Edge · Parlay Pricing

What your parlay actually costs

Build a parlay and see it priced honestly: the vig stripped out of every leg, the combined chance, the payout that would be fair against the payout on offer, and the sportsbook's cut — which compounds with every leg you add and appears on no bet slip anywhere. This page makes no prediction about any game, and does not claim to beat the market. It tells you the price of the bet you were going to make.

Build across games from the consensus close and see the cut compound with every leg. The one mode that needs no odds feed: everything in it is arithmetic on two prices.

At −110 a leg · 2 legs
8.88%
keeps 8.88 of every 100
how it compounds
one more leg
8.88% 13.03%
↑↓movespaceadd leg⌫dropMsort by margin
The one identity behind this page

De-vig each side by scaling both so they add to 100%. Then for any slip, the offered payout multiplied by the combined fair chance equals exactly 1 ⁄ Π(1+m), where each m is that market's margin.

The leg's own price cancels out. So the hold depends only on the margins — it needs no assumption about whether results are related, which is why it is still exact on a same-game slip where the combined chance is not.

Legs at −110Book keeps
14.55%
28.88%
313.03%
416.98%
624.36%

Exact fractions of 1 − (21/22)^k. Three legs is 1 − 9261/10648.

Why there is no edge column

Measured over 1,871 completed regular-season games, 2019–2025:

Home teams covered48.74%of 1,828
The over hit49.03%of 1,854
Favourites won66.52%of 1,858

Break-even at −110 is 52.38%. The favourite figure looks enormous beside that and is not: a favourite is priced at −200, not −110, where break-even is 66.7%.

npm run study:parlay pre-registered the bucketed version of that question before it ran and rejected it. So there is no edge column, no value badge and no “+EV” language here — that absence is a measured result, not an omission.

Same game, and why the number is withheld

Two legs in one game move together, and the direction flips depending on how the game goes: a blowout cover comes with the over, a 17–13 grind-out cover with the under. One correction factor would be wrong in both directions.

So nothing is discounted. You still get every per-leg chance and the full hold — those are exact either way. What replaces the combined chance is the range it must lie in, which is provable, plus how often that exact pair has actually landed since 2019.

The size of the thing being refused: a favourite of 3 to 6.5 covering and winning happens 17.7 points more often than independence would predict. Across different games the same measurement is 0.2 points. Two orders of magnitude apart, which is why the two are treated differently rather than identically.

Same-game prop pairs, counted

The panel above says the combined number is withheld because the dependence changes sign. Here it is, counted rather than estimated, over 24,621 qualifying player-weeks from 2019 to 2026. Each row is one pair of roles, how often they both landed, and the product of the two rates measured in the same games — so the last column is the dependence with the base rates already taken out.

Counted co-occurrence for 23 pre-registered role pairs, with each pair's shared denominator and its gap against independence.
PairBoth landedIf independentGamesDependence
WR1 over 50 receiving yards + WR1 over 4 receptions43.9%1622 of 369729.0%3,697+14.88pp
the QB over 250 passing yards + WR1 over 60 receiving yards28.4%653 of 229920.5%2,299+7.91pp
the QB over 225 passing yards + WR1 over 60 receiving yards34.0%781 of 229926.4%2,299+7.60pp
the QB over 250 passing yards + WR1 over 70 receiving yards24.2%556 of 229916.7%2,299+7.47pp
RB1 over 50 rushing yards + RB1 scores31.8%1152 of 362424.4%3,624+7.39pp
the QB over 225 passing yards + WR1 over 70 receiving yards28.9%664 of 229921.5%2,299+7.38pp
the QB over 275 passing yards + WR1 over 60 receiving yards22.3%512 of 229915.0%2,299+7.26pp
the QB over 275 passing yards + WR1 over 70 receiving yards19.2%442 of 229912.2%2,299+6.99pp
the QB over 225 passing yards + WR1 over 50 receiving yards39.0%897 of 229932.1%2,299+6.92pp
the QB over 250 passing yards + WR1 over 50 receiving yards31.8%731 of 229924.9%2,299+6.85pp
the QB over 250 passing yards + WR2 over 50 receiving yards23.9%516 of 215617.4%2,156+6.49pp
the QB over 275 passing yards + WR1 over 50 receiving yards24.3%558 of 229918.3%2,299+6.00pp
the QB over 250 passing yards + WR1 over 4 receptions29.8%684 of 229923.8%2,299+5.96pp
the QB over 225 passing yards + WR1 over 40 receiving yards43.9%1010 of 229938.0%2,299+5.92pp
the QB over 250 passing yards + WR1 over 40 receiving yards35.4%814 of 229929.5%2,299+5.86pp
the QB over 250 passing yards + TE1 over 40 receiving yards23.4%501 of 213918.0%2,139+5.42pp
the QB over 275 passing yards + WR1 over 40 receiving yards26.8%615 of 229921.6%2,299+5.11pp
the QB over 275 passing yards + WR1 scores14.5%333 of 229910.9%2,299+3.55pp
WR1 over 50 receiving yards + WR2 over 50 receiving yards20.6%701 of 339720.7%3,397about independent
WR1 over 50 receiving yards + TE1 over 40 receiving yards21.7%736 of 339922.0%3,399-0.37pp
WR1 over 50 receiving yards + RB1 over 50 rushing yards30.8%1108 of 359631.6%3,596-0.76pp
the QB over 250 passing yards + RB1 over 50 rushing yards22.2%500 of 225323.9%2,253-1.71pp

