2014 draft prospect rankings

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

2014 draft prospect rankings

Post by Mathketball » Sat Jun 21, 2014 7:50 pm

Hi all,

I finally finished putting together my prospect rankings for this draft. A brief explanation of my formulas. The base formula is similar to a weighted mean of stats with adjustments for Height, SoS (from Kenpom), years in college, and team performance. I adjust the weights based on position so it's really 5 different formulas.

I've made several enhancements to my formulas this time around to help curb inflated scores, increase the standard deviation of value added by certain stats, and improved how the 5 formulas compare across the board. In the past SFs and PFs skewed a little lower, this is no longer the case.

I don't include any athletic measurements (vertical, wingspan, etc) mainly because they aren't known for all the players. Because of this, players that lack NBA athleticism can be overvalued. Players that play very good on ball defense can tend to be undervalued a bit as this doesn't necessarily result in steals or blocks and my formulas are based on individual stats.

The result of my formula is displayed in the "Rating" column of the google docs below. Any player with a score over 100 should be considered a really good prospect. In the 90's would be players I'm confident can stick in the NBA.

You'll see another column titled "Success to Failure Ratio". This is a stat I've developed for evaluating a players total offensive efficiency. It's the ratio of possessions a player ends successfully to the number of possessions he ends with a failure.

Here's my 2014 sheet:
https://docs.google.com/spreadsheets/d/ ... 1547496129

Here's my Archives. This includes the full draft prospect pool from 2010-2014 as well as some notable players from drafts before that. Any player highlighted in gold has made an all-star game. Players highlighted in red pulled out of the draft that year.
https://docs.google.com/spreadsheets/d/ ... 1547496129

My top 5 picks would be:

1. Joel Embiid (if healthy): His score is second only to Spencer Dinwiddie (we'll get into that in a minute). In my Archives only Anthony Davis and Kevin Love had higher C scores. Not only do I think he's the best player in the draft but he's also an excellent fit for the Cavs at #1 because of their desperate need for shot blocking to protect Kyrie Irving's poor on ball defense. They passed on Nerlens Noel at #1 last year and Gorgui Dieng at #19 last year so will they continue to ignore their biggest need, a rim protector? Unfortunately the injury concerns are the big question here. I'm not a doctor so I won't weigh in on that but it would be understandable for that to drop him down the board. No one wants the next Greg Oden. One other note on Embiid, his foul numbers were astoundingly high. I couldn't find a good comp for him in that regard so it's tough to say what to make of that. Maybe this can be explained by the NCAA stressing hand check fouls, or maybe it's because he's only been playing basketball for 3 years. Between fouls and injuries there are plenty of reasons to worry he won't be able to stay on the court. He's an incredibly interesting prospect.

2. Andrew Wiggins: He's the 5th best SG (108) and 5th best SF (96) so why would I put him at #2? His reputation for on ball defense leads me to believe he is being undervalued a bit. I'm a firm believer that statistical analysis should go hand in hand with scouting. His score is solid, so I don't see a reason to drop him out of the top 3.

3. Jabari Parker: He's the 4th best SF (97). I didn't put him in my PF scales because I fully believe he's a SF. Him and Wiggins are so close I gave the nod to Wiggins mainly because he's considered a better on ball defender. Similar to Wiggins, Parker's score is good enough that I don't see a reason to drop him from the top 3. I will say both Parker and Wiggins scored lower than just about all the really good SF's in my Archives (Granger, Durrant, Hayward, Anthony, Leonard, George, and Gay). So there is reason to be concerned that their ceilings might not be as high as people believe.

4. Marcus Smart: He's the 3rd highest score on my PG scale (120) behind only Dinwiddie and Irving and 3rd on the SG scales (119) behind only Dinwiddie and Oladipo. Basically I love him. Dante Exum will likely be drafted ahead of him. I don't rank international players so I won't comment on if I believe Exum will be better than Smart but I will say I think Smart is going to be a great player.

5. Nik Stauskas: 5th highest SG in my archives (118) behind Dinwiddie, Oladipo, Smart, and Turner. I would put Vonleh as a close 6th but I really like them both.

Projected high picks that scored low:
1. Doug McDermott: Scored an 84 on the SF formula. I mentioned that players who lack NBA athleticism can be overvalued because athleticism doesn't factor in. That makes McDermott even more concerning. I'd be afraid of him in the top 10 or 15 picks and if he didn't stick in the league I wouldn't be shocked.

2. Aaron Gordon: Scored an 88 on the PF formula. I would be nervous about him in the top 10 but not as nervous as McDermott. Since he's a plus athlete he could be undervalued here.

