APBR-DraftExpress 2016 NBA Draft Project

Home for all your discussion of basketball statistical analysis.
Statman
Posts: 548
Joined: Fri Apr 15, 2011 5:29 pm
Location: Arlington, Texas
Contact:

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Statman »

Crow wrote: Conceivably true talent has been compromised by bad early role fit and coaching decisions about when to use and not use. Players and coaches should get better with time and the correlations should rise some. But will it be enough to look decent or better eventually?
OK, rereading what you wrote - yes, 1st season BPM (or almost any metric) is WAY too early to pronounce any kind of success or failure of any model imo.

Only 4 rookies coming from college of > 68 minutes had a BPM > -0.3 (Towns, Hollis-Jefferson, Cauley-Stein, and Richardson). Note, if you were to look at my '15 draft model upload and click on the year to year sheet (near the top of this: http://hoopsnerd.com/?page_id=823 )- my model only projected 4 drafted players to have an above NBA average Hn/48 (my "equivalent" to BPM or WS/48 or PER or whatever) - Kaminsky, Towns, Wright, & Okafor. There are OBVIOUSLY more than 4 players coming out of college in the '15 NBA draft who will end up better than NBA average - yet my projected results and actual BPM results had but 4 (only agreed on 1 though).

Adding to the point, D'Angelo Russell was my #2 prospect overall - but there were 21 guys ranked in front of him in my first season projected HN/48, yet again the scope of his career he projected up to #2. 1st season player projections often don't come anywhere close to career projections (based on age, player type, etc.).

So, in my obviously very biased opinion, one needs MANY years under the collective prospects' belts before we can come close to judging the efficacy of any draft model.
Crow
Posts: 10536
Joined: Thu Apr 14, 2011 11:10 pm

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Crow »

I looked at just one draft class, one year later. Obviously it is far from an ideal / full test. But it is a prediction test, better I think than no check and unable to do a better prediction test on these entries yet. Yeah some entries have a lot of retrodiction data for player performance further into their careers. Take each for what they are worth to you. It would perhaps be a mistake to over-react to my correlations, but it would perhaps be a mistake to not consider them either. Has anyone done any other prediction check on other draft class predictions or is everyone relying just on retrodiction quality? That is a different kind of questionable basis / potential big mistake.

If the goal is accurate prediction of actual production by year 3 or 4, then it might be worth modeling the expected intelligence / efficiency of the average team that is likely to make a draft pick at the range you expect a guy to go and adjust his expected production up or down according to that team impact. And I guess you might want to check the average team intelligence / efficiency of the guys used as significant similar for him. The quality of teams for the pick and his similar both affect their numbers and the reasonableness of the similarity. The college side and pro side. Several ways it might not be an equivalent / "fair" comparison and claim of similarity. Can we really separate player quality from context quality? The first things are recognizing the issue and then trying vs. not (which is what most or everyone is doing to my understanding).

How does the progression to year 3 or 4 or career peak go for top, middle and lower draft # picks on average and for full distribution and how does it go for projected high, middle and lower performers and how does it go for the 9 combinations? Again, if the goal is accurate projection of results, the prediction performance projection should account for expected pick # and the impact that has too on what is brought from (and when) from that player and their similars). So team impacts on players and similar and pick # impacts too, really the combination. Just predicting talent level is too simple for finding performance by year x. Maybe this part is beyond what anyone has tried to model? I am guessing so, based on not hearing it discussed. If you don't to do it, fine; but it will affect the outcomes so you are either changing the target or ignoring an aspect that affects the target. One could say it is too hard because every team is different and because traded picks affects the quality of team at spots outside the general pick # team quality curve. And I'd say yes they do, but that is part of the task. Adjust or ignore.

How does performance develop from year 1 levels? After you have year 1 performance data, which is better the draft projection or the year 1 data or what blend of the two at predicting what comes next? Predicting what comes after year 1 (or 2, 3, etc.) should be a contest too in the professional realm. Stay with the D Russell projection or modify it? Big decision. One among many.

