2013 NBA draft prospect rankings

Home for all your discussion of basketball statistical analysis.
Mathketball
Posts: 44
Joined: Tue Jun 21, 2011 4:48 pm

2013 NBA draft prospect rankings

Post by Mathketball » Sun Jun 23, 2013 9:08 pm

With the draft a few days away I thought it'd be a good time to post my 2013 rankings. I have a base formula for evaluating collegiate players numbers that is tweaked based on position. I've made a few adjustments since last year and have been pretty happy with the results. Previously I was undervaluing mid major stars like Damian Lillard so that was one thing I addressed. Also, I revamped how I handle post players that can shoot 3's.

Also, it used to be that a score of 1.6 or above was elite. To make it easier to read I adjust my results so 1.6 is now 100. So basically anyone scoring over 100 projects to be a very good player. I account for height but currently I don't account for athleticism so guys that were really good college players that lack the athletic ability to compete in the NBA will be overvalued (John Surna and Robbie Hummel are examples of this from last year).

Also, on my spreadsheets you'll see a column for "Success to Failure" ratio. This is a stat I developed for evaluating how efficiently a player ends possessions. It's the ratio of possessions ended successfully to possessions ended with a failure. It's basically Ast to TO ratio extended to the full offensive game.

Here's my 2013 results:
https://docs.google.com/spreadsheet/ccc ... GR3c#gid=0

My top 5 picks would be.
1. Porter
1a. Noel (if I was the Cavs I probably pick Porter just because of the injury. If healthy I like Noel a little more than Porter though. Either player fills a huge need for them so if they choose one of these two they should be in good shape.)
3. Burke. (He has the highest score but PGs are easier to find so that makes Porter and Noel more valuable IMO)
4. Michael Carter-Williams
5. Victor Oladipo

Guys I would avoid early.
Steven Adams
Shabazz Muhammad
Alex Len
Anthony Bennett (I'd like him at 10. Tope 5 I'm not sure about)

Sleepers
Reggie Bullock
Ryan Kelly (Could be this years most over valued player on my scales but I do like his game. The question is will he ever be able to rebound enough to stay on the court?)
Lorenzo Brown
Nate Wolthers
Ray McCallum
Andre Roberson
Carrick Felix
Chris Evans (Here's my deep sleeper. He'll go undrafted and have to come through the D-League. Played for 3 colleges and 4 years. Made dramatic improvements from his Jr. to Sr. season when he finally played for the same team. Has an NBA body and athleticism. I'd love to see how he scores on some of the other draft rankings I've seen on here.)

Here's my 2012 results updated with this year's changes.
https://docs.google.com/spreadsheet/ccc ... kaFE#gid=0

I remember last year someone on here questioned why Jon Henson was ranked so low. So I think it's worth noting that he jumped from about an 85 on the PF scale to a 97 after I ported this years changes back. Also, Lillard went from a 78 to a 92.

Thanks for reading. I'd love to hear and questions or comments anyone has about my rankings.

Brett.

VJL
Posts: 51
Joined: Mon May 20, 2013 11:18 am

Re: 2013 NBA draft prospect rankings

Post by VJL » Sun Jun 23, 2013 11:00 pm

Nice presentation of the ratings.

Can you explain a bit more what goes into your model? Is it a priori? Regression based? What variable or types of variables are included?

I built a couple regression models that come to similar conclusions to your ratings, so it would be interesting if you arrived there with a different approach

jpadula
Posts: 2
Joined: Sun Jun 23, 2013 7:57 pm

Re: 2013 NBA draft prospect rankings

Post by jpadula » Sun Jun 23, 2013 11:44 pm

Looks pretty interesting, I'm surprised to see that Shabazz Muhammed was as low as he was, but definitely looks like there is a method behind that claim. I'm wondering if there is a usage rate measure or someway to measure players based on their varying differences in terms of their usage rate. Obviously all players in the draft had major impacts on their teams offenses, but not all players had the same opportunities.
Check out my Basketball statistics Blog
The Numbers of Basketball
http://thenumberofbasketball.blogspot.com/

Mathketball
Posts: 44
Joined: Tue Jun 21, 2011 4:48 pm

Re: 2013 NBA draft prospect rankings

Post by Mathketball » Mon Jun 24, 2013 3:43 am

VJL wrote:Nice presentation of the ratings.

Can you explain a bit more what goes into your model? Is it a priori? Regression based? What variable or types of variables are included?

