Same game parlay, explained

Why the legs do not settle independently

An ordinary parlay across four games is four roughly unrelated questions. A same game parlay is one question asked four ways. If a quarterback throws for a lot, his receivers usually catch a lot; if a game goes to overtime, almost every counting stat in it rises. Those legs are correlated, which means knowing one outcome genuinely changes the probability of the next.

Correlation runs the other way too. Two running backs on the same team splitting one workload are negatively correlated: the carries that go to one do not go to the other, and a ticket built on both is less likely to land than the multiplication suggests, not more.

Why the book’s price is not the multiplied price

Multiply the four legs of a same game parlay together and you will usually get a bigger number than the book is offering. That gap is not the book being stingy: it is the book correcting an assumption the multiplication makes and the game does not honour. Books either reprice the combination through their own correlation model or refuse the combination outright, and which legs are allowed together differs by book and by market.

This is also why a “true odds” figure for a same game parlay is a number to be suspicious of wherever you see it. Computing one honestly needs a joint model of the game, not a product of marginals. Parlay Builder does not publish one, and the chat and the builder are built never to quote one. It is a real limitation and it is stated rather than papered over.

Worked example: the multiplied price against the quoted one

Illustrative numbers, invented to show the arithmetic. Not a real game, player or price, and not a pick or a prediction.

Two legs from the same game, each priced at decimal 1.90 on its own, and a book quote for the pair:

• Multiplied as if independent: 1.90 × 1.90 = 3.61.
• Suppose the book quotes the pair together at 3.20, a figure invented for the example.
• On a 10 unit stake that is 36.1 units back at the multiplied price and 32 at the quoted one.
• The 4.1 unit gap is the book’s correlation correction, written as a return.

What the example cannot tell you is which of the two numbers is closer to the truth about the game. Working that out needs a joint model of how the two legs move together, which is exactly the figure this site declines to publish. The multiplication is easy and wrong in a known direction, the book’s quote is the book’s own model, and neither is a forecast from here.

What the builder actually does about it

Two things, and both are product behaviour rather than copy. The AI parlay generator prefers legs drawn from different games when it assembles a ticket, because that is the case where the arithmetic it shows you is closest to true. And where a ticket does end up with two legs in one game, the card prints a second correlation-adjusted combined figure beside the independent one instead of replacing it, so you can see the size of the assumption rather than inherit it.

Neither of those makes a same game parlay a good bet or a bad one. They make the number on the card mean what it says.

Reading an SGP without the marketing

The useful questions are the boring ones. Which legs in this ticket share a game? Does the story that makes one leg land also make the next one land, or does it compete with it? Is the price the book is showing consistent with legs that move together, which is to say noticeably shorter than the multiplied one?

If a leg has a measurable history: how often this player has cleared this line, over how many games: that history is worth more than the narrative. Every prop on this site carries its own sample size for that reason.

Common questions

Try it on the live board

Every prop with its measured hit rate, its sample size and the live price beside it. See tonight’s ranked prop board.