How TeamRankings Makes NBA Preseason Rankings

Giannis (Photo by Adam Davis/Icon Sportswire)

This post describes our methodology and process for creating NBA preseason rankings for all 30 teams.

We’re data people, so as one should expect from TeamRankings, our NBA preseason rankings are primarily driven by stats and modeling and not less objective methods like film study or media scouting reports.

(That’s not to broadly denigrate more subjective methods of analysis. But when it comes to preseason NBA rankings, many narratives exist that aren’t supported by hard data.)

Before we dive into the details of our approach, let’s first cover a few basics.


Editors Note: You can sign up for algorithmic NBA picks and predictions (game winner, point spread, over/under, and money line) for all 2019-20 games on our signup page.


What Our NBA Preseason Rankings Represent

It’s important to know that our preseason rankings simply represent the rank order of preseason predictive ratings that we generate for every NBA team.

So the first step in our process is to calculate preseason team ratings.

Predictive Rating Definition

In simple terms, an NBA team’s predictive rating is a number that represents the margin of victory we expect when that team plays a “perfectly average” opponent on a neutral court.

This rating can be a positive or negative number; the higher the rating, the better the team. A rating of 0.0 indicates a perfectly average team.

How Ratings Translate To Predictions

Because our NBA predictive ratings are measured in points, the difference in rating between any two teams indicates the projected winner and margin of victory in a neutral-site game between them.

For example, our system would expect the Milwaukee Bucks, who have a 2019 preseason rating of +5.0, to beat an average NBA team by 5 points on a neutral court.

It would expect Milwaukee to beat Phoenix, which has a -3.0 rating, by 8 points on a neutral court. And Phoenix would be expected to lose to an average team by about 3 points.

Ratings Are More Precise Than Rankings

Understanding the nature of a predictive rating is helpful, because it is more precise than a ranking.

For example, the Los Angeles Lakers are +3.1, and the Golden State Warriors are +3.3. So yes, if you put a gun to our head and told us to rank order every team, we’d say Golden State is going to be a better team than the Lakers this season. But the difference is so small that it’s practically meaningless.

Meanwhile, the difference between the No. 8 Lakers and the No. 6 76ers is a full point in the ratings. That’s bigger than the difference between the No. 6 76ers and the No. 2 Bucks. So the Lakers are further from cracking the top 6 (even though they are ranked 8th) than any team from spots 3 through 6 is of cracking the top 2.

In short, don’t place too much stock in a team’s ranking. Ratings tell the more refined story.

Why We Make NBA Preseason Ratings

Once the season starts, our predictive ratings go on autopilot. As game results come in, our system automatically adjusts team ratings (and the resulting rankings) within a few hours of receiving a new box score.

Teams that win by more than the ratings had predicted see their ratings increase. Teams that suffer worse than expected losses see their ratings drop. Software code controls all of the adjustments and no manual intervention is required.

Generating preseason ratings, however, involves a more labor-intensive process that we go through before every new season starts. In short, we are trying to pre-calibrate our NBA predictive ratings system. We want to give it a smarter starting point than simply having every team start the season with a 0.0 rating.

Put another way, our preseason ratings are our first prediction of what we think every NBA team’s predictive rating will be at the end of the upcoming season. And we need to make that prediction before any regular season games are played.

How We Make NBA Preseason Ratings

For our NFL and college football ratings, we’ve done historical research to identify and properly value team-level stats that are highly correlated with success in an upcoming season. We then create models using those stats, and blend our model output with betting market info to create our final preseason ratings.

Once upon a time, we took that same approach with our NBA preseason ratings. However, we found that our projections using that approach were less accurate than the market, and less accurate than other, more advanced player-level systems with publicly available projections.

We suspect that the main reasons for this are:

  • The relative impact of individual player skill compared to the coaching system and franchise/program history is much higher in the NBA than it is in NFL or college football. (The main exception to this is the quarterback position, which is the one position that we do explicitly adjust for in our NFL preseason ratings.) To use a more concrete example, knowing that LeBron James is now on the Lakers, and using ratings that incorporate his projected performance, is going to be a lot more accurate than a team-level approach that assumes how the Cleveland Cavaliers and Los Angeles Lakers played two seasons ago will have much impact at all on what happens in 2019-20.
  • The NBA analytics community has made greater strides in player all-in-one value metrics than their counterparts studying the NFL. Part of this simply stems from better data being widely available for NBA players. (Go ahead, try to evaluate NFL offensive linemen based only on box score info.)

Both of these factors make constructing NBA team-level models relatively less fruitful than player-level models.

So rather than stick with our old team-level model, we now create our NBA preseason ratings by blending several player-level models from other sources with team win total projections from the betting market. We determined which player-level models to include based on their past accuracy in predicting season win totals, and we update the blend of models each season.

More specifically, here are the steps we take in compiling our ratings:

  • Collect win total projections from various sources (including betting markets, major media sites, and the APBR message board win totals prediction contest) from the past several years.
  • Use a regression model to determine what blend of those win totals has historically best predicted the actual number of wins for each team.
  • Use that model to create weighted average win totals for the upcoming season.
  • Calculate the preseason rating values that will produce the target win totals from the previous step. (To do this, we start with every team rated 0.0, calculate expected win totals based on those ratings, compare the results to the target win totals, adjust the ratings for each team a little bit in the direction that ought to move their projection closer to the target total, and then repeat until the ratings stabilize.)

What we’re left with in the end is the predictive power ratings that are implied by the consensus win totals.

Conclusion

There are many different ways to make preseason rankings for the NBA. The approaches can vary greatly, from media power rankings to “expert” analysis, from building complex statistical models to making inferences from futures odds in the betting markets.

And speaking frankly, there’s plenty of crap out there. But there’s also no Holy Grail (yet).

Within ten seconds of looking over our preseason NBA rankings, you’ll probably find several rankings you disagree with, or that differ from what other “experts” or ranking systems think. That’s to be expected.

When the dust settles at the end of the season, our NBA preseason ratings, and the various projections we generate using them, will almost certainly be way off for a few teams. As happens every year, some teams simply defy expectations thanks to surprise breakout performances, while other teams are impacted by injuries, trading away superstars, or as we saw last year with New Orleans, a star player deciding to shut it down for the season. Some franchises, let’s be honest, also go into tank mode during a season if they feel there is nothing to be accomplished by accumulating wins that put them just outside the playoffs anyway. When and if that happens in specific cases can alter results.

Nonetheless, the primary goal of our preseason analysis is to provide a baseline rating for each team (or “prior” in statistical terms) that makes our system better at predicting regular season NBA games. We’re most concerned about the overall accuracy of the system — that is, how good it is at predicting where every predictive rating for every NBA team will end up at the end of the upcoming season.

For that purpose, we’ve settled on our current approach. And so far, this approach has delivered very good results.


Editors Note: You can sign up for algorithmic NBA picks and predictions (game winner, point spread, over/under, and money line) for all 2019-20 games on our signup page.