I built a couple regression models that come to similar conclusions to your ratings, so it would be interesting if you arrived there with a different approach
Thanks VJL.

Basically the best way I can describe my base formula is a weighted mean of box score stats. The weights differ based on the players position. I've determined the weights based on previous draft classes with an emphasis on stats that translate well from NCAA to NBA. So I would say it's heavily driven by previous results. Of course, when developing it, I wouldn't make a tweak that couldn't logically be justified just to obtain better results.

After I obtain the rating from the formula there are deductions for SoS (from Kenpom), Class (currently I use class not age but I'm considering changing that next year), and a slight deduction for college winning %. The maximum a player can have deducted from his score is 25. The deductions are based on a tier system so they're discrete in nature. Also, I group post players into tiers for their 3 point shooting based on 3's attempted per X number of minutes and what their 3pt % is. The result of this is added to their score.

I hope this answers your questions. I started reading through the thread with your results. It looks very interesting. I think our approach is different but it looks like we're looking at mostly the same stats. I always enjoy seeing other approaches that come to similar conclusions.

Mathketball
Posts: 44
Joined: Tue Jun 21, 2011 4:48 pm

Re: 2013 NBA draft prospect rankings

Post by Mathketball » Mon Jun 24, 2013 4:17 am

jpadula wrote:Looks pretty interesting, I'm surprised to see that Shabazz Muhammad was as low as he was, but definitely looks like there is a method behind that claim. I'm wondering if there is a usage rate measure or someway to measure players based on their varying differences in terms of their usage rate. Obviously all players in the draft had major impacts on their teams offenses, but not all players had the same opportunities.
Yes, Muhammad is definitely one of the more bold claims of my results. I think when you look at his draft resume his wingspan is one of his most intriguing attributes. As a SF he might be a little short but he'll be more than long enough to defend NBA SFs. From a statistical stand point the wingspan is actually concerning because that long reach didn't translate into steals or blocks in college nor did he rebound particularly well. That could indicate lack of effort or lack of instincts on defense. His lack of assists and fairly average FT shooting didn't help much either. I wouldn't be shocked if he ended up being a really good player but his floor could also be pretty low.

Currently I don't include usage rate in my formula. Instead I focus on efficiency mainly because I feel just because a player shoots a lot doesn't mean he should. One thing I've considered factoring in is [points per possession used] - [team points per possession] or the ratio of the two. This would benefit high usage players that make their team better by using a possession and hurt high usage players whose teams are more efficient when they're not involved. If I did this I would probably put players into usage tiers to determine how much this should impact their score so I can make it more important for high usage players.

Mathketball
Posts: 44
Joined: Tue Jun 21, 2011 4:48 pm

Re: 2013 NBA draft prospect rankings

Post by Mathketball » Mon Jun 24, 2013 4:27 am

Here's a link to last year's thread on my results.
viewtopic.php?f=2&t=7992

These are the big changes I've made since then.
* Added a multiplier so the results are easier to read (100's, 90's, 80's etc instead of 1.6's, 1.5's etc)
* Adjusted weights to make different positions more comparable. Before post players generally scored lower than guards and wings.
* Adjusted how I deduct for SoS so I'm not killing mid major players like Lillard and McCollum.
* Adjusted how I factor 3's into post players scores.

VJL
Posts: 51
Joined: Mon May 20, 2013 11:18 am

Re: 2013 NBA draft prospect rankings

Post by VJL » Mon Jun 24, 2013 12:36 pm

Great. Thanks.
and a slight deduction for college winning %
If I am reading this correctly, you debit players for being on winning teams?

One of the surprising results I have found in my work is that not only does being on a good team (high SRS) mean that, all else equal, you will do better in the NBA, but it also is a much more important predictor than SOS. When you include both SRS and SOS in a regression SOS often becomes a useless or even negative predictor.

I still haven't found a great explanation for this pattern. My first hunch was that it was subtly picking up on some unobserved talent because top recruits go to good schools, but the effect was unchanged when controlling for RSCI.

v-zero
Posts: 409
Joined: Sat Oct 27, 2012 12:30 pm

Re: 2013 NBA draft prospect rankings

Post by v-zero » Mon Jun 24, 2013 12:45 pm

VJL wrote: One of the surprising results I have found in my work is that not only does being on a good team (high SRS) mean that, all else equal, you will do better in the NBA, but it also is a much more important predictor than SOS. When you include both SRS and SOS in a regression SOS often becomes a useless or even negative predictor.

