See what your bet is actually worth.
Your book hides its cut in every line. Paste the market and actualodds rips the juice out: fair odds, true EV, and the exact stake to fire.
Free and instant, with nothing to sign up for!
Worst-case: methods to compare
Weighted average: method weights (%)
Bankroll, Kelly & display
Estimate correlation. Test the assumptions.
Singles and a parlay quote reveal a price adjustment. Separating juice from correlation needs assumptions.
-110/-110 = both sides, your pick first. +300/7% = one side with its own vig. Bare odds use the default below; /0% treats a price as fair. American or decimal odds.Examples load prices only and keep your assumptions. Snapshot prices, not live odds.
One-sided overround reconstructs the other side, then uses worst-case devig. Two-sided legs use their actual market.
This optional floor is a modeling choice, not an observed SGP margin. It can change the correlation’s sign. Uncheck to use only your total assumption.
Math & assumptions
Worst-case per leg: run Multiplicative, Additive, Power and Shin, then keep the lowest fair probability for your pick. This is conservative across those four methods, not proof of the true probability. Two-sided quotes must cover every outcome of the same market. The leg breakdown names the method driving each FV.
One-sided overround: reconstruct the opposite probability as 1 + v − q, then devig that market using the same four methods. For +300/7%, q = 25%, so the assumed other side is 82%. If the implied opposite reaches 100%, enter actual opposite odds or lower the vig. Bare odds use your default; /0% treats the quote as fair.
Expected hold: a one-sided hold assumption directly sets p = q × (1 − h); it does not get a second devig. The assumption selector also applies to total SGP vig. Assumed margins are not observed market margins.
Regular parlays: with both sides of every binary leg, each book sum is qYes + qNo. The sum across all multiplied parlay prices is the product of those book sums. Combined market overround = product(book sums) − 1; equal-payout margin = 1 − 1 ÷ product(book sums). This assumes the book multiplies the singles for every combination and the outcomes have compatible settlement. The selected ticket’s estimated margin can differ under Power or worst-case devig. No SGP juice assumption is applied to a regular parlay.
SGP juice: I = product of fair leg probabilities is independent FV. Q = product of quoted leg probabilities. The leg-implied margin factor is Q ÷ I; a total SGP factor is 1 + v for markup or 1 ÷ (1 − h) for hold. Manual mode optionally uses F = max(leg factor, total factor), not their product or sum. A calibrated profile uses its estimated factor directly, without that floor. Fair joint probability J = final quoted probability ÷ F. The fair legs are not devigged again. A 25% total markup equals a 20% expected hold.
What prices identify: let S be the final quote’s implied probability. The raw multiplier S ÷ Q equals (J ÷ I) × (F ÷ (Q ÷ I)): dependence and relative juice are mixed together. A longer payout can come from negative dependence, lower SGP juice, or both. Even unchanged odds can hide offsetting effects. The floor assumes SGP juice is at least the compounded leg margin; the prices do not establish that. When the floor sets F = Q ÷ I, the estimated all-win multiplier simplifies to S ÷ Q, exactly the raw price multiplier.
Two-leg correlation r: r = (J − pA × pB) ÷ √[pA × (1 − pA) × pB × (1 − pB)]. It is a coefficient from −1 to +1, not a payout percentage. Positive means the legs win together more often than independence, negative means less often, and zero means no residual correlation under the assumptions. Unequal leg probabilities can restrict the attainable r range.
3+ legs: report the all-win probability lift, not a Pearson r or average pairwise correlation. A positive all-win gap does not imply every pair has positive correlation; a zero gap does not prove mutual independence.
Probability lift: (J ÷ I − 1) × 100% is shown separately. It can exceed 100% and is not a correlation coefficient. Higher assumed juice lowers J and the inferred r or all-win lift; it does not necessarily make the estimate closer to zero.
Stress test: vary the total SGP margin over your selected interval, without the optional floor. Intersect that interval with the feasible joint probabilities and report the exact endpoint range, holding the fair leg probabilities fixed. The default interval is illustrative, not a confidence interval or a measured sportsbook range. The method comparison repeats this for alternative devig methods; worst-case singles do not provide a bound on correlation. Change per-leg assumptions too when only one side is available.