3. Zach LaVine: Scored an 80 on the PG formula and the SG formula. I didn't add his SG score to the sheet but it was nearly identical to his PG rank. His success to failure ratio is almost 1 to 1 so this explains his score. I would still be concerned taking a guy in the top 15 or 20 with that score.

4. T.J. Warren: Scored an 83 on the SF scale. His low Success to Failure ratio and Assist to Turnover ratio can be attributed to this.

Guys outside the top 15-20 projected picks that stood out:
1. Spencer Dinwiddie: Top PG (141) and SG (145) in my archives. Since he didn't play a full season due to his injury there is reason to believe he was playing at an unsustainable level. I put his 2013 numbers in where he scored a 114 as a PG and a 117 as a SG. So he might be overvalued a bit but I still really like him. If he can come back from the ACL injury he could wind up being the steal of the draft. If I've got a pick late in the first round and he's there, I don't think twice about drafting him.

2. Kyle Anderson: 2nd highest SF in my archives (132). He could fall into the category of guys that are overvalued because of lack of athleticism but I still believe he's going to be a really good player. I think his long term future will be at the 3 but his score as a PF was very good as well (6th highest in archvies at 130). If I'm in the second 5 and Embiid, Wiggins, Parker, Smart, Stauskus, Vonleh, and probably Exum are off the board, I take Anderson. Probably higher than he'll go, but I love him. An interesting note here, if the Cavs trade the first pick for the 3 and 10 from Philly and take Embiid anyways, Anderson might be a great fit for them at 10. Irving and Waiters are below average distributors so an elite passing SF who has experience coexisting with 2 talented guards might make a ton of sense for them.

3. Khem Birch: #2 PF in this class on my scales at 114. He will be a late 2nd rounder pick because of his slender build (209 lbs) but I think down the road he could be a really nice player.

4. Jordan Adams: Scored a 115 on the SG scales. Good for 6th best overall in the draft.

5. Tyler Ennis: Scored a 111 on the PG scales. Good for 8th best overall in the draft.

6. Shabazz Napier: Scored a 107 on the PG scales. Good for 10th best overall in the draft.

7. Markel Brown: Scored a 107 on the SG scales. Good for 11th best overall in the draft.

8. Taylor Braun: Scored a 104 on the SG scales and a 99 on the SF scales. Good for 13th and 20th best overall in the draft.

9. DeAndre Kane: Scored a 103 on the PG scales. Good for 14th best overall in the draft.

10. Kendall Williams: Scored a 103 on the PG scales. Good for 15th best in overall in the draft.

Gary Harris (103), Rodney Hood (102), Sim Buhllar (102), and K.J. McDaniels (100) were the others to break 100.

Here's the thread I started last year on my formulas: viewtopic.php?f=2&t=8267

I'd love to hear everyone's thoughts on my work and if you have your own rankings to compare to mine I'd be very interested to see those as well.

Crow
Posts: 6248
Joined: Thu Apr 14, 2011 11:10 pm

Re: 2014 draft prospect rankings

Post by Crow » Sat Jun 21, 2014 10:43 pm

Thanks for the results and write up. I. Take it you don't want to reveal the stat weights? That is the part that draws my greatest interest.

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

Re: 2014 draft prospect rankings

Post by Statman » Mon Jun 23, 2014 4:51 am

Your success to failure thing seems to hate high usage guys (who tend to be a bit lower in efficiency) who don't get bunch of assists. I think this may make your overall ratings value lower usage guys a bit too much. Just my thought at first glance.

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

Re: 2014 draft prospect rankings

Post by Mathketball » Tue Jun 24, 2014 1:00 am

Crow wrote:Thanks for the results and write up. I. Take it you don't want to reveal the stat weights? That is the part that draws my greatest interest.
I'd prefer to avoid posting my exact formulas but I don't mind sharing more about the weights. The weights aren't all linear so the easiest way to discuss them is what percentage of the final score comes from each. The reason I made some of them non linear is to increase the variance in the interquartile range and decrease how much of an impart outliers have.

Here's the percentage breakdown with the standard deviation in parenthesis so you can see which categories are causing the biggest difference in scores. Note, I multiple the final result by a scalar to make the final scores more presentable (around 100). The StdDev below is before multiplying so that's why they seem low.

