Arbitrage Betting — The Complete Guide
Everything you need to understand and execute arbitrage bets: the formulas for 2-way and 3-way arbs, precise stake allocation math, realistic ROI expectations, the account-limit problem, and how arbitrage differs from dutching.
Quick Calculator
Arbitrage betting — often called arbing or surebetting — is the practice of covering every outcome of an event at prices from different bookmakers whose combined implied probabilities sum to less than 100%. When such a price configuration exists, proportional stakes on each side guarantee a positive return regardless of the outcome. No edge in predicting the game is required. All the edge comes from pricing inefficiency between books.
In practice, arbs are rare, small, and short-lived. Most real arbitrages offer 0.5% to 3% margin and survive in the market for seconds to minutes before one book corrects. The mathematics, however, are clean and worth mastering, because the same machinery also powers dutching, no-vig fair odds extraction, and synthetic market-making.
1. The Core Arbitrage Test
Step 1 — Convert all prices to decimal odds. Step 2 — Compute the arb sum: S = (1 / o1) + (1 / o2) + ... + (1 / oN) Step 3 — Decide: S < 1.00 → arbitrage exists S = 1.00 → fair market (no edge, no arb) S > 1.00 → standard vigged market (no arb) Step 4 — Profit margin: margin = 1 − S ROI = (1/S) − 1 = margin ÷ S Step 5 — Stake allocation on total bankroll T: stake_i = T × (1/o_i) ÷ S payout = T ÷ S (same for every outcome)
2. Worked Example — 2-Way Arbitrage (Tennis)
Book A prices Player 1 at decimal 2.10. Book B prices Player 2 at decimal 2.05. Total stake to deploy: $1,000.
S = (1/2.10) + (1/2.05) = 0.47619 + 0.48780 = 0.96399 Margin = 1 − 0.96399 = 3.601% ROI = (1/0.96399) − 1 = 3.735% Stake P1 = 1000 × 0.47619 / 0.96399 = $494.00 Stake P2 = 1000 × 0.48780 / 0.96399 = $506.00 Total = $1000.00 If P1 wins: $494.00 × 2.10 = $1037.40 → +$37.40 If P2 wins: $506.00 × 2.05 = $1037.30 → +$37.30 Guaranteed profit ≈ $37 (3.7% ROI on the capital)
Both branches pay within $0.10 of each other — the small gap comes from rounding stakes to whole dollars. Funded across two books, you lock in the profit before the event even starts.
3. Worked Example — 3-Way Arbitrage (Soccer)
Book A: Home 3.25. Book B: Draw 3.70. Book C: Away 2.55. Total stake $1,000.
S = 1/3.25 + 1/3.70 + 1/2.55 = 0.30769 + 0.27027 + 0.39216 = 0.97012 Margin = 1 − 0.97012 = 2.988% ROI = 3.080% Stake Home = 1000 × 0.30769 / 0.97012 = $317.17 Stake Draw = 1000 × 0.27027 / 0.97012 = $278.60 Stake Away = 1000 × 0.39216 / 0.97012 = $404.23 Total = $1000.00 If Home wins: 317.17 × 3.25 = $1030.80 → +$30.80 If Draw: 278.60 × 3.70 = $1030.82 → +$30.82 If Away wins: 404.23 × 2.55 = $1030.79 → +$30.79 Guaranteed profit ≈ $30.80 (3.08% ROI)
4. Realistic ROI Expectations
| Arb Margin | Arb Sum S | Frequency | Typical Lifespan | Accessibility |
|---|---|---|---|---|
| 0.5–1% | 0.990–0.995 | Very common | Seconds | Usually sharp vs soft book, small stake caps |
| 1–2% | 0.980–0.990 | Common | 30–120 seconds | Obvious price gaps between books |
| 2–3% | 0.970–0.980 | Moderate | 1–5 minutes | Often soft book error or delayed update |
| 3–5% | 0.950–0.970 | Rare | Minutes | Usually one book slow to move on news |
| 5–10% | 0.900–0.950 | Very rare | Short | Almost always error; high void risk |
| 10%+ | < 0.900 | Almost never | — | Palpable error — likely voided |
5. The Practical Risks That Kill Arbing Accounts
Soft books identify arbing patterns quickly. Within 20–100 arbs an account is often limited to $5–$20 max stake. Professional arbers cycle dozens of accounts and soft books constantly.
