Why My Draft Model Doesn't Love Andrew Wiggins

(But really liked Jordan Adams, even before the tourney)

In the latest test of my draft model tests Andrew Wiggins ranks as the 24th best prospect based on his age and statistical production.  Yet he is widely regarded as the first or second best prospect in mock drafts around the web.  

My model is still very much a work in progress, but I doubt any further development will raise its estimation of Wiggins to those heights.  I fairly confident in that because other stats models, such as Layne Vashro's model and Kevin Pelton's WARP based model have Wiggins rated very similarly to mine, both ranking Wiggins in the twenties. (Since my model includes prospects playing in Europe, he's possibly a little higher in my model than in the other two).

All of these models are based primarily on boxscore production, and there Wiggins numbers have only been good, not great, even accounting for age.   The current model was built on pace adjusted data from Draft Express as of the March 17th, and so, doesn't include either Wiggins' above average first game in the conference, nor his tough second game.  

The short and simple answer is that Wiggins production during his first, and likely only, year in college ball has been good, but not great.  But, I think it's worth looking at in a bit more detail.

Comparing Wiggins' stats production in different categories to his fellow top 100 prospects as listed by Draft Express, what stands out is the lack of standing out.  One way to express this is to look at how his numbers compare to the average in that category and scale the the numbers by the standard deviation.

•  In True Shooting percentage Wiggins came into the tourney at 57%, right at the average for the top 100 prospects,  And he is doing it with a usage that is only slightly, .15 standard deviations, above average.
• His rebounding is similar, with offensive rebounds .21 standard deviations above average and his defensive rebounds -.48 standard deviations below average.
• Assists are -.45 below average, steals are approximately average among his fellow top hundred prospects, blocks are below average, turn overs are average, and his fouls per 40 minutes is .29 standard deviations above average (meaning he fouls less often than the average prospect)

Basically Wiggins production is very much average for a top 100 prospect, until we account for the fact that he is able to achieve this at 19 years old.  It is the adjustment for age that puts Wiggins into the top 30.

Another way to understand how his production looks is to simply compare it to other wing players in this draft.  Below I have Wiggins, Jordan Adams, Jabari Parker, Doug McDermott, Nik Stauskas, and Sam Dekker.

 

Name	Team	Age at Draft	Minutes	TS%	TS-StDev	O-Reb-StDev	D-Reb StDev	Ast StDev	Steals StDev	Blks StdDev	TOV StdDev	Usage %	Draft Mod 1a
Jordan Adams	UCLA	20.0	990	61%	0.79	-0.03	-0.48	0.15	2.83	-0.86	0.68	0.24	5.87
Jabari Parker	Duke	19.3	1,044	56.9%	-0.15	1.05	1.25	-0.62	-0.01	0.28	-0.52	0.31	5.69
Doug McDermott	Creighton	22.5	1,109	65.6%	1.86	-0.24	0.56	-0.45	-1.57	-0.93	0.44	0.33	4.53
Andrew Wiggins	Kansas	19.4	1,076	57.5%	-0.01	0.17	-0.48	-0.45	-0.01	-0.17	-0.04	0.25	4.55
Nik Stauskas	Michigan	20.7	1,133	65.1%	1.75	-1.19	-1.16	0.74	-1	-0.71	0.44	0.22	4.37
Sam Dekker	Wisconsin	20.2	993	56.2%	-0.32	0.03	0.46	-0.45	-0.43	-0.33	1.16	0.21	4.2

Jordan Adams leads in steals and is among the lowest in turnovers among this group and is second in True Shooting percentage and assists.  Adams is also only 7.5 months older than Wiggins.  While Jabari Parker has the best rebounding numbers and similar scoring efficiency to Wiggins, though with a much higher usage and Doug McDermott has put up amazing scoring numbers, but is three years older.

By the way, my numbers indicate that this may be a fairly deep draft, but not one with a clear super star.  Comparing this year's draft class to those used in the training set for the model, Wiggins would score in the 80th percentile of players drafted, but still sits in the mid-twenties compared to this draft class. There are also eight or nine players that project in the 90th percentile compared to the training set, but none that would project in the top ten.

 Why the Model Might be Wrong

All this said, there are reasons to think any stats based draft model may not tell the whole story. Analytics models have to be weighted with traditional scouting, preferably in a Bayesian process.  Wiggins' athleticism is clear, as his length.  Additionally, all of the reports on Wiggins emphasize his defensive potential, which is poorly captured in box score data.  

Lastly, Wiggins' age, and Parker's, for that matter, makes predictions about development particularly difficult as any model can only project out based on typical level of development.  Any of these kids could go on to surprise on the up side, or the down side.