clock menu more-arrow no yes

Filed under:

The Offbeat: A Quick Look at 82games, or... Do The Jazz Play Their Best Players?

New, comments

82games features some fairly telling statistics about the Jazz' rotations. How can they be improved?

Steve Mitchell-USA TODAY Sports is an amazing statistical resource for NBA fans. Why? Because they interpret the NBA's play-by-play data to produce aggregate data you can use to answer all sorts of questions - those involving efficiency, position, lineups, shot clock usage, production, and more.

Today, we're going to peek around 82games to learn something about the Jazz' rotation.

Here is this season's 82games Utah Jazz basic statistics:


A quick glossary of what you see above:

  • "Min" just represents the total percent of team minutes an individual player has played.
  • Production is a slightly, but unsubstantially, modified version of Hollinger's PER. NBA player average is still approximately 15. Each player's own PER as well as their opponent's(at the position that player is defending) PER is tallied. The "net" is just the subtraction of those two stats, in other words, how much higher is your PER than your opposition?
  • On court tracks how many more points per 100 possessions the Jazz score than their opponents while the player is on the floor, off court tracks how many more points the Jazz score while the player is on the bench. Net compares the two.
  • Simple Rating simply takes a weighted average of the Production net score and the On Court/Off court net score, consisting of two-thirds Production, one-third On Court/Off Court. That proportion is not due to any analytical reason, it's just a reasonable enough balance between the two inputs. I actually quite like this statistic, and if you go around the league, you'll find the top players do extremely well in the Simple Rating statistic.

Here's the problem with the above table: for the Utah Jazz, being good at basketball (as defined by Simple Rating) is negatively correlated with playing time. This is obviously not the goal: one of the main roles of a coach is to play his best players the most minutes. Unfortunately, that isn't happening for the Jazz.

Let's take a closer look. While playing time is negatively correlated with the Simple Rating, it's a very weak correlation (-0.07). So, at a more nuanced level, it's not that Ty is playing the worst players more, it's just that he has no preference whether or not the players are good or not. Moreover, there is a positive correlation (0.32) between playing time and the PER differential of the players, so it's possible that the coaching staff is simply looking at player success on an individual efficiency level. Unfortunately, that also means that the +/- and playing time correlation is pretty strongly negative (-0.51), and while the +/- data certainly has a lot of noise, it's also what happens to win basketball games (i.e., scoring more points than the other team). To ignore +/- totally is fairly dangerous.

So, by this analysis, what should the coaching staff do?

  1. Play DeMarre Carroll at the expense of Randy Foye. Looking at the above chart they're almost totally opposites: Carroll's relatively efficient offensively, plays good defense, and has excellent +/- numbers, Foye displays none of those three. There's been much discussion about Foye's play recently, but the evidence is building: he's not a player who helps you on the scoreboard.
  2. Play Enes Kanter and Derrick Favors more. This is the classic SLCDunk criticism, and at this point, Ty Corbin himself probably agrees. But the data is building: it's not even just the plus/minus numbers showing the Jazz playing excellently with the younger bigs in the game, but also the production stats: both Favors and Kanter have better PER differentials than their starting counterparts. Defensive stats can be a little cloudy, so it's not exactly clear if Kanter and Favors are better players than Jefferson and Millsap, but the +/- data shows that we should probably err on the side of the young bigs.
  3. Play Jamaal Tinsley rather than Earl Watson. Obviously, the Jazz don't have the luxury at this point to choose just one of them, but Tinsley should be getting the majority of the minutes, probably a full 36 MPG. Tinsley's not a great option(his PER differentials show that), but he's a better option than Watson, unless Watson is able to recover to his pre-injury form. (Of course, that's what makes this so difficult. How do you determine if Watson is back in form without playing him minutes?)

Ty has a difficult job, there's no question, and his position in the locker room as a player's coach is important for the future of the franchise. It keeps stability and promotes a certain culture that is understandably important to the Utah Jazz brand. Unfortunately, at this point, keeping the highest priority on keeping players happy has meant significant underperformance on the basketball court, which ultimately is what any team is judged by. While keeping players happy is important, a priority shift towards the players who perform well may be in order.