Nylon Calculus: Using math to predict the 2017-18 NBA MVP race

SAN ANTONIO, TX - MARCH 27: LeBron James
SAN ANTONIO, TX - MARCH 27: LeBron James /

The 2016-17 NBA MVP race was one of the most exciting in recent memory, and possibly ever. Almost all of the major challengers for the award last year (Russell Westbrook, James Harden, Kawhi Leonard, LeBron James) are in position to compete again this season.

So, how can we use the history of the MVP race to predict the 2017-18 season? Who will be the 2017-18 NBA MVP?

Building a model

To start off with, I built a database of every winner of the award going back to the 1979-80 season, the first with an NBA 3-point line. Then, using MVP award shares as a target variable, I performed a linear regression of a few key variables that I found to be predictive of future results.

The final equation for MVPScore is as follows:

Where the constants are:

This equation misses on players who win MVP for narrative driven reasons more than on-court reasons, like Derrick Rose in 2011 or Allen Iverson in 2001. Both of those players were very good the year they won MVP, but may not have carried their teams in the traditional sense compared to other MVPs.

This methodology has been able to correctly predict 25 of the last 38 MVPs, including the 2016-17 MVP Russell Westbrook.

A 66.0 percent hit rate is not perfect but does a good job at identifying at the very least who the right candidates are. Of the last 38 MVPs, 34 have received one of the top three MVPScores in the season they won. In fact, only two of the last 38 MVPs did not rate as the top-5 most likely winners by this metric and both times it was the same player: Steve Nash, who came in eighth by this metric for his 2004-05 MVP and ninth for his 2005-06 MVP.

Knowing that the eventual MVP was always a top-10 player by MVPScore, and 95.0 percent of the time a top-5 player, let us look at the players for the 2017-18 NBA season who project to receive the best score.

The 2017-18 NBA MVP is…

With a projected MVPScore of 37.7 percent, the 2017-18 MVP favorite in my model is currently Kawhi Leonard. Leonard projects to have per game averages of 23.7 points, 6.2 rebounds, 3.7 assists and 1.8 steals while being the best player on a 50+ win team.

Here is the entire projected top 10 for the 2017-18 NBA MVP Race:

The biggest surprise is Jimmy Butler coming in at sixth. The MVPScore metric sees Butler as the most important player on a Timberwolves teams that is expected to win about 48-52 games. Butler projects to per game averages of 19.2 points, 5.6 rebounds, 4.2 assists and 1.6 steals. While more casual fans may view Karl-Anthony Towns as Butler’s equal, Butler rated out as one of the 10 most impactful players last season while Towns did not.

Players of note who just missed the cut for the top-10: Rudy Gobert, Kemba Walker, Kyle Lowry, Chris Paul, Damian Lillard and Gordon Hayward.

Next: How accurately can we predict NBA Playoff berths?

The 2017-18 NBA MVP race is shaping up to be a really great one once again with no clear favorite. Narrative will matter a lot. It hopefully is able to live up to the high bar the 2016-17 MVP set. The season is just under a month away. Soon we will see who steps up and leads their team.