How to Find Arbitrage Bets: Manual Checks, Free Tools and Scanners

To find arbitrage bets, compare the best available decimal odds for every outcome of the same market and add their reciprocals: a total below one identifies a potential arbitrage. Before betting, verify matching settlement rules, available stakes and current prices; a mathematical gap is not a completed hedge.

You can discover those prices manually, check them with a free calculator, or use a scanner to surface candidates. This guide focuses on that discovery-to-execution workflow. For the underlying concept, start with Arbitrage Betting Explained: Odds, Stakes and Risks. All worked numbers below are illustrative, not live opportunities.

How do you find arbitrage bets manually?

Start with one event and one precisely defined market. Your task is to collect the highest available back price for each mutually exclusive outcome, with those outcomes covering every possible result under the same settlement rules. The prices can come from different sportsbooks.

1. Choose a narrow search: one sport, event and market rather than unrelated selections.
2. Open the corresponding market at the sportsbooks you can actually use.
3. Record each selection, its decimal odds, its sportsbook and when you checked it.
4. Keep the highest usable price for each outcome, not simply the highest displayed price.
5. Calculate the reciprocal sum, then inspect the rules and stake availability before treating the result as executable.

What calculation identifies an arbitrage?

For decimal odds O, implied probability is 1 ÷ O. As sacstat explains, implied probabilities in a fair market sum to one, while an excess above one is the overround. Statometrics Academy applies the corresponding arbitrage test: add the reciprocals of the best prices across all outcomes and check whether the sum is below one.

Consider a hypothetical tennis match with exactly two settlement outcomes. Sportsbook A offers Player A at 2.10, while Sportsbook B offers Player B at 2.10. Assume both bets have identical settlement conditions and no fees for this calculation.

• Player A implied probability: 1 ÷ 2.10 = 47.619%.
• Player B implied probability: 1 ÷ 2.10 = 47.619%.
• Combined implied probability: 95.238%, which is below 100%.
• The prices therefore pass the mathematical arbitrage test, subject to execution and settlement checks.

The gap below 100% is not the same as return on total stake. In this example, the gap is approximately 4.762 percentage points, but the equal-payout return is 5% before costs: 1 ÷ 0.952381 − 1. Do not confuse a market’s reciprocal sum with the return your stake allocation produces.

How should you divide the stakes?

For total stake T and reciprocal sum S, stake each outcome at T × (1 ÷ its decimal odds) ÷ S. This makes the gross payouts equal before fees and rounding. In the tennis example, a total stake of 100 currency units splits into 50 on each player. Either winning bet would return 105, leaving a calculated surplus of 5 if both bets are accepted and settle as assumed.

Use the OddsFantasy arbitrage calculator to split stakes rather than estimating them mentally. After rounding, recalculate every outcome’s payout minus the total amount staked. A positive headline percentage is not enough if the actual rounded stakes leave one outcome below break-even.

Which free tools help you discover or verify arbitrage?

Separate finding prices from checking prices. A calculator can establish whether a set of odds works mathematically, but it does not necessarily discover those odds or confirm that a sportsbook will accept your required stake. A useful manual setup combines accessible market pages, a comparison worksheet and a stake calculator.

• Sportsbook market pages: use these to verify the current selection, price and settlement wording.
• A spreadsheet: keep one row per outcome, calculate 1 ÷ odds, and sum the results for each market.
• An arbitrage calculator: enter the best price for every outcome and calculate the stake split.
• An execution record: note accepted stakes and odds separately from prices you merely observed.

Arbocado’s Free Arbitrage Betting Calculator instructs users to enter the best price for each outcome, even when those prices come from different sportsbooks. ABC Tools’ arbitrage guide also recommends rechecking outcome returns after rounding. These are verification tasks: neither a copied price nor a calculated stake proves the complete position can still be placed.

How does a scanner differ from manual discovery?

Manual discovery means you gather and compare the odds yourself. A scanner handles that comparison across a broader set of prices and presents candidate combinations. OddsFantasy’s arbitrage finder compares odds across 50+ bookmakers across the US, Europe, Asia and Africa, including prediction markets, betting exchanges and sharp books. It accounts for fees and commissions first and shows each leg’s bookmaker, selection and odds alongside the opportunity’s return.

