Understanding Football Odds: From Price to Probability to EV

Odds Are the Price of a Probability

An odds figure is not a rating, a confidence score, or a tip. It is a price, and what it is pricing is a probability. Read that way, a lot of confusion disappears at once.

Decimal odds state the total return per unit staked, stake included. A price of 2.00 returns two units for one, so one unit of profit. A price of 1.50 returns one and a half. A price of 4.00 returns four. That convention is why decimal odds are the format worth learning first: the conversion to a probability is a single division, with nothing to memorise.

This guide does that conversion, shows what the operator's margin does to it, defines expected value precisely, and then explains the thing that trips up almost everyone — why a genuinely good price and a winning position are two different subjects.

The Conversion: One Divided by the Price

The implied probability of any decimal price is 1 divided by that price.

• 2.00. 1 divided by 2.00 = 0.50. The market is pricing that outcome at 50%.
• 1.50. 1 divided by 1.50 = 0.667. The market is pricing it at 66.7%.
• 4.00. 1 divided by 4.00 = 0.25. The market is pricing it at 25%.

Two more, because they show up constantly on handicap lines: 1.90 implies 1 divided by 1.90 = 52.6%, and 1.72 implies 1 divided by 1.72 = 58.1%.

The reverse direction is just as useful. If you think an outcome is 40% likely, the price that would make it a break-even proposition is 1 divided by 0.40 = 2.50. Anything longer than 2.50 is favourable to your estimate; anything shorter is against it. That single habit — converting a price to a probability before forming any opinion about it — is the largest single upgrade most people can make to how they read a market.

The Margin

Now the complication. Add up the implied probabilities of every outcome in a market and the total will come to more than 100%.

Take a three-way price of 2.10 for the home side, 3.40 for the draw and 3.80 for the away side.

• 1 divided by 2.10 = 47.6%
• 1 divided by 3.40 = 29.4%
• 1 divided by 3.80 = 26.3%
• Total: 103.3%

Those 3.3 percentage points above 100% are the margin — the operator's built-in cut, sometimes called the overround. It exists in every market, on every line, at every price you will ever see.

To recover the market's underlying view, divide each implied probability by the total:

• 47.6 divided by 103.3 = 46.1%
• 29.4 divided by 103.3 = 28.5%
• 26.3 divided by 103.3 = 25.5%

Those now sum to 100%, give or take a rounding fragment, and they are what the market actually thinks. Note what happened to the home side: from 47.6% to 46.1%, a gap of 1.5 percentage points that was never the market's opinion at all.

Two-way markets work identically. A handicap line priced 1.90 on both sides implies 52.6% twice, which is 105.3% — a 5.3 point margin on that line. This is why comparing a raw implied probability against your own estimate without removing the margin systematically flatters the market. How margin interacts with handicap lines specifically is covered in Asian Handicap Explained: Lines, Quarter Balls and Price.

Expected Value, Step by Step

Expected value, EV, is the number the OddsFlow cards carry, and it means one specific thing: the gap between the model's probability for an outcome and the probability implied by the price on offer for it.

Say a price of 2.00 is available, which implies 50%. Say a model estimates the same outcome at 55%. The gap is five percentage points, and that is what a card reports as positive EV.

The same disagreement can also be expressed as a return per unit staked. That is a different number, not a restatement of the five points: the per-unit figure is the gap multiplied by the price, which here is 0.55 times 2.00, minus 1, = plus 0.10 per unit. Written out step by step:

1. 1.Stake one unit at 2.00. A win returns 2.00, so the profit is 1.00. A loss costs 1.00.
2. 2.Weight each by the estimated probability: 0.55 times 1.00 profit = 0.55. And 0.45 times 1.00 loss = 0.45.
3. 3.Subtract: 0.55 minus 0.45 = plus 0.10 per unit staked.

Now change one thing. Keep the 55% estimate and move the price to 1.70:

1. 1.A win at 1.70 returns 0.70 profit; a loss still costs 1.00.
2. 2.0.55 times 0.70 = 0.385. And 0.45 times 1.00 = 0.45.
3. 3.0.385 minus 0.45 = minus 0.065 per unit.

Identical opinion about the match. Opposite conclusion. EV is a property of an opinion and a price together, never of an opinion alone. This is why a signal card is anchored to the price at the moment of publication, and why a card read an hour later may describe a situation that no longer exists.

It is also why a large EV figure is not automatically a better signal. A big gap usually means the model and the market disagree strongly, and strong disagreement is as often a sign of thin data or an unusual market as it is a sign of opportunity. What does the gatekeeping is selectivity: roughly one candidate in twenty-two survives the full filter. The EV field is one input to that decision, not the decision itself — a point expanded in How to Read an OddsFlow Signal Card: Every Field Explained.

Why a Good Price Is Not a Winning Bet

This is the section that matters most, and the one people skip.

EV is a statement about a distribution, not a match. A positive-EV position with a 55% estimate loses 45% of the time when the estimate is exactly right. Losing is not evidence that the estimate was wrong. Winning is not evidence that it was right. Single outcomes carry almost no information about the quality of the reasoning that produced them.

The estimate can simply be wrong. Every EV figure contains a model probability, and every model probability has error attached. If the model says 55% and the truth is 48%, the position looked positive and was negative. Nothing about the arithmetic protects you from that; only calibration checked over long runs does.

The price moves. EV is computed against one number at one instant. Markets reprice, and in live markets they reprice within seconds — the reason live signals carry an elapsed minute at all, as covered in Live Betting Basics: How In-Play Markets Move.

Short runs prove nothing in either direction. A selective process produces streaks in both directions as a matter of course. Any sequence short enough to feel meaningful is far too short to be meaningful. This is the whole reason a record has to be published in full, with losses included and a stated formula — the methodology is set out in How We Count Our Record: The Formula and the Timestamps, and the live figures inside the bot are always the reference.

Taken together, these four points are why nobody serious talks about a single call being right. The unit of assessment is a process observed over many decisions, with the arithmetic visible and the timestamps checkable.

The Short Version

Divide one by the price to get an implied probability. Add the outcomes up, notice the excess above 100%, and divide it out to see what the market really thinks. Compare that to an estimate, and the difference is expected value. Then remember the last part: expected value tells you a price and an estimate disagree. It never tells you what happens next.