Even if you remain focused on performance correlations in year 3, 4 or later, isn't it worth knowing how jumbled things are after year 1? Is this a normal level of jumble or higher than usual? Lots of levels to consider, if you want to get a full handle on what is happening and want to assess the value of the models and get the most out of the overall process.
Dr Positivity
Posts: 331
Joined: Thu Sep 20, 2012 6:44 pm

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Dr Positivity »

Top 60, didn't take the purist approach of just punch in the numbers and live with whatever comes out, but on the whole overwhelmingly data/retrodiction driven:

Code: Select all

1. PF Ben Simmons
2. PG Kris Dunn
3. SG Isaiah Whitehead
4. PG/SG Wade Baldwin
5. C Chinanu Onuaku
6. C Ante Zizic
7. PF/C Zhou Qi
8. PF Brice Johnson
9. SF Taurean Prince
10. PF Dragan Bender
11. C Georgios Papagiannis
12. C Ivica Zubac
13. SF Brandon Ingram
14. SG Ron Baker
15. SG Patrick McCaw
16. SF/PF Juan Hernangomez
17. PG/SG Alex Caruso
18. C Daniel Ochefu
19. C Jakob Poeltl
20. SG Thomas Walkup
21. PF/C Jameel Warney
22. SG Denzel Valentine
23. PF/C Robert Carter
24. PF/C Gracin Bakumanya
25. PF/C Thon Maker
26. PF Domantas Sabonis
27. PG/SG Jamal Murray
28. SG Buddy Hield
29. SG Terry Tarpey
30. SF Jaylen Brown
31. PF Marquese Chriss
32. SF Derrick Jones, Jr.
33. SG Furkan Korkmaz
34. C A.J. Hammons
35. PG/SG Gary Payton II
36. PG Kahlil Felder
37. PF Guerschon Yabusele
38. PG Nikola Ivanovic
39. SF Paul Zipser
40. C Diamond Stone
41. PF Pascal Siakam
42. C Deyonta Davis
43. C Cheick Diallo
44. SF/PF Jordan Fouse
45. PF/C Henry Ellenson
46. PG/SG DeJounte Murray
47. SG Caris LeVert
48. PF Stefan Jankovic
49. PF Petr Cornelie
50. SG Malcolm Brogdon
51. SF Damion Lee
52. SF Jarrod Uthoff
53. SG Malik Beasley
54. SG A.J. English
55. SF Troy Williams
56. SF Timothe Luwawu
57. PG Demetrius Jackson
58. SF DeAndre Bembry'
59. PG Fred VanVleet
60. PG Tyler Ulis
Crow
Posts: 10536
Joined: Thu Apr 14, 2011 11:10 pm

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Crow »

I haven't studied the individuals in draft class much but I noticed where you put Murray and Valentine. I don't have a firm opinion on where they should be ranked yet, but they seemed like cases worthy of my further study because of the best case scenarios that come to mind.
Dr Positivity
Posts: 331
Joined: Thu Sep 20, 2012 6:44 pm

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Dr Positivity »

Valentine is a difficult prospect, he has glaring weaknesses as a senior guard with low steal and block and when it comes to scorers the prospects with his profile of putting up 10-11 points per 40 minutes on .51-.52 TS% when they're a freshman or sophomore, but then by senior year are all of a sudden elite are ones to be wary off. The combination of low steal, block and poor early scoring would normally suggest bust. On the other hand there's just not many comparisons points for a player who's either been as good as him as he has 4th highest BPM since 10-11 with one of the players ahead of him being Walkup against lower competition, the others Kaminsky and Crowder, nor many wings who've approached either his shooting or his passing season let alone both at once. On top of everything else it's rumored he has Danny Granger's knees, which I didn't take into account in my ranking.