The practical difference is where you spend your attention: searching individual markets manually versus checking opportunities already identified by the finder. The verification burden remains. OddsFantasy does not place arbitrage bets automatically; you place each leg yourself at the relevant venues. A listing should therefore start your execution check, not replace it.

Can Telegram alerts replace watching a scanner?

Alerts can bring candidates to your attention, but they cannot confirm your eventual fill. OddsFantasy’s Telegram bot, @OFalert1_bot, sends high-EV value bet alerts on the Classic Telegram plan. The Premium Telegram plan adds arbitrage alerts. Check OddsFantasy’s plans for the relevant option, and distinguish an arbitrage notification from a value bet notification: they require different decisions.

What must you verify before placing either leg?

Apply the same checklist whether the prices came from your own spreadsheet or a scanner. The objective is to establish that the market definitions, obtainable stakes and current prices support the same calculation.

1. Match the event and market. Confirm the participants, scheduled event and exact selection. A football match-winner market and a qualification market are not interchangeable simply because the same teams appear.
2. Read the settlement rules. Check the relevant treatment of extra time, retirements, postponements, cancellations and voids. For the tennis example, do not assume two match-winner bets treat a retirement identically.
3. Cover every outcome. A two-player tennis calculation needs both players under the assumed rules. A three-way football result market also needs the draw. An omitted outcome leaves exposure, not a complete arbitrage.
4. Check executable stakes. Make sure each venue can accept the required amount and that you have the necessary balance. For an exchange leg, check the amount available at the quoted price rather than relying on the price alone.
5. Recheck costs and prices. Confirm each currently obtainable price and any applicable commission or fee. Recalculate the net result for every outcome if either changes.
6. Choose an execution sequence. Identify the harder-to-obtain leg, decide how you would respond to partial acceptance or a price move, and confirm the accepted bet before treating a leg as filled.

What happens if the odds change?

Return to the illustrative tennis prices of 2.10 and 2.10. If Player B’s price falls to 1.90 before you place either bet, the reciprocal sum becomes 1 ÷ 2.10 + 1 ÷ 1.90 = approximately 100.251%. The arbitrage has disappeared. Reallocating 100 units to equalize payouts would now return only 99.75 before costs.

If you already placed 50 on Player A at 2.10, the situation is different: one leg is fixed. Placing another 50 on Player B at 1.90 would return 95 if Player B wins, a loss of 5 on the combined stake. Do not reuse the original stake instructions after a price move. Recalculate from the accepted position, using a hedge calculator where appropriate.

Can you find arbitrage through an exchange or API?

How does back-and-lay discovery work?

Statometrics Academy describes another route: backing an outcome at a bookmaker and laying that same outcome at a lower price on an exchange. This is not the same stake calculation as backing every outcome. Account for lay liability, commission and available exchange depth before deciding whether the position works.

For this back-and-lay workflow, Statometrics recommends checking matching settlement rules and placing the harder-to-get leg first, usually the bookmaker bet. That sequencing does not remove execution risk: the exchange price or available amount may change after the bookmaker bet is accepted.

What data would a DIY scanner need?

A basic scanner needs event identifiers, matching market definitions, outcome prices and update information. The Odds API documentation says its odds endpoint returns upcoming and live games with recent odds for a selected sport, region and market. Decimal odds are the default; relevant exchange lay odds can appear under a separate h2h_lay market key.

Group matching markets, select the best back price for every outcome and flag reciprocal sums below one. Keep back and lay calculations separate. The documentation describes market-level last_update as the last time its system saw that bookmaker’s market odds—not a promise that the displayed price remains executable. A feed cannot replace final venue checks.

How should you check a Polymarket arbitrage candidate?

OddsFantasy’s finder includes opportunities involving Polymarket, but the same verification discipline applies. Compare the exact resolution condition with the other leg’s settlement rules, then verify the obtainable price and amount. Do not treat similar market titles as proof that both legs resolve identically. For that specific assessment, see Polymarket Arbitrage: Is the Price Gap Actually Tradable?.

What is a repeatable arbitrage-finding workflow?

Search narrowly by hand or use a scanner to identify candidates; verify the reciprocal sum and stake split; check rules, balances and executable amounts; then refresh prices before placing each leg manually. Record accepted bets rather than assuming every submitted stake was filled. If a required check fails, skip the candidate or reassess the existing exposure. A calculated edge is not a reason to ignore an unresolved execution problem.