Better evidence: use simultaneous two-sided singles from the same book, or explicit fair-price references. For two legs, prices for all four mutually exclusive combinations (both win, only A, only B, neither) allow a joint market to be devigged under a stated allocation method. A single both-win quote cannot identify its true margin. Compare books using the same legs, rules and timestamp; different single prices alone do not prove different real correlations.
Impossible joint probabilities are flagged, not forced into range. Pushes, voids, dead heats, and three-way markets are not modeled. Use comparable leg and SGP prices from the same time.
Method references: adjusting bookmaker odds · covariance and correlation · joint probability bounds.
Independent parlay
This measures correlation plus relative juice, rounding and any promotion. It cannot establish the sign of correlation on its own.
Enter at least two legs to see the price before and after removing vig. No final odds required.
Vig removed · no correlation.
Quoted leg probabilities multiplied. Not FV.
Leg FV & devig methods
| # | Leg | FV | Method |
|---|
Enter the book’s SGP odds to estimate dependence under your juice assumptions.
This is implied by the prices and your vig assumptions, not measured correlation or a betting edge. Different margin allocation, stale prices, or rounding can also explain the gap.
Does the sign survive different juice?
Vary total SGP juice with fair legs fixed. This test does not apply the optional leg-margin floor.
Compare leg devig methods
| Leg method | r range | Zero at total juice |
|---|
Each row uses the selected total-juice interval. Per-leg vig assumptions stay fixed; this does not cover uncertainty in those inputs.
How the estimate changes with SGP vig
| Assumed | Applied | SGP FV | r |
|---|
Calibrate SGP juice from a complete market
Enter all 4 combinations for two binary legs or all 8 for three. Y means the named event happens; N is its exact complement. Nine outcomes would describe two three-way legs, which this section does not model.
Same sportsbook, same lines and timestamp, no boosts. All combinations must be mutually exclusive and cover every possible outcome. Use X only for a logically impossible combination; never for missing or rejected quotes. Pushes, voids and dead heats need extra states and are not covered.
1 − 1 / book sum. Not the book’s realized profit rate.
| Outcome | Quoted | Fair chance | Fair odds | Cell markup |
|---|
These marginals and correlations come from the same devigged joint distribution. They are method-dependent estimates. A complete 8-outcome table can estimate all three pairwise correlations; a lone 3-leg all-win quote cannot.
Saved references & juice at your target odds
Only references matching the sportsbook, market family, leg count and pregame/live settings above are pooled. Each complete market gets one vote. Similar payout alone is not enough: match the leg price mix and event structure too. Repeated snapshots of one event are not independent evidence.
| Reference / time | Overround | Action |
|---|
References stay in this browser. Export a backup to move them between devices. Applying a profile transfers only the estimated margin model; it keeps the target’s own leg probabilities and does not copy reference correlations. It replaces manual SGP juice, turns off the leg-margin floor, updates at each new target quote, and blocks mismatched leg counts or missing comparisons. After changing the reference group or method below, click Apply again to update the active profile.
Where your edge leaks
Take a coin flip. A fair book would post +100 both ways. Yours posts −110/−110. Bet both sides and you lay out $220 to collect $210. That missing $10 is the vig: a tax sewn into every price on the board. Devigging backs the tax out. It looks at every side of a market, strips the margin, and hands you the probability the book is actually pricing. Stack that fair line against a boost or promo and the verdict is instant: positive EV, fire. Negative, walk.
Your edge, played forward
An edge doesn't pay you on schedule. It pays you eventually, through swings that will test your nerve. Dial in your numbers and watch what a real bankroll does: the dashed line is the math, the wiggly ones are your life.
Devig methods, explained
Books don't spread their cut evenly: longshots usually carry more of it than favorites. Four methods, four opinions on where the juice hides. Same market, watch +650/−900. The calculator's default takes the worst case across all four, so a play that survives doesn't depend on you picking right.
Multiplicative
Shaves every side by the same ratio. Goes easy on longshots. +650/-900 → +675
Best for: tight two-way markets near even money: spreads and totals at standard juice.
