Definition
Arbitrage betting, sometimes called “arbing,” is the practice of placing bets on all possible outcomes of an event at different sportsbooks, using odds that together imply less than 100% probability. When the combined implied probabilities are below 100%, the set of bets can return more than the total amount staked regardless of which outcome occurs — in a frictionless comparison, and before limits, fees, and execution costs.
It is a pricing disagreement between books, not a prediction about the event. The bettor is not betting on an outcome; they are buying both sides of a disagreement.
How sportsbook price differences create arbitrage
Each sportsbook sets its own odds, and books rarely agree. A book may price one outcome generously to attract action or to balance its own position, while another book prices the opposite outcome generously for its own reasons.
Arbitrage exists when the best available price for every mutually exclusive outcome, taken across different books, sums to an implied probability below 100%. That shortfall is probability slack — the distance below the 100% threshold — not the return itself. The bettor captures the underlying disagreement by placing a stake on each outcome at the book offering the best price for it.
A simple two-way example
Consider a two-outcome market. Book A offers 2.10 on Outcome A, and Book B offers 2.05 on Outcome B.
| Outcome | Book | Odds | Implied prob. | Stake | Payout |
|---|---|---|---|---|---|
| Outcome A | Book A | 2.10 | 47.62% | $493.98 | $1,037.36 |
| Outcome B | Book B | 2.05 | 48.78% | $506.02 | $1,037.34 |
| Total | 96.40% | $1,000.00 | ≈ $1,037.35 | ||
Different stakes + different odds → approximately equal payout.
The implied probabilities are 1 ÷ 2.10 ≈ 47.62% and 1 ÷ 2.05 ≈ 48.78%, for a sum S of 96.40%. Because S is below 100%, the combination meets the arbitrage condition. The distance below 100% is 3.60 percentage points of probability slack — not the ROI. For equal-payout staking, the theoretical gross return on the total stake is (1 ÷ 0.9640) − 1 ≈ 3.73%, before fees, limits, and execution frictions.
Arbitrage math at a glance
- Outcome A · 47.62%
- Outcome B · 48.78%
- Slack below 100% threshold · 3.60pp
Theoretical equal-payout ROI: +3.73%Illustrative example only — not a guarantee of profit.
Three-way markets
A three-way market has three mutually exclusive outcomes, common in soccer as home, draw, and away. The same logic applies, but now three legs must satisfy the condition, and the best price for each outcome may live at a different book.
| Outcome | Book | Odds | Implied prob. |
|---|---|---|---|
| Home | Book A | 2.20 | 45.45% |
| Draw | Book B | 3.60 | 27.78% |
| Away | Book C | 5.00 | 20.00% |
| Total | 93.23% | ||
Here the implied probabilities are 45.45%, 27.78%, and 20.00%, for a total of 93.23%. Because that is below 100%, the combination is arbbable in principle. Stake allocation follows the same pattern as the two-way case: weight each leg by its implied probability and scale the stakes so every outcome pays the same amount. Three-way markets add complexity — more legs, more books, and more chances for settlement rules to differ.
Implied probability and the arbitrage condition
Implied probability is the reciprocal of the decimal odds: 1 ÷ odds. A single book normally prices a market so its implied probabilities sum to more than 100%. That positive margin is how the book earns on balanced action.
Across multiple books, let S be the sum of the best available implied probabilities for every mutually exclusive outcome:
When S is below 1, an arbitrage exists. The distance below the 100% threshold — for the two-way example, 100% − 96.40% = 3.60 percentage points — is probability slack, not ROI.
For equal-payout staking, where stakes are weighted so every outcome returns the same amount, the theoretical gross return on the total stake is:
With S = 0.9640, that is (1 ÷ 0.9640) − 1 ≈ 3.73%. So 3.60pp is the distance below the 100% threshold, while 3.73% is the theoretical return on total stake — a distinction worth keeping straight. The condition is a pricing inequality, not an execution guarantee: it says the prices disagree, not that the bets can be placed and settled exactly as calculated.
Execution risks
An arbitrage that exists on paper can still fail in practice. The main frictions are:
- Limits. Sportsbooks cap stakes, and arbitrage typically needs meaningful stakes to be worthwhile. Accounts that consistently take the best prices may be limited or closed.
- Odds movement. Prices change quickly. If one leg moves after the first is placed, the second leg may no longer complete the arbitrage at the intended prices.
- Settlement differences. Books can settle the same event differently — dead-heat rules, overtime or regulation-only markets, and player eligibility are common sources of disagreement. A leg that settles differently can break the hedge.
- Voids. If one leg is voided while the other stands — an event is cancelled or a market withdrawn — the bettor is left with an unhedged position.
These frictions mean a detected arbitrage is a research lead, not a risk-free return. Anyone acting on one still needs to confirm the stakes, the rules, and the timing for themselves.
How SparksOdds relates to arbitrage detection
SparksOdds compares compatible sportsbook prices and surfaces arbitrage candidates in a research table, with validation state, ROI, books, legs, and age. It is a detection and inspection layer: it shows where prices disagree, not an execution system. SparksOdds does not place bets and cannot remove the frictions described above — limits, price movement, settlement differences, and voids all remain the user’s to verify.
Treat the arbitrage table as a starting point for your own checks, not as a guarantee of returns or risk-free execution.