The sportsbook's built-in margin — the percentage by which implied probabilities exceed 100%.
A fair coin flip should pay +100 / +100 — risk a dollar to win a dollar, with the two sides summing to a 100% probability total. Sportsbooks don't offer fair prices. They offer -110/-110, or worse, and the difference between the two-sided implied probability total and 100% is the vig.
On a standard -110 / -110 market, each side implies about 52.4%. The two sides sum to 104.8%, an overround of 4.8 percentage points. That 4.8% is the book's theoretical margin per dollar — but only if action is balanced equally on both sides. If everyone bets one side, the book is exposed to the result and the actual hold deviates from the theoretical. Over a season, books manage exposure aggressively enough that realized hold tracks theoretical.
Different markets carry different vig. NFL and NBA point-spread markets typically run 4-5% vig. Two-way money lines run higher, especially on heavy favorites. Player props can carry 8-15% vig. Futures markets can carry 25% or more. Vig is the single biggest cost a bettor pays, and a strategy that doesn't account for it will conclude it has an edge when it doesn't.
Strip the vig and you get the no-vig fair odds — the book's clean estimate of the true probability. Comparing a bettor's price against no-vig fair odds rather than against the raw price is the standard way to measure CLV. Comparing against the raw price flatters the bettor by counting the vig as edge they captured.
Our EV calculations are run against no-vig fair odds, not against the offered price. Comparing model probability to the raw price would inflate edges by the vig and lead to systematically over-confident publishing.