How many games are the Jazz going to win this season?

After an offseason with little change, it’s easier to project where the Jazz will end up than many other teams trying to merge new pieces into their rotations. But there are still some questions.

The Jazz probably can’t win all year long at the pace at which they finished the regular season last year, a 29-6 stretch that would mean 68 wins over a full season. But it’s probably best to expect the Jazz to start the season better than where they did last year, with young players who have improved and with more cohesion as a group.

One way to answer the question is to look at the situation with statistical modeling, which might give us some idea on how to handle teams who have had seasons like the Jazz’s before.

ESPN

ESPN’s Kevin Pelton has been projecting team wins for years, both at ESPN and at Basketball Prospectus before that. Sometimes, his predictions cause consternation. More often, they’re right.

First, Pelton uses his projection system, called SCHOENE, to estimate how each player will impact the score when he’s on the court, as measured by ESPN’s Real Plus-Minus. The past few seasons of play are all considered, as well as a player’s age, position, and tenure with his team. Then, Pelton estimates how many minutes each player will play for each team (he does this largely through guesswork), which allows his model to spit out a win total.

Pelton projects that the Jazz will win 53.4 games, good for the second seed in the Western Conference.

A model to tinker with

Jazz fan Kevin Huynh was inspired by Pelton’s work, but wanted the chance to tinker with the specifics. So he borrowed the work of the FiveThirtyEight team, which uses its CARMELO projection system to predict how each player compares to the mean. Then, he made his own estimates of playing time, and ran a quick study to find that each point above average usually leads to about 2.2 wins.

His result was predicting the Jazz winning 52.8 games, a good total. But the real fun of his system is being able to play with the variables: make a copy of his Google Sheet and experiment to your heart’s content. Want to see how the Jazz would end up if Rudy Gobert played a full season of healthy minutes? Give it a shot. Predicting a breakout season from Dante Exum? Change his projection and see what it does. What would happen if the Jazz started Georges Niang, Tony Bradley, and Grayson Allen all season long? Go nuts.

Using the schedule

Nylon Calculus writer Jacob Goldstein took a bit of a different approach. His model starts similarly from those above, although he uses a different metric: his own Player Impact Plus Minus. Like the rest, he takes into account player aging, their past few seasons of play, and lots of other factors to estimate how good each team will be.

What makes his model unique is it projects a team’s win total by going game-by-game through the schedule. For each game, his model looks at whether a game is played at home or on the road, whether or not a team has a rest advantage, and of course, how talented those teams are. Then, the model gives a percentage win probability for each team for each game. Add them all up (or run the process 10,000 times), and you can find out the most likely winning total for each team. For the Jazz, Goldstein projects 49.2 wins.

What Vegas says

Finally, there’s the predictions of Las Vegas’ sportsbooks. Many of them are waiting for the finalities of the NBA offseason to settle down, but at least one, Westgate SuperBook, is now taking bets. Yes, they take advantage of a computer model too, though of course the details are secret. But Vegas’ sportsbooks also hedge up or down to meet consumer demand, trying to get a line that will make them guaranteed money.

For the Jazz, their over-under is 48.5 wins. If you believe the computer models above, take the over.