeds your total stake regardless of which outcome hits. The leftover percentage is your guaranteed profit.
The math doesn't care about who wins. It only cares about the relative pricing of the two sides. When two books disagree enough that one's "yes" price plus the other's "no" price covers less than 100% of the implied probability space, the arbitrage exists.
Pure arb is the mathematical version of a free lunch. The catch: those free lunches are small (typically 0.5% to 2% per cycle), they don't last long (sharp markets correct mispricing quickly), and the practical risks of executing them tend to eat into the theoretical profit more than the theory suggests.
For a binary event with outcomes priced at decimal odds o1 and o2 at two different books:
Worked example. Tennis match at Pinnacle and a regional book.
That's a meaningful arb. Most real arbs are smaller (0.5% to 2%). Anything above 3% is rare and usually involves a stale line at one of the books that's about to move.
To lock in equal profit regardless of which outcome hits, you split the total stake in proportion to the implied probability of each side:
Continuing the example, with $1,000 total stake:
Equal profit (small rounding gap) regardless of outcome. That's the arb. $37 on $1,000 staked, executed cleanly between two books.
For three-way markets (soccer with home/draw/away, or any market with three outcomes), the same logic extends:
Three-way arbs are rarer than two-way because finding three different outcomes priced cheaply at three different books is mathematically harder. When they show up, the margins can be wider (2% to 5%), but the execution requires placing three bets quickly across three operators before any leg moves.
Arbitrage opportunities exist because different books are pricing the same event slightly differently. The structural reasons:
Books shading toward different sides. One book has heavy public action on the favorite and shades the favorite line accordingly. Another book has more action on the underdog and shades that side. The two books' prices on the same game can drift apart enough to create an arb.
Stale lines after news. Injury news drops, weather changes, lineup announcements. Sharp books update within minutes. Slower retail books can take 5 to 30 minutes to fully adjust. During the gap, the stale price at the slow book can pair with the new price at the sharp book to create an arb.
Different vig levels. A sharp book with 2% vig and a recreational book with 5% vig pricing the same game can produce small arbs even without any shading, just from the structural difference in how much margin each book is taking.
Promotional pricing. An odds boost or a profit boost at one book can create temporary +EV that pairs with the standard price at another book to create an arb. These are the most reliable source of consistent small arbs for retail bettors.
Exchange vs book mismatches. Exchange prices can drift away from book consensus, especially in low-liquidity markets. Pairing exchange and book prices is one of the more reliable arbing patterns where exchanges are available.
Middling is when you bet both sides at slightly different lines, hoping the final result lands between them. If it does, you win both bets. If it doesn't, you win one and lose the other (a small loss equal to the vig).
Standard example. NFL total opens at 47, you bet over 47 at one book. Line drifts to 49 by kickoff, you bet under 49 at another book. Final score: 48 total. Both bets win. The spread between 47 and 49 is the "middle" and any final total of 47.5, 48, or 48.5 produces a winner on both sides.
Middling NFL totals around the key numbers (3, 7, and the 41-49 zone for typical totals) is where the math works most often. Hitting a 1-point middle in those zones happens roughly 4% to 6% of the time depending on the specific numbers.
Middling is mathematically distinct from pure arb. Arb locks in guaranteed profit regardless of outcome. Middling locks in a small loss with a chance of much larger gain. The expected value can be positive even though most middles miss.
Scalping is similar to arbing but in live markets, where prices move continuously. The scalper bets one side, waits for the line to move, and then bets the other side at a better price than the original side, locking in profit.
Example. NBA game, Team A is -3 at the start of the third quarter. You bet Team A. They go on a 10-2 run and the live line moves to Team A -7 by mid-quarter. You bet the opponent at +7. If the final spread lands between 3 and 7, both bets win. Otherwise, you win one and lose the other, with the spread between the two prices determining whether your overall position is profitable.