Two rungs on one player are not two legs. WR1 over 50 receiving yards and WR1 over 70 receiving yards is nested: every game clearing 70 already cleared 50, so the pair lands exactly as often as the harder leg alone — 1374 of 3697, 37.2%. That is an identity, not a measurement, and it holds with no assumption about dependence whatsoever. A reader stacking them has bought the harder leg twice and paid for two legs; nothing has been diversified. It is also the only shape where the Fréchet interval this page falls back to is tight — its upper end, 37.2%, is the answer rather than a ceiling.

Counted, never fitted. A leg names a position and an opportunity rank rather than a player — “WR1” is the most-targeted qualifying receiver by his target rate entering the game, so the rank is a pre-kickoff fact and never the stat being counted. Named pairs were measured first and could not support this: their median is four shared games, and only two pairs in the league clear the 100-game minimum. Every one of the 23 pairs was declared before it was counted, and all of them cleared the minimum. Same-team pairs only: a quarterback against the opposing receiver needs a schedule join and is not counted here. None of these numbers prices a slip. Built Nov 3, 6:01 PM EST from nflverse.

How the suggested slip is chosen

By one stated rule over two named quantities, both printed in their own columns: maximise the offered payout, subject to the combined chance clearing the band's floor, at most one leg per game and at most 6 legs. There is deliberately no third number — no risk score, no confidence rating, nothing weighted together. A blended score is a number with no unit that nobody can argue with because nobody can read it, and this repo refuses to build one; the same rule shapes the start/sit board and the DFS optimizer.

Preferring well-priced legs is not a fourth rule either — it falls out. For a given fair chance, a lower-margin market offers a bigger payout, so maximising payout under a chance floor already takes the better-priced side. The search is exhaustive at this board size, so the headline slip is the highest-paying one at that floor rather than the best one found; if it ever stops being exhaustive the page says “best found” instead.

The band floors — 25%, 12.5%, 5% — are a convention and not a measurement. No data on anyone's risk tolerance exists here, so the floors are chosen to be legible and to separate, and the actual combined chance of whatever got built is always printed beside them. The leg cap is a search bound, not a recommendation.

About this page

Goal Line is not a sportsbook and accepts no wagers. Nothing here is a bet, an offer to take one, or advice to place one. This page takes prices that already exist and does arithmetic on them: it removes the sportsbook's margin from each side, multiplies the legs, and shows you what the payout would have to be to break even against what is actually offered. The difference is the book's cut.

No claim to beat the market. Measured over 1,871 completed regular-season games from 2019 to 2025, home teams covered the spread 48.74% of the time, the over hit 49.03%, and favourites won 66.52% of games against prices that already implied roughly that. Break-even at −110 is 52.38%. These markets are efficient, this page contains no prediction of any game, and there is no “value” or “+EV” figure anywhere on it because none was earned.

Where the lines come from. Lines come from nflverse's published schedules, which carry an unattributed consensus closing price assembled from several sportsbooks. They are refreshed every six hours and carry no movement history — there is one set of numbers per game and no record of what they were before. No sportsbook provides, endorses or is affiliated with any of this.

Legality and age. Sports betting is regulated differently in every jurisdiction and is illegal in some. You are responsible for knowing the law where you are, and for being old enough to bet there — which is 21 in most US states that permit it, and 18 in a few. Nothing on this page is available to anyone under that age.

If betting has stopped being fun. In the US, call or text 1-800-GAMBLER (1-800-426-2537), text 800GAM to 53342, or visit ncpgambling.org. These are the National Council on Problem Gambling's own lines and they are free and confidential. Outside the US, look up your national problem-gambling council — we would rather send you to find the right number than print one here we have not checked.

Currently unauthenticated and unpaywalled. See Players for the weather and injury facts behind the rest of the app — none of which touches this page, which contains no projection of ours at all.