NBA Play-In Odds: How Play-In Tournament Prices Are Built
NBA Play-In odds come from single-game win probabilities multiplied through a fixed bracket. Seeds 7 and 8 need one win from two games; seeds 9 and 10 need two straight wins, the 9 seed with one of them away and the 10 seed with both. So the higher seeds qualify far more often, and every price depends on games a team does not play.
How the Play-In Tournament works
The format below was checked against NBA.com on 2026-11-05. The NBA adopted the Play-In Tournament as a permanent path to the playoffs on July 12, 2022, after using it in the 2020-21 and 2021-22 seasons. The 2026 edition ran from April 14 to April 17, with the playoffs starting April 18. No change to the format had been announced for 2026-27 at the time of checking.
In each conference, the teams finishing 7th through 10th play three single games for the 7th and 8th playoff seeds:
- Game A: No. 7 hosts No. 8. The winner takes the 7th seed.
- Game B: No. 9 hosts No. 10. The loser is eliminated.
- Game C: The loser of Game A hosts the winner of Game B. The winner takes the 8th seed; the loser is eliminated.
Seeds 7 and 8 need to win one of two games. Seeds 9 and 10 must win two in a row. The 9 seed hosts the 9 versus 10 game and then travels for the second; the 10 seed is away for both. That asymmetry drives every price.
Which markets the Play-In creates
Single-game lines. Once the matchups are fixed, each Play-In game gets an ordinary moneyline, spread and total. The Game C opponent is unknown until Game B is played.
Qualification and seeding props. Some books offer yes/no prices on whether a team makes the playoffs, or prices on final seed.
Series and futures. The Play-In winners enter best-of-seven series against the 1st and 2nd seeds, so their conference and title prices inherit the Play-In uncertainty.
Listings and settlement rules vary by book.
Why one game is so much noisier than a series
A best-of-seven series filters out single-game variance. A single game does not. Take an illustrative team that wins any one game against a given opponent 60 percent of the time, and assume that probability is constant across games. Its chance of winning a best-of-seven series is the chance of reaching four wins before the opponent does:
0.6^4 x (1 + 4 x 0.4 + 10 x 0.4^2 + 20 x 0.4^3) = 0.1296 x 5.48, or about 71 percent.
In a single Play-In game, the same team is a 60 percent favourite and nothing more. This is why Play-In outcomes look upset-prone: a qualification price describes a short, high-variance path, not which team is better.
Pricing a single Play-In game
A model prices a Play-In game the same way it prices any NBA game: a rating difference, an adjustment for home court, and adjustments for availability and rest, turned into an expected margin and then into a win probability. Three Play-In specifics deserve extra attention.
Late-season information is distorted. Teams locked into a position may rest players in the final regular-season games, and teams fighting for position may not. Down-weight games where key players sat.
Availability is decisive. With one game deciding the outcome, a single star's status moves the price more than it would across a series. Model the probability of availability, not just the confirmed status, until it is announced.
Fatigue and travel between games. Game C is played a short time after Games A and B.
To compare a model price with a book, remove the margin first. Convert both sides to implied probabilities, add them, and divide each by the total. The gap between the no-vig figure and the model is the claim being made.
Pricing qualification: an illustrative probability tree
Qualification probability comes from multiplying through the bracket. The numbers below are illustrative, chosen only to show the mechanics. Each is the probability that the first-named team wins, with the first-named team at home:
- 7 beats 8: 0.60. 9 beats 10: 0.58.
- 7 beats 9: 0.68. 7 beats 10: 0.72.
- 8 beats 9: 0.60. 8 beats 10: 0.65.
No. 7 seed. It qualifies by winning Game A, or by losing Game A and then winning Game C against whichever team won Game B:
0.60 + 0.40 x (0.58 x 0.68 + 0.42 x 0.72) = 0.60 + 0.40 x 0.697 = about 0.879.
No. 8 seed. Same path, with the 8 seed's own Game C probabilities:
0.40 + 0.60 x (0.58 x 0.60 + 0.42 x 0.65) = 0.40 + 0.60 x 0.621 = about 0.773.
No. 9 seed. It must win Game B, then win on the road against the loser of Game A, which is the 8 seed 60 percent of the time and the 7 seed 40 percent of the time:
0.58 x (0.60 x 0.40 + 0.40 x 0.32) = 0.58 x 0.368 = about 0.213.
No. 10 seed. 0.42 x (0.60 x 0.35 + 0.40 x 0.28) = 0.42 x 0.322 = about 0.135.
The four figures sum to about 2.00, which is the check: exactly two teams advance. Note that the 9 seed's chances depend on Game A, a game it does not play. That dependency is why qualification prices have to be computed jointly rather than team by team.
From probabilities to fair odds
Fair decimal odds are one divided by the probability. In the illustration, the 7 seed's 0.879 converts to about 1.14, or roughly -725 in American odds. The 10 seed's 0.135 converts to about 7.40, or roughly +640.
A book's yes/no qualification market includes a margin on top of these, and props and futures often carry more margin than main game lines. Remove it before comparing. In a multi-outcome market, divide each implied probability by the sum across all outcomes. Proportional removal is only one method. Longshots often absorb a larger share of the margin, so a proportional adjustment can overstate their fair probability.
Before the bracket is set
During the season, nobody knows who will be 7th through 10th. A standard approach is a Monte Carlo simulation of the remaining schedule. Each simulated season produces a final standings table, including tiebreakers, and the Play-In tree above is applied to whichever four teams land in each conference's 7 to 10 band. Averaging across thousands of runs gives each team's probability of finishing top six, entering the Play-In, and qualifying through it.
Ratings in the simulation should carry uncertainty that widens with time remaining, otherwise it is overconfident about the standings.
Where model prices and market prices tend to disagree
Disagreements usually come from information, not mathematics: injury and rest news reach the market before most public models update. A large gap between a model and a no-vig market price is more often a missing input than an edge.
This page describes how prices are built and is not betting advice. Single-game outcomes are highly variable, and no model removes that.
Frequently asked questions
- How does the NBA Play-In Tournament work?
- Teams finishing 7th through 10th in each conference play single games. The 7 seed hosts the 8 seed, and the winner takes the 7th playoff seed. The 9 seed hosts the 10 seed, and the loser is eliminated. The loser of the 7 versus 8 game then hosts the 9 versus 10 winner, and that game decides the 8th seed.
- Why are the 7 and 8 seeds so much more likely to qualify?
- They get two chances and the 9 and 10 seeds get one. A 7 or 8 seed qualifies by winning either of its games. A 9 or 10 seed must win twice in a row, always away in the second game, and the 10 seed is away in the first as well. Even between evenly matched teams, that structure gives the higher seeds a large advantage in qualification probability.
- Does reaching the Play-In count as making the playoffs for betting?
- Usually a team must win its way through the Play-In to be in the playoff bracket, but settlement wording differs between books and markets. Some markets ask about the Play-In itself, others about the playoffs proper. Read the rules for the specific market before comparing its price with a model, because the two questions have very different probabilities.
- How do you convert a Play-In probability into fair odds?
- Divide one by the probability to get fair decimal odds. A team with an illustrative 0.879 chance of qualifying has fair odds of about 1.14, roughly -725 in American odds. A team at 0.135 has fair odds of about 7.40, roughly +640. Compare these with the market only after removing the book's margin.