Similarities between 2012 NCAA Draft Class and Current NBA Players (A Rough Draft)

(No pun intended.)

So, once I get an interesting new idea in my head, I tend to obsess about it (perhaps, too much). Yesterday, I wrote about a way to compare players to each other using a “distance” measure of statistical similarity. Some time after I wrote that, I had a Eureka! moment and thought, hey, I should just put current NBA players into the model, and see who the current draft compares to. This is my first stab at it, using college stats from the last six draft classes (going back to 2006). I only used the basic pace-adjusted stats this time around, so I think there’s a lot of room for improvement. But I wanted to put something up, because I think the results are neat. There are definitely some head scratchers (Anthony Davis compared to Demar DeRozan?!). Oh, and in case you’re wondering, Jae Crowder is the next Jeremy Lin.

Click to enlarge.


A couple of things. 1) It’s better to normalize each stat, so that scales are similar. 2) A polar dendrogram save a lot of space.

Click to enlarge.

5 thoughts on “Similarities between 2012 NCAA Draft Class and Current NBA Players (A Rough Draft)”

  1. Maybe break it up into a couple different batches and do it by position? or even by height/weight? might give you more space and make comparisons more robust. really great stuff though. thanks for sharing.

    1. I sort of agree and disagree. On the one hand, it’s the obvious choice to compare by position. On the other hand, I think it’s interesting to see how players compare regardless of position. For example, the fact that Moe Harkless and Harrison Barnes are about 180 degrees apart on the map, one clearly on the side of “bigs” and the other on the side of “wings” potentially tells you more about their game than if they were on the same diagram with just SF’s. Or another one is MKG vs. Jared Sullinger. You might miss the fact that a SF like MKG is actually closer to “bigs” than Sullinger. Just a thought. I can certainly do it both ways.

      Another thing I will do that I realized will be useful is to split out the data by year. One-and-done players should be compared with each other probably moreso than players coming out after 3 or 4 years and vice-versa.

Leave a Reply