The biggest number on a parlay slip is rarely the most important one.
A four-leg ticket may contain nothing that looks reckless. Suppose each selection has a 70% chance of winning. If the outcomes are independent, the probability of cashing is 0.70 × 0.70 × 0.70 × 0.70, or 24.01%. Every leg can appear sensible while the full ticket still loses about three times out of four.
That is the crucial shift: the relevant question is not whether each pick looks likely on its own, but whether every required outcome will occur. A parlay is all-or-nothing—one failed leg reduces the return to zero. The advertised payout may command attention, but the combined hit rate reveals how demanding the bet really is.
Every leg must win
A parlay cashes only when every selection wins. Three winners on a four-leg ticket still produce a losing bet, so the relevant event is the intersection of all four outcomes—not the percentage of legs expected to win. This all-or-nothing structure is central to how parlay bets work.
For independent events, the formula is:
P(parlay wins) = P(leg 1) × P(leg 2) × … × P(leg n)
Suppose a three-leg parlay contains estimated win probabilities of 80%, 65%, and 75%. Written as decimals, the calculation is:
0.80 × 0.65 × 0.75 = 0.39
The estimated probability of cashing the entire ticket is therefore 39%. Averaging the three probabilities would give 73.3%, but that figure does not represent the chance that all three occur together.
The multiplication rule assumes the legs are independent, meaning one result does not change another’s probability. Because each leg probability is itself an estimate, the final product should also be treated as an estimate rather than a guaranteed true chance.
Convert odds before multiplying
Sportsbook odds convert to implied probability—the break-even rate represented by the price. That rate includes the bookmaker’s margin, so it should not automatically be treated as the selection’s true chance.
Decimal odds
For decimal odds (D):
Implied probability = 1 ÷ D
Decimal odds of 2.00 imply (1 ÷ 2.00 = 0.50), or 50%. Odds of 1.50 imply 66.67%.
American odds
- Positive odds: (100 ÷ (odds + 100))
- Negative odds: (|odds| ÷ (|odds| + 100))
For example, +150 implies (100 ÷ 250 = 0.40), while -200 implies (200 ÷ 300 = 0.6667).
These conversions produce sportsbook-implied inputs. A true-probability estimate instead comes from handicapping, a statistical model, or a market estimate adjusted to remove vig. Either type can be multiplied, but the result answers a different question: implied probabilities describe the posted prices; true probabilities estimate how often the parlay should actually win.
Convert 70% to 0.70 before calculating. For two legs at 70% and 60%, use (0.70 × 0.60 = 0.42), then convert the result back to 42%. Multiplying 70 × 60 produces 4,200, which is not a probability.
Remove the vig before multiplying
A sportsbook’s opposing prices usually imply probabilities totaling more than 100%. For example, both sides priced at -110 convert to 52.38% each, for a combined 104.76%. The extra 4.76 percentage points represent the bookmaker’s margin, often called vig or overround.
A simple proportional no-vig adjustment divides each implied probability by their combined total:
No-vig probability = implied probability ÷ total implied probability
For two -110 sides, each adjustment is 52.38% ÷ 104.76% = 50%. With -150 and +130 prices, the raw probabilities are 60% and 43.48%. Dividing each by 103.48% produces approximately 57.98% and 42.02%.
This is a practical starting point for accounting for vig in probability calculations, but it assumes the margin is distributed proportionally. Sportsbooks may shade one side because of customer demand, market risk, or information advantages.
No formula can rescue weak inputs. A parlay estimate is only as credible as its leg probabilities, which should reflect injuries, expected roles, matchup conditions, current information, and uncertainty. When an estimate feels fragile, calculating a reasonable low-to-high range is more honest than presenting one precise percentage.
Three unequal legs: 60%, 55%, and 52%
- Write each estimate as a decimal
The three win probabilities become 0.60, 0.55, and 0.52. This assumes the legs are independent.
- Multiply the first two legs
0.60 × 0.55 = 0.33, so those two selections have a 33% chance of both winning.
- Include the third leg
0.33 × 0.52 = 0.1716. Among worked examples using the parlay probability formula, this shows how even moderately likely legs can produce a much lower combined probability.
- Convert the result to a percentage
0.1716 × 100 = 17.16%. That is the estimated chance that all three legs win and the parlay cashes.
- Express the long-run frequency
A 17.16% probability corresponds to roughly 17 winners per 100 comparable bets, or about one winner in every 5.83 attempts over a very large sample.
- Calculate fair decimal odds
Fair decimal odds are the reciprocal of probability: 1 ÷ 0.1716 = 5.83. This is a break-even price before considering bookmaker margin.
The calculation is only as reliable as the leg estimates and the independence assumption.