I still haven't found a great explanation for this pattern. My first hunch was that it was subtly picking up on some unobserved talent because top recruits go to good schools, but the effect was unchanged when controlling for RSCI.
SRS can be thought of, lazily, as SOS+MOV (it's not quite that, but it kind of is). That is to say that SRS takes into account the strength of schedule in such a way that each team can hence be measured from the same origin. SRS contains much more information than SOS, and all of the information that SOS contains.

VJL
Posts: 51
Joined: Mon May 20, 2013 11:18 am

Re: 2013 NBA draft prospect rankings

Post by VJL » Mon Jun 24, 2013 3:16 pm

SRS contains much more information than SOS, and all of the information that SOS contains.
I didn't realize SOS was embedded in SRS. That makes sense then. It is still interesting that the MOV component adds considerable information though. Even while controlling for RSCI and running SOS and SRS separately, SRS comes out as roughly twice as strong of a predictor.

v-zero
Posts: 409
Joined: Sat Oct 27, 2012 12:30 pm

Re: 2013 NBA draft prospect rankings

Post by v-zero » Mon Jun 24, 2013 5:34 pm

It captures all of the information not caught in the box score. Plus putting up great numbers amongst strong teammates is going to be harder in general than doing so on a weaker team, as great players on poor teams will have fewer opportunities taken away by teammates.

VJL
Posts: 51
Joined: Mon May 20, 2013 11:18 am

Re: 2013 NBA draft prospect rankings

Post by VJL » Mon Jun 24, 2013 6:49 pm

Plus putting up great numbers amongst strong teammates is going to be harder in general than doing so on a weaker team, as great players on poor teams will have fewer opportunities taken away by teammates.
Definitely. Diminishing returns in usage and rebounds are a big deal. This leads to an interesting question though. One of draft analysts' favorite excuses for players is "he was surrounded by bad players who could 'get him the ball"/"led him to force bad shots'". This theory would run the opposite direction. I suppose both factors probably exist simultaneously but on net the diminishing returns problem wins out.

The capturing non box-score stuff I am less certain of. It was my initial expectation, but because controlling for RSCI does little to impact the effect I don't know... I suppose it would capture +/- impact in a very sloppy fashion.

Statman
Posts: 548
Joined: Fri Apr 15, 2011 5:29 pm
Location: Arlington, Texas
Contact:

Re: 2013 NBA draft prospect rankings

Post by Statman » Mon Jun 24, 2013 10:04 pm

jpadula wrote:Looks pretty interesting, I'm surprised to see that Shabazz Muhammed was as low as he was, but definitely looks like there is a method behind that claim. I'm wondering if there is a usage rate measure or someway to measure players based on their varying differences in terms of their usage rate. Obviously all players in the draft had major impacts on their teams offenses, but not all players had the same opportunities.
I stated months ago in my blog - I have serious doubts Shabazz will be able to stick at all in the NBA.

Statman
Posts: 548
Joined: Fri Apr 15, 2011 5:29 pm
Location: Arlington, Texas
Contact:

Re: 2013 NBA draft prospect rankings

Post by Statman » Mon Jun 24, 2013 10:06 pm

v-zero wrote:
VJL wrote: One of the surprising results I have found in my work is that not only does being on a good team (high SRS) mean that, all else equal, you will do better in the NBA, but it also is a much more important predictor than SOS. When you include both SRS and SOS in a regression SOS often becomes a useless or even negative predictor.

I still haven't found a great explanation for this pattern. My first hunch was that it was subtly picking up on some unobserved talent because top recruits go to good schools, but the effect was unchanged when controlling for RSCI.
SRS can be thought of, lazily, as SOS+MOV (it's not quite that, but it kind of is). That is to say that SRS takes into account the strength of schedule in such a way that each team can hence be measured from the same origin. SRS contains much more information than SOS, and all of the information that SOS contains.
Agreed. I use Sagarin instead of SRS, I think it's a little better indicator on team quality - but same idea. SoS is obviously inherent within team ratings (SRS or Sagarin or Pomeroy or whomever).

