PMVP NBA Draft Guide
Updated June 12, 2018
You can find the RShiny Interactive Draft Guide here.
Projecting precisely how well a 19 year old college basketball player will perform in the NBA is about like projecting the next stock market bubble or whether someone will contract an infectious disease. While not impossible, its quite improbable. But, it is not to say that a rigorous projection model cannot be useful.
As statistician George Box famously once said, “All models are wrong, some are useful”. We don’t embark on building models, like an NBA Draft model, in hopes of perfection, but in hopes to limit bias, make sounder decisions, and add to the discussion of both fans and teams.
I have written a few times in the past few months about my Player Market Value Projection (PMVP) model for the June 21st draft. I have now equipped the model with an RShiny web application to allow everyone to engage with the model and compare players statistics and their forecasts to be found here. Please spend some time playing with the tool and dive deeper below as I refresh the logic of the model and provide the landscape of this year’s draft.
PMVP is intentionally unique in two-ways. First, many models differ in what they are predicting. Most are either using a dependent-variable that is either a kitchen sink statistic or a plus-minus-like statistics. There is nothing inherently wrong with these, but they can at times lack information or miss out on particular parts of the game. Thus, PMVP projects a player’s 5th year contract value as a percentage of that year’s salary cap (adjusting for inflation). Why 5th year? Because this is generally the first year of a player’s second contract which is more representative of their market value and, therefore, what NBA General Managers think of the player.
Secondly, many models simply give you a one-number summary of a player. Many times this is the easiest way to convey information about a player to a coach or scout, but it severely misses the point. NBA teams do not all have the same levels of risk, just ask Danny Ainge. Thus, it is important to know all the possible outcomes a player’s career could take. What is the likelihood they are a bust? What are the chances they are an all-star? Questions like these are just as important as determining the average of a player’s career, because you might be more willing to take a “safer” pick even if they are projected to perform just slightly worse.
To solve the variability issue, as well as for a laundry list of other reasons, I chose to model the NBA Draft using a Bayesian Additive Regression Tree (BART) model. I found this model performed significantly better than both linear regression (no surprise!) and random forest models. This model builds a prior distribution for each player and then additively updates the distribution until we can create a posterior distribution of the player’s PMVP — their 5th year contract value.
The RShiny App is a great tool for visualizing these distributions for each player and comparing (based off 10,000 simulations of a player’s career). Unfortunately, I have currently only built the model for collegiate players. Continue to check back in on this page and the application in the near future for further updates and modeling. Now, lets dive into both my rankings and the model.
1. Luka Doncic
PG — Real Madrid (Slovenia)
Doncic has been heralded as a prodigy having led Real Madrid to the EuroLeague championship as the EuroLeague MVP at just 19 years old. Doncic has shown elite skill in shot creation and passing having played far more games against high level competition in the past 2 years than any other player in the draft.
The only real questions about Doncic are his athleticism. He has the skill and IQ to run an NBA offense right away. Doncic is as much of a sure-thing as one could hope for, but may have a lower ceiling than some of the other players at the top of the draft.
It is easy for NBA decision makers to pick holes in the game of someone they have seen for many years either in a college or European setting, but we should know that with time we can find similar holes in the games of players up and down the draft board. There is real value in seeing a player grow on a large stage over the course of a several dozen game more schedule than one-and-done prospects. Thus, Doncic in my mind has been in another tier from the rest of the class since last spring and has only separated himself since.
2. Deandre Ayton
It appears the Phoenix Suns might just take Ayton at №1. He is the best rebounder in the draft by far and has the ability to become an elite offensive talent with a 65.8% true shooting percentage. Ayton’s physicality and size, matched with good speed and touch give him the potential to eventually become a dynamic offensive option. We can see below that Ayton’s posterior distribution is pretty well-defined — a low-risk pick at the top of the draft.
Yet, the looming questions for Ayton will be his motor and defensive ability. Though one would think he has all the tools to be a stellar rim protector he didn’t impress on that end of the court. If he is able to learn to set better screens and defend on switches — no small task — he could easily become one of the unicorns of today’s game. While Doncic can likely fit in any backcourt from day one, Ayton is in need of a coaching staff willing to help him make the leap to become the backline of a modern NBA offense.
