EV analytics price is the number a bet must beat — the fair spread, total, or moneyline implied by a defensible win probability. Our own game-odds data show when the market's price is truly settled and when a quoted number is only a thin single-book opener.
Several games posted a home spread of 1.5 or minus 1.5 at exactly six books with zero dispersion . Lopsided college openers, by contrast, quoted single-book numbers as large as minus 54.5 . Both facts are price signal, and neither one is an edge by itself.
What is an EV analytics price?
An EV analytics price is what an event is worth once a model turns evidence into a win probability and then into odds. Expected value is the long-run average of a decision: estimate the chance something happens, multiply by the payoff, and compare the result with the cost. The number that decision must clear is the analytics price.
There are always two prices in the picture. One is the sportsbook offer. The other is your estimate of the fair number. The distance between them is the candidate edge. A neat expected-value figure only states that distance; it does not create it.
The betting odds guide shows how to turn a moneyline or spread into the break-even rate a bet must clear. That conversion is the arithmetic. The judgment lives in the probability underneath.
Where our own data show a settled price
Our game-odds rows show several games where the books did not argue. San Diego was a 1.5 home underdog at Cincinnati with zero dispersion across six books . Toronto was a 1.5 home underdog at Cleveland, again at six books with zero dispersion .
The favorites agreed just as tightly. The Mets were minus 1.5 at Tampa Bay across six books . Seattle was minus 1.5 at Boston across six books . San Francisco was minus 1.5 at Pittsburgh across six books .
Zero dispersion is the honest part of that story. Dispersion is how spread-out the books' numbers are; a zero means every captured book quoted the same number. That agreement is a real measurement of the price. It tells you the market settled on a number rather than shopping for one. The closing-line value guide explains why beating a settled number is the whole game.
Where the market is thinner
Not every game gets that treatment. Several college openers carry only one book in the captured rows, and the numbers swing wide. Missouri opened minus 54.5 at home across one book . Idaho opened minus 35.5 at Utah across one book . Coastal Carolina opened minus 21 at West Virginia across one book .
A single-book price is still a price, but it is a thin sample. One source quoting a number is not the same as six independent books agreeing on it. Before calling a wide spread a market, you want more than one quote and more than a snapshot. The spread mechanics guide and the against-the-spread guide walk through what a spread number actually encodes.
Why agreement is not an EV edge
The common mistake is treating a settled price as if it were value. A six-book number is the market doing its job, not a gift. Expected value appears only when your own estimate clears the offered number by enough to beat the vig. The vig and hold guide shows how much cushion the book builds in before you even get to your edge.
Agreement is also evidence against you. A number six books independently landed on is a number that has already absorbed a lot of analysis. Beating it usually needs a fresh estimate, measured on evidence the market has not priced in yet, and graded on games it did not train on. The CLV and sharpness deep-dive explains the discipline behind that estimate.
What would change our mind
We would flip this take on one condition: a calibrated model, graded out of sample, showing a fair price that consistently clears the 1.5 number six books already agree on . A graded sample that keeps clearing a settled price would turn agreement from a measurement into an edge worth betting.
Until that sample exists, the honest read is unchanged. Six books agreeing on 1.5 is a settled price. A single book quoting minus 54.5 is a thin one. Each is evidence about the market, and neither one, on its own, is expected value. The track record page shows where the graded samples live when the evidence is real.
Model calibration from graded predictions
Calibration points render only when a verified source binds prediction probabilities to settled outcomes for the same observations.
Expected value from graded outcomes
Expected-value cells render only when a verified source binds observed win outcomes to the price paid for the same bets.
