Arturo avatar

The 2014 Draft Preview Part 3: The Deepest Draft?

(Editor Arturo's note: I screwed the data for two players. Kyle Anderson from UCLA and Spencer Dinwiddie from Colorado. I have fixed this as of 2AM EST on 6/26/14)

C-3PO: Sir, the possibility of successfully navigating an asteroid field is approximately three thousand, seven hundred twenty to one. 
Han Solo: Never tell me the odds!

— Star Wars Episode V: The Empire Strikes Back
The draft in the end comes down to understanding and managing the odds. 
In the first two (and a half) parts (here, here and here), I built and shared with you the tools I would build to analyze and understand the odds and the value of each and every player. You want to know about NCAA players?

NCAA Interactive Player Ratings



NCAA Player Comparison Tool

You need to see the international and the Dleague players as well?

Euro/Dleague Interactive Player Ratings


With these tools in hand I can see and compare every prospect (with one notable exception) for the 2014 draft and I can make the most informed analysis possible. 

Thing is analysis of the odds is not enough. There is a point when you have to make a decision. To paraphrase Mark Twain, 99 percent of success is built on failure. You have to take chances to succeed. The draft is a key illustration of this. You're queasy about taking the potential franchise altering center because of his injury history? The tall point guard intrigues you but you have zero meaningful data on him? The killer Euro prospect is enticing but euro data is somewhat murky? Tough, you need to make a choice and take a risk.
The trick is to manage that risk appropriately. Never put all the eggs in one basket. The wrong approach is to bet everything on one pick. The right approach is to diversify and increase the amount of plays that you make. Look at what the Sixers, Celtics and Suns are doing. The Sixers in particular are super interesting. They won last year's draft handily and there is a real chance they're about to repeat. I could totally see them winding up with the three best players in the last two drafts by the end of the night. 
We will be here to talk you thru it.
Let's get to the final draft rankings shall we? We will be using two flavors to rank the prospects this year depending on the professional league they play in. For all those plying their craft under the yoke of the NCAA, I'm using my updated version of the two models to predict the future performance of NBA draft picks (go here for the model build parts 1 & part 2  and here for the detail on the update). In very general terms, the models use the available data to predict future performance for each player coming into the draft from the NCAA. Based on that prediction, a ranking is done and a draft recommendation is generated.

It has performed at a very high level. For the full history you can go to:

A neat little summary of how the model does is this:

As promised in a previous post, I will be updating this to include the 2010 class and doing a review on the 2013 class over the summer as it really deserves its own post. 

European and DLeague prospects will be rated based on their actual professional competition numbers as projected to the NBA as shown in part two of the analysis. I will be doing this for all ranked prospects in the Draft Express database as well as any seniors that the draft model recommends be drafted. I did model every possible player but 1000+ prospects do not show well on a table. Feel free to request any player that you might like to see in the comments and I'll share their numbers.

There are four notable exceptions. Dante Exum (#5 in Draft Express), Artem Klimenko (#38), Cristiano Felicio (#78) and Maximilian Kleber (#85), where the data is either non-existent (Exum and Artem) or from a less than great league (Cristiano and Maximilian). For the last two it's not a real problem, as if you're interested, you can probably just invite them to summer league and your Dleague squad. Artem you can have for a second rounder but there are safer options that will be available. Exum is the big question mark. I'm going to link to the excellent piece Ryan Russillo did on Grantland. Here's the critical exchange for me:

Scout 3: I have no comparisons for him. None.

Unless you are some super-scout that saw him on some JV vocational team in Australia, you have only seen him live twice in like 14 months. The workouts give you nothing; we already know he’s big and an athlete. It’s a risk, but last time we saw him at the under-19 he dominated.

It’s all upside, but he’s never played against men. He has all the physical skills, great feel for the game, so that is where you start when you think about position in the NBA. We project PG will be his primary position.

To think he’s the third pick in the draft, and hasn’t played in 12 months … man, that is impressive.

Russillo: I swear we should just replace the word “upside” with the phrase “we don’t know how bad he could be.” Like everyone else, I’m not sure. He is a product of human nature; we don’t know enough about him, so we assume he is going to be good.

Caveat emptor. I wouldn't touch this kid with a ten foot pole. Way too much risk for the money, guards are a dime a dozen in the NBA,there are better and safer options in Europe

If I'm going to risk it all on the mystery box, I'd prefer it to contain a freakishly athletic big man. 

Let's get to the numbers. We start with all the NCAA ranked prospects and seniors rated as picks by our model. You'll note that unranked Seniors will show up highlighted in black. I've also put in the net player rating for the 2013-14 NCAA season as per part 1, this is to give you an idea of how productive the players actually were in college. 

We have 18 two star prospects (i.e selected by the two models) and 10 one star prospect (selected by only one) amongst the NCAA players. Of these, 10 and 4 are likely to see the court in an NBA game. How does this compare historically?