Murray would have rated lower than I had him, I gave him some benefit of the doubt because if Devin Booker becomes an all-star he will be one of the biggest outliers in my system for bad stats and NBA success. My theory is Booker's UK stats were tainted cause they only made him play as a spot up shooter when he had other talents. Therefore my benefit of the doubt with Murray would be that as a player projected as a possible PG coming in, playing with Ulis could have messed up his skillset and his stats. But I still wouldn't count on him to be worth a lotto pick. However by the time of rating non top 20 prospects, anything with high variance is probably a good thing. In a successful model the 30th ranked NCAA player or whatever should not a sharp bet to become a starter/all-star. However for example this 18 year old French big Bakumanya who there's no real data about (playing against other French prospects only in some development set-up they have, it looks like), by variance and lack of data alone maybe he is more likely to be an all-star, and certainly would take the absence of data over a prospect like Skal who has plenty, it just all says he sucks
kjb
Posts: 124
Joined: Fri Apr 15, 2011 6:40 pm
Contact:

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by kjb »

I'm working my way through this year's draft class. This is 4th or 5th season I've analyzed players (college only -- I haven't had time to dig into international players). This year I'm seeing big differences between where players rate in my analysis and where they're showing up in the various mock drafts.

Has anyone seen Thomas Walkup play? He's older and he played against weaker competition, but was very productive. Looks like he'll go undrafted. Might be a nice UdFA for someone.
Statman
Posts: 548
Joined: Fri Apr 15, 2011 5:29 pm
Location: Arlington, Texas
Contact:

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Statman »

kjb wrote:
Has anyone seen Thomas Walkup play? He's older and he played against weaker competition, but was very productive. Looks like he'll go undrafted. Might be a nice UdFA for someone.
Thomas Walkup is an exceptional player. I watched him a few times last season - including every minute his stand out play in the NCAA tourney against supposedly "better" competition.

There's an element of play that can't be easily pinpointed that results in exceptional production at the college level (adjusting for comp, etc) - it's a court awareness, hand eye coordination, alpha male type thing. Combine data (length, standing reach, etc) usually does it no justice. In football - it's like Tom Brady going into the draft. Or Tony Romo. There's just something, innate ability to feel the rush, read the defense, stay cool under pressure. In college basketball - it usually manifests itself in ACTUAL high level performances again & again that greatly outpaces many players who will be drafted well ahead of said player. Now, these players do tend to be older - so there's an easy excuse to overlook them. Draymond Green was one of these guys coming out of college. Denzell Valentine is one. Brice Johnson is one. Buddy Hield would have been one if he wasn't breaking three point records - which got him lottery attention.

Obviously my ratings love Walkup, my projections will love him (despite low three point attempts for a guard). He has a very nice shooting stroke - I don't see any reason why he won't be an effective 3pt shooter. To me, I FULLY EXPECT him to be a BETTER Dellavedova type player. He was easily THE star (NO teammates anywhere close to a next level player) on a top 30 college team. Stephen F Austin ended up MUCH better than LSU & Simmons (NOT saying he's a better prospect than Simmons btw, just making point about player impact). They were MUCH better than Washington - who somehow could have TWO players in the lottery who weren't even close to the best players on their own Huskies team. Stephen F Austin was a fairly high level college team almost solely because of Thomas Walkup. He's obviously a immense gym rat by looking at his level of usage, assist rate, mid range efficiencies. He's built like a 6'4" piece of granite.

Slightly related note - would Larry Bird or Charles Barkley be high lottery picks if they came out of college in today's game? Bird was OLD, slow, not athletic, and starred on a team not much different than Stephen F Austin. Would be an elite three point shooter - that'd maybe get him in the lottery?

Barkley was a 6'4" fat post player with little face up game yet - although he ran the court great for a fat guy. I can't imagine an NBA team giving him the time of day in today's draft (fear of injury, eating himself out of the league, not being able to defend any position).

