Fansided

Nylon Calculus: Highly Plausible Win Projections for the 2017-18 NBA season

By Andrew Johnson

HOUSTON, TX - JULY 14: The Houston Rockets introduce Chris Paul toss t-shirts to the fans on July 14, 2017 at the Toyota Center in Houston, Texas. NOTE TO USER: User expressly acknowledges and agrees that, by downloading and/or using this photograph, user is consenting to the terms and conditions of the Getty Images License Agreement. Mandatory Copyright Notice: Copyright 2017 NBAE (Photo by Bill Baptist/NBAE via Getty Images)
HOUSTON, TX - JULY 14: The Houston Rockets introduce Chris Paul toss t-shirts to the fans on July 14, 2017 at the Toyota Center in Houston, Texas. NOTE TO USER: User expressly acknowledges and agrees that, by downloading and/or using this photograph, user is consenting to the terms and conditions of the Getty Images License Agreement. Mandatory Copyright Notice: Copyright 2017 NBAE (Photo by Bill Baptist/NBAE via Getty Images) /
facebooktwitterreddit

For the fourth year in a row I am putting out my Highly Plausible Win Projections. The basis of the projections is a blend of my own player rating called Player Tracking Plus Minus (PT-PM), a combo of box score stats and SportVU, and Regularized Adjusted Plus Minus (RAPM). Last year was both an up year and a down year for the Highly Plausible Win Projections. After finishing at the top of the APBR projection board for two years, my projections finished in the middle of the pack last year despite having a lower average error rate than the prior years and beating both Vegas and 538’s CARMELO for the third year in a row.

Despite a few tweaks the projection system is similar to prior years; past performance in PT-PM and RAPM for each player is adjusted for age and playing time then weighted by their expected playing time in the upcoming year. Changing teams leads to an additional regression to the mean, as past performance has indicated that players changing teams have more uncertainty in their performance.

The minutes projections are done with a depth chart style system, but some hand adjustments are needed especially in cases where a player is headed into the season with a known injury, like Isaiah Thomas or Jabari Parker. In general, my research has shown that changes in minutes played is responsible for 10-15 percent of the error between the projections and the teams’ actual final record.

Read More: Projected 2017-18 NBA season win totals from Jacob Goldstein

Once the expected overall strength of each team was estimated I ran them through their schedule with adjustments for home teams and appearances in back to back games.

Eastern Conference

All that said, here are the projections by conference, with the Pinnacle regular season line as of Sept. 23, a couple of teams with big personnel changes had been removed from the line at that time. I also calculate the deviation of my projections from the Pinnacle line, with a negative meaning I am projecting under the line and a positive over.

In the last column, I have the playoff odds based on 500 simulations of each conference using my strength projection and rate of error from my past models.

A couple of points from the table:

  • I have Cleveland as a decent size favorite in the East, though this assumes a reasonable return by Isaiah Thomas both in terms of playing time and performance. It also, assumes only a minor drop-off in minutes for LeBron James.
  • Boston, Washington and Charlotte are in a surprise tier with little difference. Boston has huge roster turnover and six rookies on their roster, 55.5 wins would take almost everything going right for them.
  • My projection is much lower on Milwaukee too in the next tier, in large part due to depth issues. That said, with the good size gap between Miami and Milwaukee and the next tier, both landed in the top eight almost 80 percent of the time in the simulation.
  • The lower middle class of the East is the point where my model disagrees with the Pinnacle line the most. I would have the under on Philly and over on Indiana and Atlanta but then my model has always been a bit irrationally high on Ersan Ilyasova.
  • Brooklyn gets a four percent shot at the playoffs, which would be kind of incredible.

Below is a distribution of the how the over/under line for the Celtics and Pacers looks based on my projection’s error rate over the last three years with a normal distribution:

Organizational motives aren’t factored into the projections for a team like the Pacers, but given the weakness in the East a playoff spot opportunity and scads of worse teams below them could motivate the front office to play out the season.

Western Conference

Below is the same table for the Western conference. Overall the projections line up better with Pinnacle line, with one big exception.

  • Short of the “Big One” in California the Warriors are making the playoffs.
  • The big surprise is in the next tier with the Clippers. Part of their projection is driven by a very high rating on DeAndre Jordan, which will be tested especially on offense without Chris Paul. The players coming over from Houston were all rated as at least average, even with added mean reversion trading teams. This is a Clippers team with actual depth, hard to fathom, I know.
  • Oklahoma City is almost the opposite, despite their stars, there are some depth issues that could hurt them in the regular season. Despite the talent, Carmelo Anthony hasn’t had a significant impact on winning in the last couple years, whether Olympic ‘Melo shows up in Oklahoma or not is an open question.
  • The projections show a legit race at the bottom of the playoff ladder. The simulations had Denver in 67 percent of the time and Utah, New Orleans and Portland all with odds hovering in sight of 50/50.
  • Sacramento was a top eight Western Conference team in 0.6 percent of the simulations, or three out of 500. It’s hard for me to understand the scenario where that happens, maybe throwback Vince Carter and Buddy Hield really does shoot like Curry while being big like Klay.

Next: 25-under-25 -- The best young players in the NBA

And, yes, my model hates your team.

Home/Nylon Calculus