That would prospectively make this the best draft in our numbers. It gets even more fun when we add in all the international prospects.

Here comes the full draft table.Please note that I've divided prospects into 6 categories:

  • 1-Great: Zero red flags. Models love him. Statistics want to date him.
  • 2-Good: One red flag. Could be less than stellar performance in league. Could be one model passing on him. Could be too little minutes.
  • 3-Rotation: 2 red flags but with some indicator of good performance.
  • 4-Potential: 3 red flags but with some indicator of good performance.
  • 5- Bench: No indicators of good performance but a projection that says he'll be an effective bench player. See Young, James.
  • 6- Bust: Zero indication of even mediocre performance. If your team is drafting one of these player's , vaya con Dios my friend. See Hairston, P.J.

A nice graphical way to view this follows.

Let's break it down by position so I can then give my two cents. We start, as you always should in the draft, with the bigs.

  • Capela is easily the best big prospect and I'd have him number 1 on my draft board. The knocks on him (shooting, turnovers) are things that will go away with proper training. You can't teach tall, handle,athleticism, motor or ball hawking and he's got those in spades and he's already using them to produce at 20. If the Celtics pass with him, I may need to break up with Danny Ainge.
  • Tavares and Nurkic both look to be great value as well. I'd avoid drafting the rest of the euro bigs.
  • Embiid looks every bit the franchise changing big if he can stay healthy and on the board. You also have to play the long game with him and rehabilitate him properly. I'd need a detailed medical report but I'd definitely draft him number 1 if I were Cleveland. I expect he'll go with the three to Philly but we will see.
  • Vonleh, Stokes and Birch are all two model recommendation bigs with sparkling NCAA production. Good value picks at their slots.
  • Single recommendation players are Mitch McGary and Randle. I'd recommend Randle with zero issues. McGary has had some medical issues and a low minute sample. Again, consult with your medical expert. 
  • The players to avoid are: Aaron Gordon (short, young, unproductive), Adreian Payne, Patric Young, McAdoo, O'Bryant,Powell and Cory Jefferson (older and unproductive). 

Let's do the wings next:

  • Kyle Anderson's numbers are killer across the board and he rates higher than his numbers given that he put them up on a loaded UCLA team. Definite must draft candidate and a great value at any pick.
  • Jordan Adams is loved by the models and his stats are stellar. He's also favored by the fact that he played on a very good UCLA team as that is generally an indicator that the model may be undervaluing him. Great late first round value.
  • Dario Saric is another easy recommendation. Good size, good numbers and a decently large sample size in a competitive pro league. The only thing that stops it from being higher is I wish his production was a bit higher.
  • Gary Harris is the other two model recommendation for the wings. His numbers in the NCAA were a bit lower which is why I have him as a second tier recommendation. 
  • Doug Mc Dermott does not quite get a full recommendation from the model.
  • Spencer Dinwiddie is porjected as an early second round steal.
  • Are there ever some landmines here. Jabari Parker looks to have some potential but he's severely overvalued based on his numbers in the NCAA. The numbers suggest he will be a good NBA player eventually but right now? Not so much. The numbers are less kind to Andrew Wiggins. He was not great in the NCAA. Spending a lottery pick on him is downright irresponsible. I think both these guys will be NBA level talent but not for the team that drafts them.
  • Some other to note: Nick Staukas (just ok), TJ Warren (bad across the board), Jerami Grant ,K.J. McDaniels, James Young and Zach Levine (enough potential to warrant a Dleague shot but not a lottery pick).
  • P.J Hairston is terrible and if the Celtics draft him, I will become a Spurs fan full time.
  • Every other wing is just no. 

Let's end with the smalls:

  • Payton, Napier and Smart are one, two and three in some order. The models love all three and their numbers are very similar. I'd give the edge to Payton probably on size and that he's going to be relatively inexpensive. 
  • Tyler Ennis is good but young. The model docks him for his youth. He might not be that good on his rookie contract. 
  • All the other smalls are so/so and I wouldn't spend any draft capital on them. You can find and sign that level of player on the street or in Europe and develop them in the Dleague (remember Jeremy Lin).
  • We talked about Dante Exum before. Zero data on him. If you draft him, you better be prepared for something like the Otto Porter, Darko, or Anthony Bennett experiences. I would not take him. (Update: I ran the numbers from Exum at the Under 17 and Under 19 World Championships. I get a 6.28 Win Score per 40 in a very small sample size of 487 minutes. If those numbers were taken as similar to NCAA numbers,Dante would rate as the fifth small behind Ennis and be recommended by one of the Models. The implication seems to be that he is not quite ready and you might not get full value for the pick. I would still recommend a pass. Better options are available.)

With that we come to a close. I swear I've put a month's worth of work into this post.

Oh, there is one more thing. Here is everything in a worksheet so you can play with it and follow along during the draft. 

Cleveland you are officially on the clock now.