In terms of college production, Bird & Barkley would have easily been right at the top of their draft classes according to my draft model. Teams OFTEN drafted back then more according to relative college production (although at times did a poor job of adjusting correctly for pace, SoS, etc - think Bo Kimble) - and it seems when they ventured away from that (like they do often now) they usually failed miserably (think Chris Washburn & Olowokandi). It seems when high school kids became part of the draft - NBA teams started to put MUCH more importance into combine results and private workouts, in many cases almost completely ignoring relative college production. I think this is part of the reason why the draft has turned so much into a crapshoot for many teams - it'd be like baseball teams almost completely ignoring minor league production when considering who to call up - relying instead on 40 times, bench press, shuttle runs, and private workouts to judge players. Makes it easy for a draft model like mine (through retrodictions) to historically outperform actual draft positions while ignoring 95% of the data (combines, workouts, interviews, medical records, etc) most NBA teams consider.

So, again, my draft model will love Walkup. My eye's saw a next level player - I saw a college player that produced in a way I would have expected a 23 year old Manu Ginobili to produce if he played on that SFA team - high motor, high usage, great efficiency across the board, good assist rate, etc. I'll bet about anything he'll prove to be well worth the "risk" of a late 2nd round pick or FA signing.
Dr Positivity
Posts: 331
Joined: Thu Sep 20, 2012 6:44 pm

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Dr Positivity »

Walkup's stats are as good as I get but doing it at 23 in the Southland conference is worth being concerned about especially with athletic and 3 point shooting concerns. Should have passing and basketball IQ ability though and can see him being a rotation player
Statman
Posts: 548
Joined: Fri Apr 15, 2011 5:29 pm
Location: Arlington, Texas
Contact:

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Statman »

Dr Positivity wrote:Walkup's stats are as good as I get but doing it at 23 in the Southland conference is worth being concerned about especially with athletic and 3 point shooting concerns. Should have passing and basketball IQ ability though and can see him being a rotation player
He has easily been the best player for SFA the last 3 seasons - leading them to an 89-14 record those seasons. It's not like this season was his break out year.

But yes - obviously his age, quality of competition, and his lack of three point proficiency are his three pretty big red flags - enough reason to slip all the way through the draft.

I haven't run it yet - oh the suspense - but I'm guessing he'll project north of an optimized 50 career WAR in my model. Looking at the small school guards who projected over 50 WAR in the retrodiction (1998-2014 drafts):

Steph Curry 200.7 (best projected guard ever outside of Kyrie Irving & his very small Duke data sample)
Delonte West 138.5
Gordon Hayward 112.8
Bonzi Wells 102.3
CJ McCollum 98.9
Jameer Nelson 98.5
Speedy Claxton 70.0
Larry Hughes 66.8
Damian Lillard 62.5
Nate Wolters 61.5
Eric Maynor 60.2
Elfrid Payton 51.6
Kyle Korver 50.6

That's the complete list. I'd say there's 1 "bust" there - if you can call a 2nd round pick a "bust" (Wolters). It is VERY difficult for a small school guard to project so well - the team SoS and their age are usually too much to get north of projected 50 career WAR. One has to be an exceptional college player. But, it does happen (about 2 guys every 3 drafts) - and the players that make the list all (so far) do something in the NBA.

