Genesis Arbitrage

Genesis Arbitrage · · 9 min read

Arbitrage Betting, Value Betting and Middles: How They Differ and How to Calculate Each

In short

  • An arbitrage bet (surebet) returns a profit on any outcome, as long as every bet is accepted and settled under the same rules. A value bet pays off only over hundreds of bets, and a middle is a bet on a narrow range where both sides win.
  • A value bet is odds above the fair price. Expected value is p × k − 1. With a fair probability of 50% and odds of 2.20, that is +10% of the amount staked over the long run.
  • Fair probability comes from several sportsbooks' odds with the margin removed, and the median of those shows which price is too high. Odds far above the rest are often a pricing error.

Arbitrage betting, value betting and middles are three ways to bet with an edge. What separates them is when you know the result: before kick-off, after hundreds of bets, or only once the final score is in.

All examples below are football (soccer), settled on regulation time, with decimal odds.

Arbitrage bet: the result is known before kick-off

Hypothetical example: Arsenal vs Chelsea. Arsenal to win at Fonbet @2.45, the draw at Winline @3.90, and YES on Chelsea to win on Polymarket at 30¢. A 30¢ contract is the same as decimal odds of 1 / 0.30 = 3.33.

The check: 1/2.45 + 1/3.90 + 0.30 = 0.9646. The total is below 1, so there is an arb. Out of $1,000, $423 goes on Arsenal, $266 on the draw and $311 on Chelsea YES. You collect $1,037 on any outcome, a 3.67% profit.

The risk of an arb sits outside the match: a bet is rejected, the odds move, the sportsbook voids the bet or settles the market under different rules. Stake sizing is covered in How to calculate a surebet, the types of arbs in Types of surebets, and the risks in Surebet risks and working rules.

Value bet: odds above the fair price

Fair probability is the chance of an outcome with the bookmaker margin removed. Odds of 2.00 match a fair probability of 50%. If a sportsbook offers 2.20 on that outcome, it is a value bet: the price is higher than the outcome is worth.

Plugging in: 0.50 × 2.20 − 1 = 0.10, or +10% over the long run. At the fair price of 2.00, EV is zero. At a typical sportsbook price of 1.90 it is 0.50 × 1.90 − 1 = −0.05, minus 5%.

On prediction markets, decimal odds equal 1 / contract price. So for buying YES the formula becomes EV = p / price − 1.

Variance: what 100 bets look like

Hypothetical example: 100 bets of $10 at odds of 2.20, each with a fair probability of exactly 50%. A win is +$12, a loss is −$10. Turnover is $1,000 and expected profit is +$100.

Metric100 bets
Expected profit+$100
Spread (standard deviation)±$110
Chance of being downabout 18%
A run of 6+ straight lossesabout 55%
A run of 8+ straight lossesabout 17%

You end up down with 45 wins or fewer: 45 × $22 = $990 on $1,000 turnover. The spread is larger than the expected profit, so 100 bets tell you almost nothing about whether you had an edge. Over 1,000 such bets, the chance of being down drops to about 0.2%.

All of this assumes the probability estimate is correct. If the fair probability is really 45%, a bet @2.20 gives 0.45 × 2.20 − 1 = −0.01. A 5-point estimation error turns +10% into −1%.

How to check your bets had value

Value bettors compare the odds they took with the odds just before kick-off, the closing line (see betting line). You took 2.20 and the price closed at 2.05: the market moved your way. If that happens on most of your bets, your probability estimate is probably sound, even while results are still negative.

Middle: a chance to win both bets

A middle appears when two sportsbooks post different lines on the same match stat. Most often that is a total (over/under) or a handicap.

Example: over 2.5 and under 3.5

Hypothetical example: over 2.5 goals at Fonbet @2.10 and under 3.5 goals at Winline @1.80, $500 on each.

GoalsWhat winsResult on $1,000
0–2Under 3.5: $500 × 1.80 = $900−$100
Exactly 3Both: $1,050 + $900 = $1,950+$950
4 or moreOver 2.5: $500 × 2.10 = $1,050+$50

Outside the range you lose $100 or make $50; on exactly three goals you make $950. Basketball handicaps work the same way: team A −4.5 and team B +5.5 both win if A wins by exactly 5 points.

When a middle is worth it

A middle is a bet on the narrow outcome "exactly 3 goals", and it is judged by the same expected value formula. A Poisson model with an average of 2.6 goals per match gives: 0–2 goals 52%, exactly 3 goals 22%, 4 or more 26%.