Scalping has higher volume and smaller per-trade margins than pre-game arbing. The EV is similar in expectation but the execution is much faster. Live betting tools and APIs are required for any serious scalping volume.
Theoretical arb math shows guaranteed profit. Real-world arbing has much messier numbers because of the practical risks below.
The annualized math looks great until you account for what eats it: account limits, account closures, single legs getting taken before the second leg lands, tax treatment of gambling income, and the time it takes to actually find and execute arbs across multiple operators.
Single-leg taken. The single biggest risk in arbing. You place the first leg successfully. By the time you try to place the second leg, the line has moved, the book has limited you, or the bet is rejected. You're now exposed to the first bet alone with no hedge. If that bet loses, you're down the full stake. A handful of these in a year can wipe out months of arbing profit.
Account limits. Books recognize arbing patterns quickly. Bets at unusual prices, immediate hedging at other books, non-round stakes, and bet timing patterns all flag accounts. Once limited, the per-cycle stake drops dramatically and the strategy stops being economical at the limited account.
Account closure. More aggressive than limiting. Some books will close arbing accounts and return the balance. Some will close and forfeit pending bonuses or balance from "promo abuse" clauses, even when no specific promo was abused. Read the operator's terms before depositing.
Tax treatment. In the US, gambling winnings are taxable income at federal and (sometimes) state level. Gambling losses are only deductible if you itemize. For an arber, each winning leg is taxable income while the losing leg may not provide an offsetting deduction. This can turn a 1% gross arb into a negative net arb depending on bracket.
Time cost. Finding arbs manually across multiple books is slow. Paid arb scanners exist but cost $50 to $200/month. Even with a scanner, executing an arb takes a few minutes per cycle. At 1% margin and 5 minutes per cycle, you're earning roughly $12 per hour at $1,000 stake, less if you're getting limited or hitting single-leg-taken issues.
For a retail bettor in the US, the realistic options are limited:
Sharp-friendly books in your jurisdiction. Circa Sports in states where it operates accepts arbing customers. Pinnacle accepts arbing in regulated international markets. These books understand the arb business and price accordingly, which means margins are smaller but accounts are stable.
Exchanges. Novig in the US (where available), Betfair Exchange (UK/Ireland/Europe), Smarkets, Matchbook. Exchanges have no economic reason to limit arbing customers because the exchange takes commission rather than depending on customers losing.
Promotional arbs. Boost-driven arbs where one leg is a profit-boosted bet that pairs with an unboosted bet at another book. The arb is short-lived (the boost expires or sells out), but margins can be 5% to 15% per cycle when they show up.
Futures arbs. Long-running futures markets with stale prices at one book paired with current prices at another. Capital is locked up until the event resolves, which limits annualized return, but the margins per cycle can be 5% to 10%.
For most US retail bettors, pure arbing as a primary strategy isn't realistic. The combination of account limits, taxes, and operational time cost grinds the theoretical edge down to something not much better than the average +EV-betting strategy, while requiring much more capital deployment and operational complexity.
The places where arbing actually works for retail customers are mostly the legitimate ones: betting at sharp-friendly books, using exchanges, hitting boosted promotional offers when they appear, and accepting that the volume will be moderate rather than continuous.
Middling around key numbers and structural mismatches is more realistic than pure arbing, because the per-cycle margins are higher when middles hit. The downside is that most middles miss, so the variance is much higher even though the EV is positive.
If you want a stable, low-variance approach to capturing arb-like edge, focus on consistent line shopping plus sharp-book betting plus opportunistic promotional pickups. That collection of habits captures most of what arbing offers in theory, with much less operational complexity and far lower risk of account closure.
The content on Compare n' Bet is published for educational and informational purposes. By reading this guide you acknowledge:
If you or someone you know has a problem with gambling, the National Council on Problem Gambling helpline is 1-800-522-4700 (US) or visit ncpgambling.org. International readers can find local resources at gamblingtherapy.org.