NBA play-in odds are not playoff odds: simulating the play-in tournament instead of guessing it

In basketball, unlike baseball, American football and hockey, the path to the playoffs is not straightforward. American sports have only two zones: finish above the cutoff and you're in, or below and you're out. However, basketball operates with three distinct zones, not two, and understanding these is crucial when analyzing odds.

The NBA play-in tournament acts as a bridge between the regular season and playoffs, fragmenting the playoff qualification into separate markets. The play-in tournament consists of a bracket where teams compete in one or two games to determine their seeding, with the losers being eliminated. Kalshi, a betting platform, lists these as distinct markets, with a clear distinction that qualifying for the play-in tournament does not automatically qualify a team for the playoffs.

To simulate the NBA play-in tournament, one must go beyond simply determining whether a team is in or out of the playoffs. The odds for the play-in tournament and playoffs are separate, often behaving almost oppositely. While a title contender is almost certain for the playoffs, they are almost certain to be out of the play-in. Conversely, a 44-win team could be as uncertain as a coin flip between the two.

To model this accurately, the simulation must account for each step of the process. This involves playing out the remaining regular-season schedule, assigning seeds, conducting the play-in games, and then running the full bracket. This approach provides multiple probabilities for each team: top-six, play-in, playoff, conference finals, conference title, and championship. These probabilities are derived from a simulation that runs thousands of seasons, producing a range of outcomes rather than a single prediction.

However, this model requires substantial data input. It uses the Pythagorean exponent specific to basketball, which better predicts future results than a team's record alone. It also incorporates uncertainty in team strength by drawing game outcomes from team ratings after each season. This distributional approach reduces the likelihood of extreme outcomes, such as a 100% chance of making the playoffs for a given team.

Despite the simulator's accuracy, there are discrepancies between the model and betting markets. For instance, the model might predict a 99% chance of making the playoffs for Oklahoma City, while the market might only reflect a 98% chance. This gap can be due to various factors, including offseason changes the model hasn't yet accounted for, such as free agency, trades, or injuries. The model advises caution until at least ten games have been played, labeling any gaps as 'watch' for observation.

The NBA Playoff Odds API, which includes the Monte Carlo Simulator & Value Bets, is a resource for such simulations. It pulls in standings and schedules from ESPN and betting prices from Kalshi's public API, providing a comprehensive tool for anyone interested in NBA odds and value betting.