Of course, after that, watch Walkup not project north of 50 WAR. Even if he doesn't, he'll be close.
Crow
Posts: 10536
Joined: Thu Apr 14, 2011 11:10 pm

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Crow »

Are J Winslow and S Johnson slow starters, were they given too many minutes, too big a role (for ill now, for good maybe down the road) or were they overrated by many draft experts (analytic and scouting)? In last season's draft analytic article Layne Vashro and Jesse Fischer ranked them as high as DX did. BPM and Nick Restifo rated them lowest but the players did way worse than that even, especially Johnson. Just misses, wait a few more years before positing or concluding much, or anything to learn or weight differently?
Crow
Posts: 10536
Joined: Thu Apr 14, 2011 11:10 pm

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Crow »

Seems like a fairly wide range of ranks for Deyonta Davis and Brice Johnson. Super surface review, I'd lean towards better end of their ranges or even better. Maybe I overrate tall athletic types. KJB, are either of these guys discrepancy cases between your analysis and conventional wisdom? Anyone else high on them or not? I see Dr. Positivity is high on Johnson but not so much on Davis.
Crow
Posts: 10536
Joined: Thu Apr 14, 2011 11:10 pm

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Crow »

Winslow and Johnson were both highly rank out of high school and went 1 year to a big name school. There is not much superlative to their stats except Winslow's 3pt fg% from NCAA distance. Neither took a lot of 3s. Both are strong but being strong in a tweener NBA body doesn't mean the same thing as it does in NCAA on a talented team mainly playing less talented teams. Looking at these details, I can see how they in sum could lead to them being overrated. A lot of analysis seems to be position neutral or blind. If Winslow and Johnson were evaluated as SFs I'd think their projection would be lower than under a system that wasn't position oriented. 3pt shooting is a big deal for NBA SFs and strength? Isn't really that much.


Also, does anyone have the data for and take the time to adjust NCAA 3pt fg%s for frequency of open shots? I'd think that would be a pretty important adjustment as some guys on potent NCAA offenses might get higher than normal open looks and some not so dangerous 3 pt shooters might get more open looks because they aren't that frequent or comparatively good with the shot, apart from the team effects.
Dr Positivity
Posts: 331
Joined: Thu Sep 20, 2012 6:44 pm

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Dr Positivity »

Crow wrote:Seems like a fairly wide range of ranks for Deyonta Davis and Brice Johnson. Super surface review, I'd lean towards better end of their ranges or even better. Maybe I overrate tall athletic types. KJB, are either of these guys discrepancy cases between your analysis and conventional wisdom? Anyone else high on them or not? I see Dr. Positivity is high on Johnson but not so much on Davis.
Deyonta is kind of a one category wonder for me. His block rate (3.9 blk/40 pace) is elite but his steals, assists, rebounding, and scoring combination is pretty blah. The poor offensive combination of a low FT drawing rate and lacking shooting range. He does have a low turnover rate but I don't value that stat as much as others

To put Brice Johnson's season in perspective the median for Anthony Davis, Kevin Love, Chris Bosh, Draymond Green, Lamarcus Aldridge, Derrick Favors, Paul Millsap, Carlos Boozer, David West their draft year was 1.35 steals, 2.1 blocks, 12.85 reb, 1.6 ast, 2.65 TOV per 40 minutes, and Brice put up 1.5 steals, 2.1 blocks, 15 rebounds, 2.1 assists, 2.5 TOV. Scoring is a complicated stat to compare across age and conference, but Brice put up 24.2 points on .65 TS%. He had a dominant season that maybe was a little overlooked because it only came in 28 minutes per game. If he played 35 MPG at the same rate he would have put up 21 and 13 and likely got a lot more hyped than his 17 and 10. With that said I don't love him as either a floor spacer or a defender which makes him a hard fit in the modern game (David Lee type Pts/Rebs player may be a possibility for him). He isn't a lost cause in either category though
Crow
Posts: 10536
Joined: Thu Apr 14, 2011 11:10 pm

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by Crow »

Thanks for the response.

I don't know Davis beyond his DX profile but I'd probably take him well before 40 and expect his per minute productivity to outperform most of those taken above him.
kjb
Posts: 124
Joined: Fri Apr 15, 2011 6:40 pm
Contact:

Re: APBR-DraftExpress 2016 NBA Draft Project

Post by kjb »

Statman wrote:
kjb wrote:
Has anyone seen Thomas Walkup play? He's older and he played against weaker competition, but was very productive. Looks like he'll go undrafted. Might be a nice UdFA for someone.
Thomas Walkup is an exceptional player. I watched him a few times last season - including every minute his stand out play in the NCAA tourney against supposedly "better" competition.

