Nylon Calculus: NBA free agency over- and under-rated signings

NEW ORLEANS, LA - FEBRUARY 26: Elfrid Payton #2 of the Phoenix Suns reacts during the first half against the New Orleans Pelicans at the Smoothie King Center on February 26, 2018 in New Orleans, Louisiana. 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. (Photo by Jonathan Bachman/Getty Images)
NEW ORLEANS, LA - FEBRUARY 26: Elfrid Payton #2 of the Phoenix Suns reacts during the first half against the New Orleans Pelicans at the Smoothie King Center on February 26, 2018 in New Orleans, Louisiana. 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. (Photo by Jonathan Bachman/Getty Images) /
facebooktwitterreddit

The 2018 NBA free agency period has been a whirlwind. Numerous big names made a lot of big moves and stole the majority of the headlines. However, there were also smaller moves with lesser-known players who have the ability to make a big impact on their new teams.

At it’s core, free agency is about value — production for the cost. For that reason, the best way to evaluate who has done well and poorly is to look for teams who paid players either well below or well above value based on their 2017-18 stats.  I used data from only the last two free agency years for free agent contracts, per-36 and advanced statistics. I am only including data from the last two years due to the significant salary cap increases which began two years ago. The statistical data was gathered from Basketball-Reference.com  while the contract data from Spotrac.com.

NBA free agency correlation matrix
To start I looked at the correlation between the various statistics and the player’s average yearly salary for their new contract. The chart below shows the correlation for each stat compared to one another. Dark blue shows is a correlation close to 1, dark red is -1 correlation, and white is a correlation close to 0. /

For this analysis, we really only care about the first column. The first column shows the correlation between the average salary and the each player’s statistics. Average salary and VORP are the highest correlated at 0.82. BPM, games started (GS) and minutes played (MP) also have high positive correlations with a new salary. The only significant negatively correlated stat is personal fouls (PF).

To start coming up with a way to measure players and their new contracts I started with a simple linear regression. I initially included VORP, GS and PF and split the data into a training set and a testing set. However, I was only able to get an R^2 value of 0.74. Which is not bad but we could do better. I decided to try a ridge regression with cross-validation, or ridgeCV regression. The ridge regression helps to limit the effect of collinearity between the statistics, such as BPM and VORP. Cross-validation is the process of randomly splitting the data into multiple partitions and averaging their results to get an estimation. The ridgeCV regression produced a better R^2 value of 0.81. It also produced a lower root mean squared error even when penalizing for the number of independent variables.

This model isn’t perfect and has some drawbacks including a small sample size. I only used data for the 200 players from the last two free agent markets. We might be able to find better relationships with more data, and it might be advantageous to compare a player’s salary as a percentage of team cap space. This would allow us to use data from the lower cap seasons of the past. This simple model allows for easier estimations and interpretations. What it lacks in complexity it makes up for with simple interpretability.

Using this model I predicted the value for each free agent from the last two years. The graph below shows the residuals for each of the recent free agents. In this graph, a positive residual means that the actual salary was higher than the predicted. In other words, the player is overpaid when compared to this model. A negative residual is just the opposite and the player could be considered a bargain.

NBA free agency residuals
NBA free agency residuals /

An obvious drawback of this model is it only predicts salary from statistics from the previous season. So players who could significantly improve or decline over the life of their contracts would change the valuations.

In the above graph, we can consider anyone below 0 to be bargains and anyone above to be overpaid. The model predicts Chris Paul to be the most overpaid player in this free agent market, although he contributes in other ways such as leadership and on-court organization. Another extreme to look at is LeBron James. LeBron is still a bargain even though his new contract provides over $38 million per year. The model predicts his value closer to $41 million due to LeBron’s outstanding season with a VORP of 8.9 and starting all 82 games.

Moving on to the more underrated players and signings. DeMarcus Cousins has the most negative residual and according to the model he is worth closer to $18 million. The big question for Cousins is the Achilles injury. Most players don’t easily bounce back from an Achilles. If, and that is a big if, he can play anywhere near his past production he will be the steal of this free agency class.

On the other hand, some lesser-known or possibly undervalued players include Elfrid Payton, Dwight Howard and Tyreke Evans. Howard at only $5.4 million per year is a great deal if he can get along with teammates. Evans is also interesting as he is coming off his best season since he won Rookie of the Year honors in 2009-10. If Evans can stay healthy he could be a great contributor to the Indiana Pacers.

Payton could end up being the best bargain from this summer’s free agency market. He’s coming off his rookie contract and is on a one-year, $2.7 million contract with the New Orleans Pelicans. While he hasn’t performed up to expectations he has the ability. Maybe being on a one-year contract could provide extra motivation. More importantly, Payton has spent his first seasons on the Orlando Magic and Phoenix Suns. Not exactly the most talented teams in the league. With the Pelicans, he has an All-NBA talent in Anthony Davis, quality players such as Jrue Holiday and excellent shooters in E’Twaun Moore and Darius Miller. Payton was in the top ten in the league in potential assists with 6.2 assists on 12.9 potential assists per game. Playing with higher quality players will only help him improve his assist numbers.

Next: Big men are getting under-valued in free agency

Another benefit for Payton and the Pelicans is similar to the departing Rajon Rondo. Both players are quality defenders with a lack of shooting ability. On this team Payton doesn’t have to be a great shooter, he can be a creator and passer. This could be a great deal for the Pelicans and an opportunity for Payton to prove his worth. It will be interesting to see how some of these underrated contracts perform this season.