What arbitrage betting actually is
Arbitrage betting (an "arb," or "sure bet") is placing money on every possible outcome of the same event, at different sportsbooks, at prices that add up to less than 100% implied probability. When two books disagree enough about a game, the better price on each side can be combined so that whichever result comes in, your total payout exceeds your total stake. You are not predicting the winner. You are exploiting a pricing gap.
Why do these gaps exist? Every book prices independently and updates at its own speed. One book may hang a stale number after an injury report; another may shade a line to balance its own lopsided action. Across 40+ books, small disagreements are constant. Arbitrage is the purest expression of line shopping: instead of taking the best price on one side, you take the best price on both sides.
The core math: implied probability
First, convert American odds to decimal, because decimal odds make the math clean.
- Positive odds: decimal = (odds / 100) + 1. So +105 → 2.05.
- Negative odds: decimal = (100 / |odds|) + 1. So -100 → 2.00.
The implied probability of a decimal price is simply 1 / decimal. An arb exists when the sum of implied probabilities across all outcomes is less than 1.00.
A worked two-way example
Take a two-outcome market (say a tennis match, or any moneyline with no draw). You find:
- Book A: Player X at +105 → decimal 2.05 → implied 1 / 2.05 = 48.78%
- Book B: Player Y at -100 → decimal 2.00 → implied 1 / 2.00 = 50.00%
Sum of implied probabilities = 48.78% + 50.00% = 98.78%. That is below 100%, so an arb exists. Your margin is 1 / 0.9878 − 1 = about 1.23%.
Splitting the stakes
To lock the return you must size each leg so both outcomes pay the same. The stake on each outcome is proportional to its implied probability:
Stake on outcome i = Total × (implied probability i) / (sum of implied probabilities)
With a $1,000 total bankroll for this arb:
- Player X: $1,000 × (0.4878 / 0.9878) = $493.83 at 2.05
- Player Y: $1,000 × (0.5000 / 0.9878) = $506.17 at 2.00
Now check both branches:
- If X wins: $493.83 × 2.05 = $1,012.35
- If Y wins: $506.17 × 2.00 = $1,012.34
Either way you get back roughly $1,012 on $1,000 staked — a locked ~1.2% return on this event, regardless of who wins. That is the whole idea: the profit lives in the price gap, not in the outcome.
How this connects to vig and +EV
A single book builds in a vig (hold) by pricing both sides so the implied probabilities sum to more than 100% — often 104–110% on a standard -110/-110 market (each side implies 52.38%, summing to 104.76%). That ~4.76% overround is the house edge. Arbitrage is what happens when you assemble a synthetic market from two books whose combined overround has flipped negative.
The same de-vig logic powers +EV betting, arbitrage's close cousin. De-vigging means stripping the hold out of a market to estimate the fair, no-vig price, then betting only when a book's actual price beats that fair number. Arbs are rarer and thinner; +EV spots are more frequent but carry variance. Both start from the same skill — reading true probability out of noisy prices. You can see live examples of both on today's edges.
Why "risk-free" is the wrong word
The math is airtight; the execution is not. Arbitrage carries real, practical risks that no spreadsheet removes:
- Lines move. You place leg one, then the price on leg two shifts before you fund it. Now you are half-hedged and exposed. Thin arbs (under ~2%) can evaporate in seconds.
- Bet limits. Sharp books cap how much you can wager, and your stake-split math may demand more than they will accept on one side.
- Account restrictions. Books actively profile and limit or close accounts that only bet arbs and steam. This is the single biggest reason arbitrage is hard to scale.
- Voids, pushes, and rule mismatches. If one book grades or cancels a bet differently — a scratched player, a different overtime rule, a "action/no action" clause — one leg can void while the other stands, turning a locked arb into an open position.
- Withdrawal and float friction. Capital gets tied up across many books, and holds slow you down.
Treat displayed odds as a snapshot, not a promise: a price you see can be gone before you click. Arbs are a genuine edge, but they are operationally demanding, not free money.
Where a scanner earns its keep
Finding arbs by hand across dozens of books is impractical — prices change faster than you can refresh tabs. This is the boring, mechanical work software does well: pulling live odds from 40+ books across MLB, NBA, NFL, NHL, college, and 3-way soccer, converting every price to implied probability, and flagging the moments a market's combined overround dips below 100%. EdgeFinder does exactly this — surfacing arbs, middles, and de-vigged +EV plays, and backing it with a public, verifiable model track record of logged pregame predictions graded on win rate, ROI, and closing-line value, so you can judge the tool before you trust it.
The free tier shows top edges; a free account and the $9.99/mo tier unlock the full board and the stake-split math on each play. Whatever tool you use, verify prices at the book before staking. 21+ only, and bet responsibly — arbitrage is an analytical exercise, never a guarantee.