There's an element of play that can't be easily pinpointed that results in exceptional production at the college level (adjusting for comp, etc) - it's a court awareness, hand eye coordination, alpha male type thing. Combine data (length, standing reach, etc) usually does it no justice. In football - it's like Tom Brady going into the draft. Or Tony Romo. There's just something, innate ability to feel the rush, read the defense, stay cool under pressure. In college basketball - it usually manifests itself in ACTUAL high level performances again & again that greatly outpaces many players who will be drafted well ahead of said player. Now, these players do tend to be older - so there's an easy excuse to overlook them. Draymond Green was one of these guys coming out of college. Denzell Valentine is one. Brice Johnson is one. Buddy Hield would have been one if he wasn't breaking three point records - which got him lottery attention.

Obviously my ratings love Walkup, my projections will love him (despite low three point attempts for a guard). He has a very nice shooting stroke - I don't see any reason why he won't be an effective 3pt shooter. To me, I FULLY EXPECT him to be a BETTER Dellavedova type player. He was easily THE star (NO teammates anywhere close to a next level player) on a top 30 college team. Stephen F Austin ended up MUCH better than LSU & Simmons (NOT saying he's a better prospect than Simmons btw, just making point about player impact). They were MUCH better than Washington - who somehow could have TWO players in the lottery who weren't even close to the best players on their own Huskies team. Stephen F Austin was a fairly high level college team almost solely because of Thomas Walkup. He's obviously a immense gym rat by looking at his level of usage, assist rate, mid range efficiencies. He's built like a 6'4" piece of granite.

Slightly related note - would Larry Bird or Charles Barkley be high lottery picks if they came out of college in today's game? Bird was OLD, slow, not athletic, and starred on a team not much different than Stephen F Austin. Would be an elite three point shooter - that'd maybe get him in the lottery?

Barkley was a 6'4" fat post player with little face up game yet - although he ran the court great for a fat guy. I can't imagine an NBA team giving him the time of day in today's draft (fear of injury, eating himself out of the league, not being able to defend any position).

In terms of college production, Bird & Barkley would have easily been right at the top of their draft classes according to my draft model. Teams OFTEN drafted back then more according to relative college production (although at times did a poor job of adjusting correctly for pace, SoS, etc - think Bo Kimble) - and it seems when they ventured away from that (like they do often now) they usually failed miserably (think Chris Washburn & Olowokandi). It seems when high school kids became part of the draft - NBA teams started to put MUCH more importance into combine results and private workouts, in many cases almost completely ignoring relative college production. I think this is part of the reason why the draft has turned so much into a crapshoot for many teams - it'd be like baseball teams almost completely ignoring minor league production when considering who to call up - relying instead on 40 times, bench press, shuttle runs, and private workouts to judge players. Makes it easy for a draft model like mine (through retrodictions) to historically outperform actual draft positions while ignoring 95% of the data (combines, workouts, interviews, medical records, etc) most NBA teams consider.

So, again, my draft model will love Walkup. My eye's saw a next level player - I saw a college player that produced in a way I would have expected a 23 year old Manu Ginobili to produce if he played on that SFA team - high motor, high usage, great efficiency across the board, good assist rate, etc. I'll bet about anything he'll prove to be well worth the "risk" of a late 2nd round pick or FA signing.
Very interesting. I wondered because Walkup also shows up very high in my ratings, and I'd literally never heard the guy's name. It reminded me of the first time I rated draftees and Jae Crowder showed up near the top. One key difference: Crowder played in the Big East. For guys who face less stout competition, I usually look for them to dominate. That's exactly what Walkup did. Seems he'll go undrafted.
Post Reply