Additive
Takes an equal bite of raw probability from each side, which lands harder on the longshot. +650/-900 → +757
Best for: two-way markets with a modest favorite: moneylines around −150 to −250.
Power
Fits an exponent to the whole market. Harshest on longshots, kindest to chalk. +650/-900 → +815
Best for: lopsided markets: heavy favorites, big dogs, alt lines and boosted longshots.
Shin
Models a book defending itself against sharp insider money. Matches additive on two-way markets, splits from it on three or more. +650/-900 → +757
Best for: multi-way markets: futures, outrights, first-scorer props.
Speak its language
Type odds the way you already think about them. Five short lessons, ten seconds each. Tap any example to load it live.
Legs & markets start here
A comma starts a new leg. A slash separates the sides of one market. Your side goes first. That's 90% of it.
| Your side, then theirs | +240/-310 | try |
| Parlay: stack legs with commas | -115/-105, +130/-155 | try |
| Multi-way market: list every runner | +360/+220/+700/+1100 | try |
| Backing two runners? Glue them with ^ | +475^+625/+900/-170 | try |
| Either bet can cash (independent) | -130/+105||-160/+130 | try |
| Leave yourself a note | +240/-310 // out by Sunday |
Only one side posted?
No other side on the board? Tell it the juice and it rebuilds the market for you. Already holding a devigged number? Type it alone and it's used as the fair value, untouched.
| Assume the juice yourself | +310/6% | try |
| Borrow the juice from a market you can see | +310/[-122/-102] | try |
| A lone price is your fair value, no devig | +707 | try |
| Fair probability works too | 55% | try |
Blend books & lines
One book's price is an opinion. Several books are a consensus. Mix them inline.
| Average a few books | AVG(-142,-138)/AVG(+118,+122) | try |
| Shortest or longest price wins | MIN(-145,-150)/+125 | try |
| You already know the hit rate | HIT(58%)/4% | try |
| Your line sits between posted ones: ladder whole markets, any number of rungs | INT(31.5; 29.5=-132/+110, 34.5=+208/-250) | try |
| One-sided rungs work too, just add juice | INT(25.5;23.5=-175,27.5=-115)/4.5% | try |
Boosts, promos & strange payouts
The Final Odds box prices whatever the promo desk dreams up. Whole payout tables included.
| A boosted price | +350 | try |
| Final odds quoted as a parlay | -135,-250 → it prices to +144 | try |
| Parlay insurance: 3 hit pays the parlay, 2 hit refunds half | 3=-260,-220,-190;2=0.5x | try |
| Risk-free bet (r = what the refund is worth to you) | 1=+250;r=0.65x | try |
| Win & get a bonus on top | 1=-240;w=0.6x | try |
| Get it win or lose | 1=-550;n=0.3x | try |
| Boost box does math | 100%*65% (full boost, paid in 65¢ bonus dollars) |
Correlated legs
Same-game legs move together, and pretending they don't will torch your EV math. Tell it how much.
| You know the correlation | 0.25 | try |
| Let the book confess: their SGP price vs their own legs | +155=-190,-120 | try |
| Full SGP market: all four combos, then each leg | 135/550/850/125=-125/-105,110/-135 |
The juice
Short reads on betting math. No picks, no fairy dust.
Get the next read first
One short piece on betting math when it ships. You'll never get picks or spam, and one click gets you out.
Devig your first boost in 60 seconds
The three-step routine that turns any "Boosted!" banner into a yes or a no before the line moves.
Read →Are odds boosts ever worth it?
One real boost, one marketing boost, one monster. The math that tells them apart before you tap.
Read →Which devig method should you use?
Four formulas, four answers, sometimes 140 points apart. Why worst-case is the one to trust.
Read →Why a winning bettor can lose for months
20,000 simulated bankrolls with a genuine 3% edge. A quarter were still down after 500 bets.
Read →How to turn a free bet into real cash
A $50 free bet is not $50. Where the value hides, and the odds range that cashes out the most.
Read →What is vig? The tax inside every betting line
The $10 the book keeps on every coin flip, what it costs you a year, and how to see through it.
Read →