Thank you Arturo. I don't post much at all, but read all your stuff. Truly awesome work.

Will need some time to sort through, but quite shocked that so many of the likely top 5 are projected much worse. Teams 5-10 are sitting nicely, but you never know on draft night. Not so sure of Boston ;)

Question: You say "I'd rate Anderson as the third best wing in draft about even with Harris", which is not what your model says. Is that only because he played on a loaded team (and the model needs some tweaking for this)? Any other players that could use some correction?
great preview as always arturo, thanks. with this loaded draft and possible big name movement (lebron, melo, love, chandler, others(?)), will be fun to see how this next era in the NBA shapes up...
What the hell happened to Kyle Anderson?! He was your TOP DE ranked NCAA prospect not named Joel Embiid in part 1! He's got a 98% net rating. But now he's a rotation guy? How does that happen?
He's was on a really good team. In the NCAA, we've found that depresses the stats for the model. The WP stats (which are fairly new) adjust for this and thus rate him higher. He's darn close on the models and given the fact that he played on a good team and rocked it in the NCAA, I'd be perfectly happy to have him on my team.
Nice, I've been pimping Capela/Adams/TJBray since just before the Tourney (and to a lesser extent Russ Smith). Nice to be able to point to some back up.

I thought Dinwiddie was a pg? Does your model just say he plays like a 2 guard or are you projecting him there because of his size or am I just not correctly remembering what position he was playing?
I'm also very confused as to how Part 1 relates to Part 3. I think this deserves an explanation. Otherwise, I very much enjoy this, though to my mind it places far to much emphasis on later seasons and doesn't acknowledge the risk/reward of very young prospects like Aaron Gordon. Something seems off about this sentence "Tyler Ennis is good but young."
Hey man, thanks. Thanks for all this work, this data, and these tools. Seriously. You rock.
Agreed. Thanks, Arturo.

For anyone who likes their spreadsheets all up in the cloud, I trimmed the rows and added Google Sheets friendly filtering here.
I just don't understand how Dinwiddie is so low, exactly. Doesn't turn the ball over much, has historically good shooting numbers, steals, etc. Is it THAT big of a knock to not rebound well?
Conterintuitively, the model wants more production from taller and older players. So even though his numbers were good, the models want more from a player of his profile.

Honestly? I'd consider taking a flier on him.
Might have to do with the underdevelopment and one and done movement. Pop doesn't like these hyped high school guys but I suspect he would have drafted Noel or Davis.
How would Aaron Gordon fair if you list him with the wing prospects?

I'm sure there are teams who like him as a 3 rather than a 4. I personally think he'll be able to guard both but his physical attributes probably place him more in the wing category.
Dude, you have Kyle Anderson as a 4 and Spencer Dinwiddie as a 3. You've got to switch away from the BMI-position calc.
There was a point in the college season looking at the numbers that Anderson aka Slo-mo was looking like a Magic Johnson clone. He is a guard to me.
Awesome works as always.
I screwed the data up with Kyle Anderson and Dinwiddie. Both are recommended draft prospects with Anderson being top 5 and Dinwiddie being a second round steal. Everything is fixed now.
Ok, I don't get why some prospects dropped from the reccomended lists.
For example, in this post, TJ Bray ranks at 19th in the NCAA Clean Model. With one of two models liking him, a averaged projected WS/48 of 0.091, a 99.8 NCAA Net Rating. His position is assessed as 2. One model doesn't like him - that's one red flag - so he shows as a tier 2 prospect. OK.
He doesn't show up in the Draft Scorecard for 2014 because DX thinks he won't get drafted.
But how does he not end up in your reccomendations for wings? His project Ws/48 (0.091) is 4th in the table, and he grades as a tier 2 prospect and you've included some players rated as 6th tier prospects. Doesn't this make him potentially a pretty big late 2nd round steal, even better than Dinwiddie, and his projected .072 WS/48?
Can you explain why you use Simple Wins? Simple Wins make Jabari's net rating significantly better than it should be. Also, is there any reason that model 1, which I believe is different in that it treats age and height as quantitative, also has a higher cutoff WP48. It seems like changing two (or more?) wholly unrelated things between the models is unwise.

How are you treating aging curves? I think that you ought to calculate league-wide probabilities that a player in year 5, 6, 7, etc. is on the team that drafted him, and try to model wins over the expected time with the team. After all, a team shouldn't be penalized for taking a 19 year-old when they have a 60% (random guess) chance of still having him at his peak. Also, an old Baseball Prospectus article shows that 17.5 year olds did significantly better than 18-18.5 year olds in the MLB. You might be able to improve the model with more accurate ages, because the aging curve is very steep in the teens.

Have you ever considered adding the team effect feature to NBA WP? I remember an old article from your blog (found it: that said that Kevin Love stood to lose .060 WP48 moving from a historically bad team to a great team.
I remember last year being the best draft. Did "Cindy" change the order of things?

Sign in to write a comment.