Statman
Posts: 548
Joined: Fri Apr 15, 2011 5:29 pm
Location: Arlington, Texas
Contact:

Re: 2013 NBA draft prospect rankings

Post by Statman » Mon Jun 24, 2013 10:55 pm

Mathketball wrote:With the draft a few days away I thought it'd be a good time to post my 2013 rankings. I have a base formula for evaluating collegiate players numbers that is tweaked based on position. I've made a few adjustments since last year and have been pretty happy with the results. Previously I was undervaluing mid major stars like Damian Lillard so that was one thing I addressed. Also, I revamped how I handle post players that can shoot 3's.

Also, it used to be that a score of 1.6 or above was elite. To make it easier to read I adjust my results so 1.6 is now 100. So basically anyone scoring over 100 projects to be a very good player. I account for height but currently I don't account for athleticism so guys that were really good college players that lack the athletic ability to compete in the NBA will be overvalued (John Surna and Robbie Hummel are examples of this from last year).

Also, on my spreadsheets you'll see a column for "Success to Failure" ratio. This is a stat I developed for evaluating how efficiently a player ends possessions. It's the ratio of possessions ended successfully to possessions ended with a failure. It's basically Ast to TO ratio extended to the full offensive game.

Here's my 2013 results:
https://docs.google.com/spreadsheet/ccc ... GR3c#gid=0

My top 5 picks would be.
1. Porter 186 rating, ranked 3rd nationally
1a. Noel 160, 26th
3. Burke. 189, 2nd
4. Michael Carter-Williams 165, 20th
5. Victor Oladipo 174, 9th

Guys I would avoid early.
Steven Adams 139, 144th
Shabazz Muhammad 125, 355th
Alex Len 144, 105th
Anthony Bennett 159, 28th

Sleepers
Reggie Bullock 153, 46th
Ryan Kelly 153, 48th
Lorenzo Brown 136, 178th
Nate Wolthers 165, 19th
Ray McCallum 147, 81st
Andre Roberson 155, 37th
Carrick Felix 132, 243rd
Chris Evans. 146, 92nd (EDIT 6-26-13, previously posted rating of a different Evans)


I put my national rating (100 around D1 average) and ranking (out of 2993 players) in "Impact" (ignoring games missed due to injury and or suspension) next to all your projections above - just for fun. My impact ranking is PURELY a performance ranking in college - it's not supposed to be a predictor of NBA success. But, historically, guys that rank higher in my ratings are definitely more likely on average to be successful at the next level. This last year's ranking (in HnR - which missing games hurts a player's ranking - "Impact = HnI" is the ranking on the far right for all players) of all D1 players:

http://www.hoopsnerd.com/uploads/4-8-13 ... atings.pdf

Now, I'm curious what you think of my highest HnI guys that had GREAT seasons last year:

#1 Mike Muscala 194 rating
#4 Kelly Olynyk 182
#5 Erick Green 179
#7 CJ McCollum 175
#8 K. Caldwell-Pope 175
#10 Cody Zeller 173
#11 Mason Plumlee 171
#12 Taylor Smith 171
#14 Gorgui Deng 168
#15 Pierre Jackson 168
#16 Jeff Withey 168

To comment on your post - I wouldn't drop Burke because there are other PGs - PGs who are ELITE in college in BOTH scoring and assist rate against good/great comp usually become good/great NBA players. However, wings are more a dime a dozen imo - although Porter is a STUD.

I can't really complain about your top 5 - looks as good as anybody's to me.

I agree wholeheartedly with your avoidees EXCEPT for Bennett - he was a true frosh. True frosh that are THAT dominant (they MOVED ASIDE - changed positions and gave more pine to their star from the previous year Mike Moser - BECAUSE Bennett was such an immense talent) HAVE to be taken seriously.

The idea of Len possibly going #1 makes me laugh - there are plenty of bigs in this draft that will have that will have better NBA careers if given the opportunity imo.

I'm cool with your sleepers except Brown & Felix - look at the 11 names I listed for alts (Green/McCollum/Jackson for Brown, Caldwell-Pope for Felix).

Great job though - I don't think people really how tricky it is to really separate players with similar looking college stats.

My NBA projections for college players is really down the road at least a month btw - still compiling the 15 years of college data. Didn't come close to my dream deadline of a few days before the NBA draft. Oh well.
Last edited by Statman on Thu Jun 27, 2013 4:41 am, edited 1 time in total.

bchaikin
Posts: 286
Joined: Thu May 12, 2011 2:09 am

Re: 2013 NBA draft prospect rankings

Post by bchaikin » Tue Jun 25, 2013 1:57 am

so is victor oladipo the 2nd coming of sidney moncrief?

Post Reply