During the 2023-24 NBA season, Messenger is taking a look at who is most likely to reach the Finals and who is most likely to win the title. The following ratings are updated daily during the NBA season. To learn more about how it works, check out our explainer below.
Messenger’s 2023-24 NBA predictions
How our NBA model works
taking inspiration from Baseball-Reference’s MLB ForecastThis NBA model keeps going schedule-adjusted ratings Scores for each team based on their performance on offense and defense (per 100 possessions), the location of each game and the quality of their opponents. The ratings are then converted into win probabilities for each game on the schedule, and the rest of the season is simulated 5,000 times to find who is most likely to win the NBA championship.
A different problem for the NBA (compared to MLB) is that we maintain two separate ratings for each team, which have different forecasting objectives.
First of all, we keep a regular season ratingswhich is based on a team’s most recent 50 games – including the end of the previous season in terms of early season ratings – and is went back from the meaning By 15 games. It is used to create win probabilities for regular season games, and implies that a team’s recent form during the regular season matters a lot (not surprising, but still interesting to look at the quantity. Is).
For each team, we also have one playoff ratings, which is used to simulate playoff games. It is based on the team’s most recent 110 games, and does not include any regression-to-the-mean factors, although postseason games receive triple the weighting in the ratings. This means that ratings pay more attention to a team’s performance when looking toward the playoffs, which makes sense. In terms of research showing that there is a team Previous playoff resume matters a lot Relative to the regular season.
Another factor that gets included in simulation is uncertainty. The odds of any two teams winning in a regular season or playoff game depend on how far in the future it is, with tomorrow’s game considered more certain than a game eight months from now.
The challenge for any model is to strike a balance between simplicity and complexity, which may mean deciding how much relevant information to include.
This forecasting style is even simpler than Which was previously maintained by FiveThirtyEight, which was based on player ratings and accounted for trades, injuries and other player activities. Our basic model will not be able to accommodate those differences immediately. However, this will evolve over time as the rating sample includes a mix of players that the team will use going forward.
To help mitigate some of this, the default for our model is what we call composite modeWhich blends pure ratings-based forecasting simulations with underlying betting-market probabilities from fanduel, In theory, this should steer our predictions toward more qualitative factors like injuries and turnovers, which a statistical power rating-based system can’t easily take into account.
However, if you want to see the pure stats-based version, you can toggle the interactive above Messenger-only modeWhich removes the betting-market feature.
To learn more about how such models work, you can read about baseball-reference models HereAnd This is a lecturer The underlying mathematics behind the rating methodology used by both systems. For a history of the ratings I helped create at FiveThirtyEight, Read this explainer, And for more information on the logic and science behind power ratings, I would recommend the books “Who’s #1?: The Science of Ratings and Rankings“, by Amy N. Langville and Carl D. Meyer, and “mathletics“, by Wayne L. Winston, Scott Nestler and Constantinos Pellekrinis.