Some of you may have noticed a while back that Draft Express had an article showing a "consensus" from the analytical draft models. I've been following the APBRmetrics forum since then, since that is the group that has been producing that analytical model consensus.
Apparently, they're going to have another article out in DX sometime soon, with an updated consensus (adding a few more models to the mix) and addressing some of the questions that arose from that first article.
That new consensus will be based on averaging the draft-order rankings of the various models. If you're interested in this stuff, keep an eye out for that in DX.
But the rankings that will appear there, for just-a-bit tricky mathematical reasons, aren't the best way to calculate the consensus. Their method is the most understandable for the reader who is not very familiar with statistics, and that's why they went with it. But the better method is to calculate z-scores within each of the various models before averaging.
One of the contributors to that forum did just that:
https://docs.google.com/spreadsheet...KuuMpI5eFxqby-QU8SobLgIJME/edit#gid=529517843
I wanted to see the scores in a bit more understandable way, so I changed the z-scores into a (approximately 0-100, bell-curve distributed) scale. So the average ranking for about the 100 or so prospects ranked is about a 50. Here's what these results show:
1. Russell 94.5
2. Towns 90.5
3. Okafor 87.6
4. Winslow 78.1
5. Mudiay 78.0
6. Johnson 75.8
7. Looney 74.3
8. Jones 73.0
9. Turner 71.7
10. Wright 66.1
11. Portis 66.0
12. Kaminsky 65.8
13. Payne 63.4
14. WCS 63.3
15. Oubre 62.0
16. Wood 61.4
17. Hunter 60.0
18. RHJ 58.6
and others:
21. Porzingis 56.3
27. Dekker 53.2
29. Lyles 52.9
31. Anderson 52.5
33. Booker 51.6
34. Hezonja 50.9
The obvious suspicion is that these models highly under-rank Kentucky and Spanish-league players. But it's interesting since these are the consensus of seven different models, not just one that may idiosyncratically value things oddly.
Anyway, the usual provisos apply: mathematical models should never be used in place of trained eyes and discerning background checks, etc., but presumably in concert with with them.