The first time I priced a two-leg parlay by hand I got a number the venue disagreed with by a lot, and I assumed the venue was skimming. It was skimming, but that was not the whole story. Most of the gap was correlation, and once I understood where it came from I stopped trusting any parlay price I had not decomposed myself. If you trade event contracts, this is worth an hour of your attention, because the mistake is invisible from the outside and it always costs you in the same direction.
When multiplying probabilities is actually right
A parlay is a bet that several things all happen. You win only if every leg resolves yes. The textbook price is the product of the individual probabilities. Two legs at 40 percent each gives you 0.4 times 0.4, so 16 percent. Three legs at 50 percent gives you 12.5 percent. Simple, and for a certain kind of parlay, correct.
The multiplication rule holds under one strong condition. The legs have to be independent. Knowing the outcome of one leg tells you nothing about the others. A football game in one league and the weather in a different country next month are, for practical purposes, independent. If your legs really do not talk to each other, multiply away and the answer is honest.
The problem is that most parlays people actually want to build are stuffed with legs that do talk to each other. And the moment they do, the product of the probabilities is the wrong number, sometimes badly wrong, and the direction of the error is not obvious until you think it through.
Where correlation breaks the math
Correlation means the legs move together, or against each other. Positive correlation is when a yes on one leg makes a yes on the others more likely. Negative correlation is the reverse. Both break the simple multiplication, and they break it in opposite directions, which is exactly why you have to know which one you are holding.
Say you build a parlay out of two contracts that are really measuring the same underlying thing from slightly different angles. A rate decision and a contract on whether a certain inflation print comes in hot. Those are not independent. If inflation runs hot, the rate decision leans one way, and if it runs cool, it leans the other. When you multiply their standalone probabilities you are pretending the second flip is fresh information when it is mostly a rerun of the first. For positively correlated legs, the true joint probability is higher than the product. The naive math underprices the parlay, which means if the venue also multiplies naively, you are getting the combined exposure cheap.
Now flip it. Take two legs that are mutually exclusive-ish, where a yes on one makes a yes on the other less likely. Two different candidates each winning the same election. Two different teams each winning the same bracket. Bundle those into an all-must-happen parlay and the true joint probability is far below the product, often near zero, because they are fighting over the same outcome. Here the naive math overprices, and a venue that sells it to you at the product price is handing you a contract worth almost nothing.
The rule of thumb I keep in my head is short. If a yes on leg A would make you more optimistic about leg B, the parlay is worth more than the product, and multiplying makes it look cheap. If a yes on A would make you more pessimistic about B, the parlay is worth less than the product, and multiplying makes it look expensive. Independence is the only case where the product is the truth, and independence is rarer than it feels.
How venues embed margin, and why it hides so well
Venues that sell combined contracts as a single product do not have to show you the leg-by-leg math, and that is the point. The quoted price already bundles three things you would otherwise see separately. There is the honest joint probability. There is the venue's correlation assumption, which may not match yours. And there is margin, the spread they keep for making the market.
On a single contract you can eyeball the margin. Yes plus no should sum to a dollar, and the amount over a dollar is roughly the vig. On a packaged parlay that check is much harder, because you cannot see the individual leg prices they used, so you cannot tell how much of the number is real probability and how much is the house. A parlay can look like a generous payout while quietly carrying several times the margin of the same exposure built from single legs, because the fee compounds across legs and nobody shows you the seams.
This is why I almost never take a packaged multi-leg product at face value. I price the legs myself from the single-contract markets, form my own view on how correlated they are, and only then compare against what the venue wants for the bundle.
Build it from legs, or buy the bundle
The practical question is always the same. Do I buy the combined contract, or do I build the same payoff from single legs myself. Here is the workflow I run.
- Price each leg from its own single-contract market. Not from the parlay. From where the standalone contract trades.
- Multiply them for a first-pass number. This is your independence baseline, and it is wrong on purpose, a reference point rather than an answer.
- Ask which way the legs are correlated. Same underlying driver means positive, adjust the fair value up. Competing for one outcome means negative, adjust it down. You will not nail the exact number, and you do not need to. Direction and rough size are enough to avoid the trap.
- Compare the venue's bundle price against your correlation-adjusted number. If the bundle is cheaper than your honest estimate, and you can actually get filled on the legs, the bundle might be the better buy. If it is richer, build it yourself.
- Watch execution. Building from legs means you can get filled on three of four and stuck on the last one, which leaves you with an exposure you did not want. The bundle removes that risk, and that convenience is worth something. Sometimes it is worth the margin. Sometimes it is not.
The failure mode I have watched people walk into, more than once, is the negative-correlation parlay that looks like free money. Two outcomes that cannot really both happen, packaged together, quoted at the product of their probabilities, so the payout looks fat. They see a low price and a big multiple and they take it, and the contract was worth close to nothing the whole time. The mirror-image failure is skipping a positively correlated parlay because the naive math made it look too expensive, when it was actually the cheap side.
None of this needs heavy machinery. It needs the discipline to decompose before you buy, and a habit of asking whether a yes on one leg would move your view on the next. When I want to see the standalone leg prices side by side instead of squinting at a bundle, I pull them up on Blockcircle and price the parlay myself, because the number I trust is the one I built. If you only remember one thing, remember that correlation, not the payout multiple, decides whether a parlay is a bargain or a trap, and the venue is under no obligation to tell you which one you are looking at.