Projecting the NBA Draft

After posting the results from my similarity scores system, I discovered that my system is by no means the only one out there. I was already aware of Ed Weiland's work over at Hoops Analyst but not aware of the recent postings by Kevin Pelton at Basketball Prospectus or the work of Richard Lu at Draft Outcast.

All of these systems provide a different perspective - Weiland's system uses data from college players who didn't make it; Pelton's system compares projected rookie seasons and Lu's looks at how players will fit into the team whom they were drafted by. To add something new to the discussion, I decided to make career PER projections.

First off, I made some modifications to my similarity scores system. The categories are now weighted as follows:

5: Age
4: Height, Stls/PF, Blks/PF
3: Weight, Pure point rating, Rb/40, 3p/FGA
2: FTA/FGA, Pts/40

Honestly, adding weights didn't change the rankings very much, so even though the weights are more art than science, I'm comfortable that I'm not adding much bias to the results.

Having generated lists of comps for every draft pick, I then averaged the career NBA PER for their comp list, weighted by the strength of the comparison.

Without any further ado, here are the results:

Player	Mean PER	Comps
Blake Griffin	18.79	9
DeJuan Blair	17.51	6
Jonny Flynn	17.09	13
James Harden	16.69	10
Nick Calathes	16.07	9
Tyreke Evans	15.45	28
Stephen Curry	15.28	13
Patrick Mills	15.28	16
Austin Daye	14.67	12
DeMar DeRozan	14.54	18
Jrue Holiday	14.17	5
Ty Lawson	14.14	10
Jeff Teague	13.76	21
Lester Hudson	13.70	1
Darren Collison	13.69	28
Tyler Hansbrough	13.52	18
James Johnson	13.38	78
Ahmad Nivins	12.95	49
Jordan Hill	12.93	112
Jon Brockman	12.87	50
B.J. Mullens	12.85	2
Danny Green	12.80	58
Derrick Brown	12.76	65
Chinemelu Elonu	12.69	38
Eric Maynor	12.61	57
DeMarre Carroll	12.61	80
Earl Clark	12.60	56
Chase Budinger	12.60	103
Terrence Williams	12.56	33
Gerald Henderson	12.54	85
Jeff Pendergraph	12.50	100
Hasheem Thabeet	12.45	2
Dante Cunningham	12.40	95
Marcus Thornton	12.33	89
Taj Gibson	12.33	28
Wayne Ellington	12.28	103
A.J. Price	12.23	59
Toney Douglas	12.01	77
Sam Young	11.98	36
Jodie Meeks	11.97	77
Robert Dozier	11.83	31
DaJuan Summers	11.67	49
Jack McClinton	11.66	5
Jermaine Taylor	11.61	49
Taylor Griffin	11.47	55
Robert Vaden	11.40	15
Goran Suton	11.23	27
Calvin Brock	11.07	33
Chester Frazier	10.41	15

Interestingly, these results line up pretty well with my own subjective feelings about the draft. Blake Griffin, James Harden and Tyreke Evans are viewed here as strong top 5 picks while DeJuan Blair and Nick Calathes stand out as sleepers. Interestingly, the system is not as high on Ty Lawson as I am, as undersized points don't have a great history, and prefers Jonny Flynn and even Patrick Mills. I'm not sure Flynn or Mills are quite as good as their comps, but they both fit the mold of quality NBA guards. Also don't read anything into the low ranking of Hasheem Thabeet - with only 2 comparable players and neither being strong comps at that, he is too unique of a player for this system to really apply to him.

What I like about this system is that most later draft picks are seen as 'generic' players, meaning they have tons of valid comps and not great projected PERs. For the higher picks, this is not so much the case, suggesting they have much more of a future as stars. For high picks like Jordan Hill and Gerald Henderson who have a ton of comps, this may be a sign that they are destined to be role players. I also included Illini alums Chester Frazier and Calvin Brock, two players who were not drafted and are not expected to make a summer league roster, to test the system. Interestingly, they are the two lowest-ranked players here, which gives me more faith in the system. Now we will just have to sit back and see if the 2009-10 season confirms our predictions!