EV+
Back the price, not the hunch.
A fair probability derived from the sharp consensus, compared against every bookmaker's price — so a mispriced line is something the board shows you rather than something you go looking for.
Fair vs market price
What EV+ means
Expected value is the average return a bet would produce if you could place it many times over. A price is EV+ when it implies a lower probability than the outcome's true probability — the bookmaker is paying you more than the risk is worth.
The arithmetic is unglamorous: multiply your probability estimate by the decimal odds, subtract one. Above zero is positive expected value. The difficulty was never the formula; it is arriving at a probability you can defend, and then finding the prices that disagree with it before the market closes the gap.
This is the difference between backing a price and backing a hunch. A hunch says a team will win. An EV+ read says this specific number is wrong by this much, which is a smaller and far more checkable claim. Consistently taking prices that are wrong in your favour is what separates a process from a run of luck — over a large sample, and only over a large sample.
It is worth being precise about what expected value is not. It is not a forecast of this bet, and it is not a claim about the outcome at all. It is a statement about a price relative to a probability — the same bet at a different number can be positive or negative expected value with nothing about the match having changed. That is exactly what makes it checkable: you are asserting something about a bookmaker's arithmetic rather than about twenty-two people on a pitch.
How PhotonOdds surfaces it
We derive a fair probability for every outcome from the sharp consensus, with the bookmaker's margin removed, and then compare that fair value against each bookmaker's posted price. Where a book sits above fair value, the gap is flagged as an EV+ signal on the board.
De-vigging is the step most hand calculations skip. A posted price includes the bookmaker's margin, so taking implied probability straight off the odds overstates every outcome and the book's total probability sums past 100%. Removing that margin across the market is what makes a fair price comparable to a posted one.
The same computation is available programmatically rather than only on screen: fair_odds and fair_prob carry the derived fair value, and vs_fair carries the delta against the book's price. Developers building their own models get the identical numbers the board renders, not a separate feed.
Reading an EV+ card
The discipline is in reading the fields before deciding, not in scanning the edge percentage and stopping there. The edge is one number among several: the fair price it was derived from, the book price actually on offer, the market and line the signal applies to, and the timestamp telling you how current any of it is.
A large edge on a market you do not understand is not a better signal than a small edge on one you do. The percentage tells you how far the price is from fair; it says nothing about whether the fair price itself is well-founded for that particular market, which is a judgement the board cannot make for you.
Timestamps deserve more attention than they usually get. A signal is a claim about a price at a moment, and both halves of it decay — the book may have moved, and the fair value it was measured against may have moved further. A card that looked strong twenty minutes ago is not a card that is strong now, and the board does not pretend otherwise.
How much edge is actually bettable
Less than the board shows, and this is the part most tools decline to say. An edge is only real to the extent you can get money on at the price. Limits, stake ceilings, and how quickly a book accepts a bet all sit between a displayed percentage and a placed wager.
Small edges are also the ones most sensitive to error in the fair price itself. A one-percent edge derived from a thin market is inside the noise of the estimate; the same one percent on a deep, heavily-traded market means considerably more. Identifying edges and being able to use them are two different problems, and both matter.
Staking is the other half of the question and a separate discipline entirely. An edge tells you a price is wrong; it does not tell you how much to put behind it. Kelly and its fractional variants size a stake against the edge and the bankroll rather than against conviction, which is the part most bettors do by feel.
Why the public underrates it
Expected value is unsatisfying in the way most durable ideas are. It offers no story about the match, no conviction, and no result today. A positive-EV bet that loses looks exactly like a bad bet, and the feedback arrives too slowly and too noisily to feel like feedback at all.
That is precisely why the edge survives. Prices are set against what the market believes, and a discipline that is unpleasant to hold is one fewer people hold. The trade-off is that you are asked to judge your process on something other than this week's results.
There is a second reason, less flattering to everyone involved. Expected value asks you to hold a position you cannot defend socially — no story, no read, no conviction worth sharing. Most people bet partly in order to have an opinion, and an opinion about a decimal is not one anybody wants to hear.
Where EV+ signals go wrong
EV is a long-run expectation, not a per-bet promise — variance means individual bets still lose. No outcome is guaranteed, and no size of edge changes that.
False positives have recognisable shapes. A stale price on a book that has not updated. A market where the sharp consensus is thin enough that the fair value is barely an estimate. A line that reads as mispriced because it is quietly a different line — a different handicap, a different rule on extra time. Checking the market terms before the percentage catches most of them.
Questions, answered
Straight answers to the questions skeptics actually ask.
The edge is in the math, the caveats are on the table. Here's what a sharp bettor wants to know before paying.
Billing, trials & cancellationNo. Expected value is a long-run average, not a per-bet promise. A positive-EV bet can and often does lose; the case for it rests on repetition across a large sample, not on any single result.
From the sharp consensus with the bookmaker margin removed. A posted price includes the book's margin, so implied probabilities taken straight from odds sum to more than 100%. Removing that margin across the market produces a fair probability that is comparable against any individual book's price.
A signal that reads as mispriced but is not: a stale price on a book that has not updated, a market too thin for the fair value to be more than a rough estimate, or a line that differs from the one it is being compared against in its handicap or its rules.
No. PhotonOdds computes fair value and the edge for you and surfaces them as ready-to-read signals. The same numbers are available as API response fields, and the free EV calculator lets you run the arithmetic by hand if you want to see it.
Not on its own. A large edge on a thin market can be less trustworthy than a small one on a deep market, because the fair price it is measured against is itself less certain. Liquidity, limits, and market terms all bear on whether an edge is usable.
Start your free trial and see the board move before the market does.
No commitment, no hidden setup. Just sharper pre-match insight from day one.
PhotonOdds is a sports-betting data & analytics service — not a bookmaker. We never accept wagers or guarantee profit.
18+ only. Betting carries risk. Please gamble responsibly. Learn more