The pair above: 0.52 × (−100) + 0.22 × 950 + 0.26 × 50 = +$170. Now a normal line with margin: over 2.5 @1.90 and under 3.5 @1.40, $500 each. Results: −$300 on 0–2 goals, +$650 on three, −$50 on four or more. Expected value: 0.52 × (−300) + 0.22 × 650 + 0.26 × (−50) = −$26.

The difference is one price. Under the same model the fair odds on under 3.5 are about 1.35, so @1.80 is a third higher. That happens when one sportsbook has not moved its line yet, or when the line is wrong.

Comparison: arbitrage, value and middles

Arbitrage betValue betMiddle
One match resultProfit on any outcomeWin or lose the whole stakeSmall ± or a large win
Main riskVoids, limits, rule mismatchVariance and bad estimatesRare hits
How oftenRare, until the line movesMore often than arbsLeast often
What you needAccounts at 2–3 booksFair probability and a bankrollDifferent lines at two books

An arb is checked with arithmetic before you bet; value and middles only with statistics over hundreds of bets. That is why many bettors start with arbs and look for value using the same data: odds from several sportsbooks at once.

How to find value bets with fair probability

One sportsbook's fair probability is a division: 1/k of the outcome ÷ sum of 1/k for all outcomes. This strips the margin out of the odds; the method is covered in How to calculate bookmaker margin, and converting odds to probability in Betting odds explained.

One sportsbook can be wrong. The median across several is robust to a single outlier.

Hypothetical example: the same match, 1 · X · 2 outcomes.

Sportsbook1 · X · 2MarginFair Arsenal win
Fonbet2.30 · 3.40 · 3.105.15%41.35%
Winline2.25 · 3.50 · 3.154.76%42.42%
Marathon2.35 · 3.40 · 3.005.30%40.41%
Betcity2.28 · 3.45 · 3.105.10%41.73%
Melbet2.55 · 3.30 · 2.854.61%37.49%

The median fair probability of an Arsenal win is 41.35%, so fair odds are 1 / 0.4135 = 2.42. Melbet offers 2.55: 0.4135 × 2.55 − 1 = +5.44%. Fonbet, by the same estimate, is −4.9%. If YES on Arsenal costs 38¢ on Polymarket: 0.4135 / 0.38 − 1 = +8.8% before the platform's fee.

In Genesis Arbitrage, fair probability is shown on the match market page, in the "Odds" block. The "Fair probability" column is each sportsbook's price with its margin removed, and the median across N platforms is shown below the table. The "Vs market price" column shows how many cents a sportsbook's fair probability sits above or below the market's YES price.

  1. Open the match market

    Go to the market page from the Sport mode table, or from the "All markets" catalog in "Analytics", where the "Odds ×N" tag marks markets with sportsbook odds.

  2. Take the median

    In the "Odds" block, the median fair probability is shown under the table, and the scale shows the spread: "fair probability from … to …".

  3. Calculate expected value

    Plug the median into p × k − 1 for a sportsbook price, or p / price − 1 for buying YES. An edge of a fraction of a percent will be eaten by fees and slippage.

  4. Check the settlement rules

    Confirm whether the market is settled on regulation time or including overtime, and what happens if the match is postponed. Prediction markets and sportsbooks can settle differently.

How to read the table and risk tags is explained in the Sport mode guide, and the surebet calculator sizes stakes against the order books. Where arbs come from and why they close fast is covered in Surebet scanner and Prematch vs live. Arbs using YES and NO contracts are covered in the lesson Surebets with Polymarket and Kalshi.

Questions and answers

Which is more profitable, arbitrage or value betting?

Value bets usually carry a higher expected value per bet, but the result only shows after hundreds of bets, and a losing run can last for weeks. An arb earns a smaller percentage, but every bet can be checked before the match. Many bettors do both: arbs when they appear, value bets from the same fair probability estimate.

How big a bankroll do you need for value betting?

Big enough to survive a losing run. A common approach is a fixed share of the bankroll, 1–2% per bet: at 1%, eight straight losses cut the bankroll by 7.7%. The Kelly formula (p × k − 1) / (k − 1) gives 8.3% of the bankroll for the 2.20 and 50% example, and many bettors stake only a fraction of that, because probability estimates are imprecise.

Why do sportsbooks limit value bettors too?

A sportsbook sees that a customer regularly takes odds above the closing line and cuts their stake limits; this is account restriction. It makes no difference to the book whether that is arbitrage or value betting: either way the customer is taking mispriced odds. More in Why bookmakers limit arbers.

Is a middle a type of arbitrage bet?

Only when both sides outside the range show a profit: then it is an arb with an extra chance to win both bets. If the outside results are negative, a middle is closer to a value bet on a rare outcome and is judged by expected value. See also surebet.