3. Marvin Bagley III
The questions about Bagley are his defense. Bagley dominated the college game with his athleticism, but will he remain an elite athlete against NBA big men?
Bagley should easily be able to average 15 points and 10 rebounds a night for a bad team, but it is a matter of whether he can become more than athletic, offensive big that remains the question. Bagley is a top the models ranking and as the highest floor due to his efficiency and pure talent on one of the nation’s best college teams. Bagley will be challenged at the next level. How he responds could easily make him a multi-time all-star or a player will all scratch our heads over.
4. Mohamed Bamba
A fairly mobile big man with a 7'9" wingspan who can potentially shoot threes at a high clip. That is ideal for the modern NBA. But, whether those viral videos of Bamba making threes will carry over is yet to be seen. Bamba will be picked purely on potential. His offensive load was remarkably limited in his lone season at Texas, which is why the model has him as just the 17th best collegiate player.
A player who is incredibly similar to Joel Embiid and Rudy Gobert as a prospect, if Bamba grows offensively he could become another unicorn. But, if Bamba fails to adjust to the NBA, learn to be an offensive factor, and translate as an elite rim protector he could be the bust of the 2018 draft.
5. Wendell Carter Jr.
At times Carter was overshadowed by the other stars at Duke, including Marvin Bagley III. But Carter is one of the most NBA-ready prospects in this years draft with his ability to rebound, score around the basket, and make plays out of the post. His ability to guard quicker players in today’s NBA will be the biggest question on his future. But, even his floor is remarkably high. Carter likely won’t be drafted in the top-5, but we could look back in three years and see Carter as the one of the best of the glut of bigs at the top of the draft.
6. Trae Young
Trae Young quickly became the darling of college basketball with his long range shooting, but fell from grace as defenses seemed to find ways to wear him out late in the season. Young has become known for his pull-up shooting — a much desired quality for NBA guards in today’s game — but, his skills as a playmaker in the pick and roll, creating for himself and others at three levels, gives him the highest ceiling in the PMVP model.
Many have criticized his diminutive size, defensive liabilities, and turnover problems. While these are all serious concerns that cause me to move him away from the top-2 ranking given by the model to number 6 on my board, we should not lose sight of what fascinated scouts earlier in the year. He has an offensive craftiness tailor made for the modern game. On the ball play making is something that projects remarkably well into the pros according to the PMVP, but I think his off the ball abilities, though in a limited sample size are a high commodity.
Anyone who watched this year’s NBA playoffs saw the effect of gravity — how a player’s shooting distorts a defense even when they don’t have the ball. Young could potentially be this player if a team can make do with his liabilities. Pairing Young with a a bigger, more defensive minded backcourt mate will give him the best opportunity to succeed. Young is surely a high variance pick, but one who could just be a gamble worth taking in the mid to late lottery.
7. Jaren Jackson Jr.
A blank slate.
That is the easiest way to describe Jaren Jackson Jr. He played the least amount of minutes of any of the collegiate players in the top of the draft, having to share minutes with big-man Nick Ward. Jackson, as one of the youngest players in the draft, is an good athlete with elite defensive abilities and a solid perimeter shooter. However, he could easily get in foul trouble — a big negative for big men in the model — and had many games where he completely disappeared: scoring just 2 points in Michigan State’s NCAA tournament loss, for example.
But, if Jackson gets the proper polish in the NBA he could become an elite modern NBA center. He has one of the widest range of outcomes in his distribution, showing why he has been one of the most highly debated prospects.
8. Kevin Knox — Kentucky
9. Miles Bridges — Michigan State
10. Michael Porter Jr. — Missouri
11. Mikal Bridges — Villanova
12. Shai Gilgeous-Alexander — Kentucky
13. Troy Brown — Oregon
14. Collin Sexton — Alabama
15. Lonnie Walker IV — Miami
For the rest of my big board and more on the model check out the RShiny App here.
Continue to check back in the days leading up to the draft as this page is expanded with more and more content. And feel free to reach out to me with questions or comments at aepowell95@gmail.com.
