How Fair Odds Are Calculated

Fair odds are the price at which a bet would break even forever: no profit, no loss, over an infinite number of identical wagers. They are the benchmark any book price is measured against. Anything a sportsbook offers that is better than fair is, by definition, a positive-EV bet.

Step 1: implied probability

Every price already contains a probability. To read it out, convert the American odds. A -150 favorite implies 150 / (150 + 100) = 60%. A +150 underdog implies 100 / (150 + 100) = 40%. That is the win rate the price is quietly assuming.

Step 2: remove the vig

Add up the implied probabilities on both sides of a market and they sum to more than 100%. A standard -110 / -110 line totals 104.76%, and that extra 4.76% is the vig, the house margin baked into the price. Fair odds are what is left once that margin is stripped out and the two sides are scaled back to a true 100%.

Remove the vig yourself

Steps 1 and 2 above work on any two-sided line. The example below carries one market through both steps, a favorite priced at -125 and an underdog priced at +105.

A -125 price risks 125 to win 100, which implies 125 / 225 = 55.6%. A +105 price risks 100 to win 105, which implies 100 / 205 = 48.8%. Add the two probabilities and the total comes to 104.3%. The extra 4.3% over 100 is the vig baked into both prices.

Dividing each side by 104.3% strips that margin back out. The favorite’s fair probability is 55.56 / 104.34 = 53.2%. The underdog’s fair probability is 48.78 / 104.34 = 46.8%. Converted back to American odds, fair is -114 for the favorite and +114 for the underdog.

This is the whole computation behind any no-vig calculator. The numbers Fairline shows are this same arithmetic, run against every book it carries.

Step 3: use the fair price as a reference

Fairline labels the source behind each fair price and places it beside current book prices for neutral line shopping. Automatic model-versus-book thresholds are no longer presented to users. Internal measurement continues for calibration, closing-line tracking, and the tracked MLB dog record. The model never reads book lines to build its projection. It prices the game on its own, then compares against the books only at the very end.