If a book concludes the line was a palpable (obvious) pricing error, they will void the bet and return your stake. You remain exposed on the other leg(s) at full risk.
You place leg 1; by the time you get to leg 2 the price has moved and the arb is gone. Best prevention: deploy funded balances across all target books and place legs in parallel (scripts or multi-tab workflows).
Different books may rule on dead heats, retirements, or cancellations differently. A tennis retirement can mean push at one book and loss at another — your arb dissolves.
Soft-book stake limits on side markets can be $100 or less. Your scalable bankroll is constrained not by capital but by market depth.
Some operators slow-roll withdrawals, require documentation, or close accounts when withdrawal patterns look like arbitrage. Plan for multi-week capital lock-up.
6. Arbitrage vs Dutching
| Aspect | Arbitrage | Dutching |
|---|---|---|
| Number of books | Multiple — cross-book price gap | Single book (usually) |
| Outcome coverage | All outcomes of an event | Subset of selections (e.g., several horses) |
| Profit guarantee | Yes — mathematical guarantee when S < 1 | No — depends on whether a selection wins |
| Primary goal | Lock in risk-free margin from price inefficiency | Equalize return across backed selections |
| Typical ROI | 0.5–3% per arb | Depends on edge in the selected subset |
| Best fit | Soft book vs sharp book pricing gaps | Horse racing, sports with many selections |
The mathematics of the two methods are nearly identical — both rely on 1/odds-weighted stake allocation — but their purposes differ. Arbitrage exploits price inefficiency between books. Dutching manages exposure within a single book across multiple selections. See the dutching guide for the in-book stake-equalization formula.
7. Frequently Asked Questions
How do I find arbitrage bets?
Either monitor many books manually for price gaps (labor-intensive and slow), or use an arb-finder service that scrapes dozens of books in real time. Paid arb-finders cost $50–$150/month and can be profitable if you have unrestricted accounts and fast execution.
Can I get banned for arbing?
Banned, no — limited and closed, yes. Most bookmakers reserve the right to restrict any account. Arbing is legal in most countries but violates virtually every commercial book's T&Cs. Expect rapid account degradation at soft books.
How much bankroll do I need to arbitrage?
Because margins are small (often 1–2%) and stake caps apply, meaningful dollar profit requires capital spread across many books. $5,000–$20,000 is a practical minimum for a serious operation; smaller bankrolls struggle to absorb book-limit friction.
Is arbitrage risk-free?
Mathematically, yes — if all legs settle. In practice, no: palpable error voids, settlement mismatches, leg-out risk on price movement, and withdrawal friction all create real-world losses. Expected value remains positive over many arbs, but variance is higher than the pure math suggests.
What's the best arb % to target?
Above 1% net margin (after bet-shading on soft books, which often shade stakes). Below 0.5% the variance from the risks above tends to consume the expected profit. Above 5%, you should suspect an error that will be voided.
Do exchanges (Betfair, Smarkets) help?
Yes — exchange lay prices combined with bookmaker back prices produce many small arbs. Liquidity on exchanges matters; for non-headline markets, exchange liquidity can be thin.
Run potential arbs through the arbitrage calculator. Verify that the implied probability gap is real by checking no-vig fair odds on both books. For in-book stake distribution, use dutching.
Responsible gambling notice. This article is educational only and does not encourage placing bets. Arbitrage is not risk-free in practice; account closures, voided bets, and withdrawal friction create real losses. Never deploy capital you cannot afford to lose. If gambling is causing harm, support is available at BeGambleAware.org or 1-800-GAMBLER. Must be of legal betting age in your jurisdiction.