Event-contract guide
How to Devig Sportsbook Odds, By Hand
Sportsbook moneylines aren't probabilities — they contain the book's margin (the 'vig'), so the implied probabilities of both sides add to more than 100%. Removing it is two lines of arithmetic. This is the exact method our screener uses, and you can do it with a phone calculator.
1. Convert each moneyline to an implied probability
For negative American odds: P = |odds| / (|odds| + 100). For positive: P = 100 / (odds + 100). Example: Eagles −160 → 160/260 = 61.5%. Cowboys +140 → 100/240 = 41.7%. (Decimal odds are simpler still: P = 1/odds.)
2. Notice they sum past 100% — that's the vig
61.5% + 41.7% = 103.2%. The extra 3.2 points is the book's margin. Quoted probabilities are all slightly too high; the book profits because you pay those inflated prices whichever side you take.
3. Divide each by the total (multiplicative devig)
Fair P(Eagles) = 61.5 / 103.2 = 59.6%. Fair P(Cowboys) = 41.7 / 103.2 = 40.4%. They now sum to 100%. This 'multiplicative' method spreads the vig proportionally and is the standard first tool. (Fancier methods like power devigging adjust for longshot bias; for two-way NFL/CFB moneylines the difference is usually fractions of a point.)
4. Use several sharp books and take the median
One book can be off or slow. Devig the same game at two to five sharp books (Pinnacle is the canonical benchmark) and take the median fair probability. The median resists one outlier line in a way an average doesn't.
5. Now compare to the event contract
Fair 59.6% vs a Kalshi ask of 55c (≈56.6c with fees) → about +3 points of edge. Fair 59.6% vs a 61c ask → negative EV; pass. That's the whole workflow: devig, consensus, compare against all-in cost.
6. Respect the caveats
Sharp lines are the best public estimate, not oracle truth — they move on news, and pre-game lines say nothing about in-play. Multiplicative devig overrates longshots, and not slightly: against the power method it prices a +300 dog about 2 points high and a +600 dog about 3 points high, while the two agree exactly on an even matchup. That gap is larger than the edges worth chasing, so on a lopsided game an "edge" can be an artifact of which method you picked — we take the lower of the two so an edge has to survive both. And any comparison is only as good as both quotes being current — stale inputs manufacture fake edges.
See the numbers live
Polykal runs this exact math continuously — de-vigged sharp consensus vs Kalshi prices for every NFL and CFB game, plus cross-platform spread alerts via Discord + webhooks.
Create an account →Free while paid alerts are still closed. Paste any Kalshi or Polymarket link and we'll find its counterpart and what the gap clears after costs, and keep a watchlist of the markets you care about.
Frequently asked
Why trust sportsbooks as the truth benchmark?
Because sharp books take millions in action from professional bettors and survive — their line is a price that informed money couldn't profitably beat. That doesn't make it perfect; it makes it the best free estimate of true probability an individual can access.
What's the difference between multiplicative and power devigging?
Multiplicative divides each implied probability by their sum. Power devigging solves for an exponent that normalizes them, which corrects better for favorite-longshot bias in lopsided games. For typical two-way football moneylines they agree within a few tenths of a point — start multiplicative.
Can this be automated?
Yes — it's deterministic arithmetic on public prices. Polykal's +EV screener runs this exact devig-and-median pipeline across sharp books for every NFL and CFB game and compares the result to live Kalshi asks net of fees, refreshed continuously. The output is analytics, not advice: it shows you the number, you decide what to do with it.
Impersonal market analytics and education computed from public prices. Nothing on this page is investment, betting, legal, or financial advice, or a recommendation to buy any contract. Polykal is not affiliated with Kalshi, Polymarket, or any sportsbook.