August 13, 2019
David Chung

NBA Free Agency: Who Does Data Say Won?

We analyzed the value of each free agent to see which contracts made sense.

Another NBA free agency has come and gone. As avid sports fans, we at Nuview love diving into sports analytics. We eagerly count the days to the start of the new NBA season, we either are salivating or despising the idea of many of the NBA’s brightest stars moving throughout the league. But how much are these teams paying for these stars and are they overpaying them?


Like many NBA fans, the joy of signing the most coveted star in the market often outweighs the concern of over payment ,however, there is always a lingering fear that the team is overestimating that star’s capabilities. While it is easy to automatically give our stamp of approval and since we are analytics folks, let’s take a closer look at the numbers.


First, we collected data we believed would impact a player’s annual salary such as PPG, RPG, USG%, and 3p%, from 2016-2018. In order to eliminate as many outliers as possible, we took out any player who played less than 15 minutes-per-game. After collecting and filtering the necessary data, we then see which variables have the greatest effect on annual salary by looking at a correlation matrix. By observing their relation to the amount of cap the salary takes, we can eliminate insignificant variables that do not strongly affect the salary.


Next, it is we need to understand multi-collinearity. If you look at the heat map down below, we can see which variables have a strong correlation to percentage of cap and also see that variables such as FGM and MIN have a strong correlation to each other meaning it will hinder the accuracy of our model if we keep both variables.

Correlation Matrix of Attributes

After filtering the necessary variables, we ran our regression and see the overall statistics based on players’ data. We then can further filter out variables that are statistically insignificant and run our regression again. Down below is an overall summary of our final regression and depicts our most significant variables. According to our data and regression our R-squared is roughly .376, and our most impactful variables are TOVTOT, PPG, APG, and DREBTOT. Clearly not our best model but something that can be useful nonetheless. When analyzing our fitted vs residual plot, there does appear to be some gap between the two, however there are some points that are very similar with very little difference. Overall, there is some accuracy, but there also a level of inconsistency in our model.

In the first visual, we see who has the highest annual salaries from 2016-2018.Unsurprisingly, superstars like LeBron James and Stephen Curry are the cream of the crop. The next visual shows who has had the highest PPG for their respective seasons. When comparing the two tables, we see a few similar names but also a couple of players who can score consistently, yet are not among the leagues most paid players. In the next diagram, we see a plot that depicts the relationship between PPG and percentage of cap. While there is a positive trend, it appears to be primarily scattered across the plot with high scoring players not having as a high of a salary.

The next visual conveys the league’s best distributors from 2016-2018. Much like top scorers, not all the top distributors are on the top earnings list however,the plot indicates that there is a positive correlation with a significant amount of scatter throughout the plot. As the NBA has moved toward a faster positionless game, I expect assists to play a significant role in evaluating players for new contracts.

Our next variable that we focused on was the total number of defensive rebounds. As the diagram shows, the top rebounders are generally forwards and centers and unlike PPG and APG, the top rebounders are seemingly not the top earners in the league with the exception of Russell Westbrook. However, the plot does convey a positive correlation indicating that rebounding does play a role in player’s overall salaries. Though rebounds alone cannot give a player an outstanding contract, it does point to a positive direction.

The last variable observed is total number of turnovers in a season. Understandably,turnovers are generally viewed negatively, but the top players that turn the ball over are also earning the greatest contracts. Both James Harden and LeBron James are among the top players that earn the most annually and turn the ball over. I believe that this is due to their high usage with their respective teams and the fact that they are game pace managers that can control the tempo of the game. Interestingly enough, the plot does indicate a positive trend with total turnovers meaning that because many of these stars are ball-dominant,there is a higher chance of those players turning the ball over.

So now what everyone wants to know is are their favorite teams overpaying for the players?  In the final two tables, we have collected data on 5 random players from the 2018-2019 season who have recently received new contracts and plugged them into our regression.

Derrick Rose has had arguably one of the most stunning statistical seasons in spite of his injury history averaging 18 ppg and 4.3 apg. Pistons rewarded him with a new contract that takes roughly 7% of their salary cap. However, based on our regression model, his statistics tell us that he deserves to be paid 6% more than his newly signed contract.


Jamal Murray was also rewarded with a new contract that takes a whopping 32% of the Nugget’s salary cap. While the potential star showed flashes of brilliance, his numbers suggest that he is 18% overpaid,which may give a lot of fellow Nugget’s fans grief. Interestingly enough, the table shows that Jared Dudley is only 1% underpaid meaning the Lakers have successfully signed a player close to his expected value giving Lakers more wiggle room to sign more impactful players to help LeBron and Anthony Davis.

As you can see, there is some evidence in the model’s accuracy, but it is still not completely reliable. The model does take into account of all general and advance statistics, but it does not take into account personal accolades, age, and injury history. Obviously, every team will leap through hoops in order to sign a player of Kevin Durant or Kawhi Leonard’s caliber, but it does raise questions as to whether or not they are given more than what they deserved. As the salary cap has increased in last few years, the number of lucrative deals to subpar players have increased resulting in those teams being unable to sign more talented players and struggling to trade away those horrible contracts.

David Chung

Contact Us
Thank you for your inquiry - we'll be in touch shortly!
Oops! Something went wrong while submitting the form.