clock menu more-arrow no yes

Filed under:

DraftExpress Analytics Article Suggests Draft Sleepers

New, comments

By examining the players scouts and models most disagree on, we can perhaps gain insight into the players most underrated by traditional mock drafts.

Kirby Lee-USA TODAY Sports

DraftExpress released an article this week that elevated draft models based on algorithms to a higher status than they had previously enjoyed among many scouting communities. The article gave insight into the machinations of 5 different models, then looked at them in aggregate, by seeing how these models graded players on average.

These predictive models take a player's numbers from whichever league they are currently ply their trade in,and attempt to translate them to the NBA based on how previous numbers have or have not translated with players who have previously made leaps from those leagues.

The accuracy of any single model is certainly up for debate, but ignoring their conclusions seems like folly in the modern era of NBA analysis. As is the case with every new tool to help executives evaluate talent, these models are meant to be applied in addition to traditional scouting, not instead of it.

What we, as laymen, can perhaps glean from these models are the players who are being over- or underrated based on their physical tools. Players for whom there is a large disagreement between the model consensus and the scouting consensus are likely to be steals if taken where they are projected to be taken, or else they may be far overvalued given their traditional placement. This is one of the best parts of the DX article, as it actually shows a comparison of where players were placed by the predictive models' average and by DX's own ranking.

We can, in turn, take these numbers and try to determine which players might be worth targeting at specific ranges, especially in the second round, as Hinkie will have 5 selections there. Alternately, we can use these numbers to single out players who are on the fringe of the two rounds who might be worth consolidating assets to take a stab at. Sean O'Connor wrote about the possibility of taking a stab at these late first players, and pouncing while they're still available may be worth it. This can be a tool to help determine which players Sean's proposal could provide value for.

Here are the players that DX currently projects in the final third of the first round who may be undervalued:

Player

DX Rank

Model Rank

Difference

Kevon Looney

17

5

+12

Delon Wright

25

9

+16

Christian Wood

22

11

+11

Chris McCullough

27

18

+9

Conversely, these are the players that the models show to be overvalued by DX, and may not be worth their current placement on most big boards:

Player

DX Rank

Model Rank

Difference

Rondae Hollis-Jefferson

10

17

-7

Sam Dekker

13

21

-8

Trey Lyles

16

24

-8

Devin Booker

9

33

-24

As for second round steals, the biggest winners are Wesley Saunders and Seth Tuttle, who both jumped more than 40 spots in the aggregate model rankings.

Player

DX Rank

Model Rank

Difference

Richaun Holmes

34

22

+12

Terry Rozier

40

26

+14

Wesley Saunders

72

28

+44

Vince Hunter

48

30

+18

Seth Tuttle

74

31

+43

Josh Richardson

50

40

+10

TJ McConnell

49

38

+11

Larry Nance

54

41

+13

Branden Dawson

61

42

+19

Pat Connaughton

53

43

+10

Chasson Randle

62

44

+18

Derrick Marks

69

48

+21

Unsurprisingly the biggest losers are old seniors, who were unproductive early in their careers but began to dominate as their peers graduated. This includes Norman Powell, Rakeem Christmas, Anthony Brown, and Jonathan Holmes.

Player

DX Rank

Model Rank

Difference

Michael Frazier

31

45

-14

Michael Qualls

32

46

-14

Olivier Hanlan

42

55

-13

Norman Powell

35

57

-22

Rakeem Christmas

29

60

-31

Anthony Brown

33

59

-26

Joseph Young

45

62

-17

Chris Walker

43

66

-23

Jonathan Holmes

21

71

-50

It's important to remember that these numbers might not necessarily be correct. But they are more tools to help inform decisions and roster choices, which is something we know Hinkie values highly. While he may put equal emphasis on qualitative and quantitative analysis, he will certainly be using this data and more analysis like it.

Perhaps he will find data that disagrees with the conclusions of these five modelers. At the very least, we can use what we know to help us better guess that players that Hinkie may target on draft night, and we can pretend to have been in on it all along.