A 17.16% chance does not mean exactly one of every six tickets will win. Several losses—or several wins—can occur consecutively. The estimated frequency becomes meaningful only across many comparable bets, and actual results may still differ if the inputs are inaccurate or the legs are correlated.
Why “safe” legs become a risky parlay
A 70% chance sounds reassuring for a single selection. Stack several such selections, however, and the ticket becomes an underdog surprisingly quickly. Assuming the legs are independent, the cashing probability is 0.70 raised to the number of legs.
| Number of 70% legs | Parlay cashing probability |
|---|---|
| 1 | 70.00% |
| 2 | 49.00% |
| 3 | 34.30% |
| 4 | 24.01% |
| 5 | 16.81% |
With four legs, a ticket made entirely from individually likely outcomes cashes only 24.01% of the time—roughly once in every four attempts under the model. A fifth identical leg lowers that chance to 16.81%, or about once in six.
The decline happens because every new selection adds another condition that must be satisfied. Multiplying the current probability by 0.70 preserves only 70% of the outcomes that were still capable of producing a win; the other 30% are eliminated by the added leg.
This is why adding a “safe” pick never makes a parlay safer. Unless a leg is certain, its probability is below 100%, so multiplying by it must reduce the ticket’s overall chance.
From probability to expected value
A +600 parlay has decimal odds of 7.00, so its break-even probability is:
1 ÷ 7.00 = 14.29%
The earlier estimate of 17.16% sits 2.87 percentage points above that threshold. On paper, this suggests a positive edge: the bettor’s estimated chance is higher than the chance implied by the offered price.
Expected value makes the comparison concrete. At +600, a $1 stake earns $6 profit when it wins and loses $1 otherwise:
EV = (0.1716 × $6) − (0.8284 × $1) = +$0.2012
That equals a theoretical 20.12% return per dollar staked over a very large number of equivalent bets. The same framework can apply the probability formula when calculating potential payouts, provided stake, profit, and total return are kept distinct.
This is not a prediction that one ticket will win, nor proof that +600 is mispriced. The apparent value depends entirely on the 17.16% estimate being credible. Biased leg probabilities, overlooked correlation, lineup changes, or stale information can erase the edge—and may turn a seemingly positive wager into a negative one.
Ordinary multiplication works only when the legs are independent: learning that one result occurred does not change the probability of another. Same-game selections often fail that test.
For two dependent legs, the correct formula is:
P(A and B) = P(A) × P(B | A)
Here, P(B | A) means the probability of B given that A wins. If both legs have estimated 50% probabilities but B becomes 70% likely when A occurs, their joint probability is 50% × 70% = 35%, not 25%.
Relationships can work in several directions:
- Positive: A quarterback passing-yards over can make a receiver’s receiving-yards over more likely.
- Negative: Two players competing to be the game’s leading scorer can reduce each other’s chance of winning that market.
- Mutually exclusive: Home-team moneyline and away-team moneyline cannot both win in a two-way market, so their joint probability is 0%.
For longer parlays, every added term is conditional: P(A) × P(B | A) × P(C | A and B). Estimating those conditions usually requires historical joint data or a simulation model, not merely each leg’s standalone probability.
A quoted probability such as 18.43% may look rigorous, but it is only as credible as the correlation assumptions behind it. Same-game parlay pricing may reveal that a sportsbook adjusted for dependence; it does not establish the ticket’s true probability.
A repeatable parlay evaluation
- Record every leg and settlement rule
List each selection, market, price, and estimated win probability. Note whether a push voids one leg, reduces the parlay, or causes the entire ticket to be refunded.
- Convert prices and remove vig
Turn both sides of each market into implied probabilities, then normalize them to 100%. Multiplying sportsbook probabilities without this adjustment compounds the bookmaker margin.
- Check whether the legs interact
Independent legs can be multiplied directly. For related outcomes, use conditional probabilities or a correlation-aware estimate; otherwise, the result may materially overstate or understate the chance.
- Calculate probability and fair odds
Multiply the adjusted leg probabilities, retaining several decimal places until the end. Convert the final probability to fair decimal or American odds only after the calculation to limit rounding drift.
- Compare fair value with the offer
Set the estimated fair price beside the sportsbook payout and account for uncertainty. The estimate can also support probability-based hedging decisions if circumstances change after the bet is placed.
Recalculate whenever a leg, price, or settlement condition changes.
Losing only one leg does not show that the parlay was “almost certain,” nor does it make the next ticket more likely to cash. Results should be judged against the original probability and price over many bets, not by how close one slip appeared.
A parlay estimate is strongest when its inputs are vig-free, its dependencies are addressed, and its settlement rules are explicit. Even then, it remains a disciplined estimate—not a guarantee—and small input errors can grow as more legs are multiplied.