Nylon Calculus: A nearest neighbor approach to 2017 NBA Draft statistical comparisons
Before drifting off to sleep, my mind tends to wander. Sometimes it goes over the events of my day. Sometimes it recalls an obscure childhood memory. Most recently, it wondered how in Adam Silver’s good name did we compare Adam Morrison to Larry Bird?
Many player comparisons are based on the eye-test. However, these comparisons suffer from a loss of information when they don’t factor in statistical similarity. In this post, I combine both analytical and visual methods to identify how the 2017 draft class could project at the next level.
A Little Clustering
Before diving right into the player comparisons, I’d like to take a look at identifying the types of players in the college game. The visualization below reveals some interesting findings. The skree-plot and the cumulative explained variance of the data both suggested there are four separate clusters to the data. Taking a look at the four different groups, we have the following:
Facilitators
These players are looking to create for their teammates or are in control of the team’s offense and as a result, have high assist and turnover stats. These players include Chris Paul, John Wall, and Stephen Curry.
Snipers
These players can flat out shoot the ball and are characterized by their high field-goal percentages.
Bigs
As the name suggests, these guys are big bodies that can get rebounds, block shots, and post high 2-point field goal percentages. Typical players include Festus Ezeli and DeAndre Jordan.
Hy-“Bigs”
In an attempt to not take up too much space, and give an informative annotation, I present the hybrid big-men. These are your versatile big bodies who can score, rebound, and distribute equally well. Some examples include Carmelo, Blake Griffin, and Paul Millsap.
Placing the 2017 Class
As a sanity check, we see Malik Monk as part of the snipers and Lonzo Ball as part of the facilitators. Players like Semi Ojeleye are interesting because they are located on the classification borders between two different player types. They present a combination of qualities between the two and represent a type of versatility. We see that players like Josh Jackson and Jayson Tatum could potentially be “point forwards” while players like Lauri Markannen and Zach Collins are sweet-shooting bigs.
Making Comparisons
Using a KNN approach, I sought to quantify similarity between this year’s class and past NBA prospects. I’m only compiling a short list of the most interesting statistical comparisons.
Lonzo Ball
Mike Conley, Ty Lawson, Kyrie Irving
It’s interesting to see that these comparisons are similar to 538’s comparisons. We can only hope Lonzo becomes as gifted as Kyrie and as defensively capable as Conley.
Markelle Fultz
Dwyane Wade, Stephen Curry, O.J. Mayo
I wholeheartedly agree with this list. Fultz could be this good. He has the touch to be like Curry and the mid-range finishing ability like Wade.
Jayson Tatum
Luol Deng, Kyle Singler, J.R. Giddens
Tatum’s floor doesn’t look too good here, but Luol Deng seems like a feasible comparison.
Josh Jackson
Paul George, Will Barton, Josh Howard
Jackson’s statistical company looks great. A recurring theme among these players is their ability to play positions 1-3.
Dennis Smith Jr.
John Wall
Smith definitely has the explosiveness to be a Wall-like player. Let’s see if he develops the same vision and leadership that Wall has.
Malik Monk
J.J. Redick, Andrew Goudelock
The commonality between these players is their sharp-shooting abilities. The lack of dimensionality is concerning, though Monk’s athleticism can take him further.
Next: Nylon Calculus -- A lineup-based analysis of the Chris Paul trade
De’Aaron Fox
Isaiah Thomas, Kemba Walker, Derrick Rose
This is some serious company. The scary part is, I think Fox could be this good, or better.