Based on around 120 PGs in my archives the weight breakdowns are as follows:
PPS: 21.2% (0.92)
FT Shooting: 8.4% (0.40)
3Pt Shooting: 8.8% (0.26)
Rebounding: 11.8% (0.42)
Steals: 9.1% (0.38)
Blocks: 3.7% (0.39)
Ast:To: 26.4% (1.07)
Height: 10.5% (0.26)

Based on around 147 SGs in my archives the weight breakdowns are as follows:
PPS: 23.6% (0.93)
FT Shooting: 8.5% (0.38)
3Pt Shooting: 8.4% (0.27)
Rebounding: 14.6% (0.60)
Steals: 8.2% (0.31)
Blocks: 6.1% (0.58)
Ast:To: 19.9% (0.94)
Height: 10.7% (0.25)

Based on around 153 SFs in my archives the weight breakdowns are as follows:
PPS: 24.4% (1.05)
FT Shooting: 7.2% (0.39)
3Pt Shooting: 7.9% (0.42)
Rebounding: 17.2% (0.66)
Steals: 7.5% (0.40)
Blocks: 7.9% (0.67)
Ast:To: 16.2% (1.04)
Height: 11.6% (0.18)

Based on around 149 PFs in my archives the weight breakdowns are as follows:
PPS: 25.3% (1.21)
FT Shooting: 8.5% (0.49)
3Pt Shooting: 3.6% (0.54)
Rebounding: 21.2% (0.84)
Steals: 5.7% (0.33)
Blocks: 12.7% (0.96)
Ast:To: 12.6% (0.67)
Height: 10.5% (0.19)

Based on around 75 Cs in my archives the weight breakdowns are as follows:
PPS: 25.9% (1.70)
FT Shooting: 8.3% (0.54)
3Pt Shooting: 1.8% (0.40)
Rebounding: 21.3% (0.81)
Steals: 5.4% (0.34)
Blocks: 14.4% (0.91)
Ast:To: 12.3% (0.82)
Height: 10.5% (0.20)

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

Re: 2014 draft prospect rankings

Post by Mathketball » Tue Jun 24, 2014 1:15 am

Statman wrote:Your success to failure thing seems to hate high usage guys (who tend to be a bit lower in efficiency) who don't get bunch of assists. I think this may make your overall ratings value lower usage guys a bit too much. Just my thought at first glance.
I actually don't include Success to Failure ratio in my formula. I've never done research to see how well it translates to the next level so I don't feel comfortable including it in my rating formula right now. I just include it on the sheets as a way of showing how efficient the player ended possessions. In the original post when I mentioned Success to Failure as the reason some guys scores I meant it indicated they weren't being particularly efficient which was reflected in their rating. Sorry, I didn't make that very clear in the original post.

With that said, you might be right that I undervalue high usage players. I'll look into that more.

Crow
Posts: 6248
Joined: Thu Apr 14, 2011 11:10 pm

Re: 2014 draft prospect rankings

Post by Crow » Tue Jun 24, 2014 1:45 am

Thanks. I'll think about those weights for possible questions or comments.

At first blush the way the weights change by position makes sense to me in all cases with the possible exception of PPS.

Height slightly more important at SF than elsewhere? Is that deliberate or a function of other choices? Not a big deal, just asking.

Use of shooting percentages and A/TO without some consideration of usage is an issue I'd think about. On that I agree with Statman.

Actually now I have to ask if 3pt shooting and FT shooting just measure success rate or also look at frequency of attempts, on its own or relative to 2 pt attempts?

In general, sounds pretty good. Those are my first reactions for your consideration.

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

Re: 2014 draft prospect rankings

Post by Mathketball » Thu Jun 26, 2014 2:30 am

Crow wrote:Thanks. I'll think about those weights for possible questions or comments.

At first blush the way the weights change by position makes sense to me in all cases with the possible exception of PPS.

Height slightly more important at SF than elsewhere? Is that deliberate or a function of other choices? Not a big deal, just asking.

Use of shooting percentages and A/TO without some consideration of usage is an issue I'd think about. On that I agree with Statman.

Actually now I have to ask if 3pt shooting and FT shooting just measure success rate or also look at frequency of attempts, on its own or relative to 2 pt attempts?

In general, sounds pretty good. Those are my first reactions for your consideration.
SF height being slightly higher isn't intentional. I think the cause is the fact that the average height between 2's and 3's increases more than any other positions from all the players in my archives. The average heights are as follows listed from Pg to C. 74.4, 76.5, 79.3, 81.1, 83.3. When I adjust the height formula per position I probably need to account for the jump between 2 and 3 more than I do.

3pt shooting and FT shooting just measure success rate for PG, SG, and SF. For PF's and C's I break them up into tiers based on the number of 3's they attempt per game and their percentage. I've been toying with another way to score 3pt shooting that would better account for volume across all positions. So that's definitely something I want to address.

A lot of FTA or shooting a lot of 3's at a decent percentage will certainly help PPS. So the overall formula isn't completely ignorant of FT and 3Pt volume. Ultimately I think it might be a good idea for me to move away from PPS and move to PPWS and explicitly include FT and 3Pt volume elsewhere.

Thanks for your thoughts. I appreciate your insight.

Post Reply