NBA Draft 2021: An Analysis.
Winners, losers, and a machine learning analysis of how to draft well.
We are a little more than a month away from the start of the NBA season and coincidentally just as many days since the NBA draft. In this while, I have had the chance to look at the drafted players and data relating to their performances prior to the draft and I think I have a definitive answer to who won the 2021 NBA draft.
[Spoiler: It was the Houston Rockets.]
Before we get into the fun (read: controversial) stuff, I think we should be on the same page about what makes a good draft prospect.. and long answer short? It depends.
See, drafting is a very complex process and drafting well is for lack of better words, a real hassle. For starters, it depends on what kind of role you want the player to play? Do you want them to be a compliment to your star players? Do you want to draft a star to build around? On ball threat? Off ball threat who provides defensive coverage? It’s difficult to evaluate draft prospects because it’s difficult to come up with an evaluation metric that makes sense.
Fortunately, for the purposes of this analysis I have taken a generalist approach and thus will provide an objective and unquestionable report of who the best prospects to draft are. SIKE. I want to reiterate that this is a task that 30 multi-million dollar franchises fail at (for the most part) every year. With that out of the way, let me introduce to you the evaluation metric for our analysis: VORP, or Value Over Replacement (Level) Player.
As the name suggests, VORP tries to evaluate a player's value to their team above a replacement level player ( which can be looked at as an undrafted/G-League level player). It has been defined by Basketball Reference as a box score estimate of the points per 100 team possessions that a player contributed above a replacement-level (-2.0) player, translated to an average team and prorated to an 82-game season. Further details about the stat can be found at this page.
We’ve covered the what, but let’s talk about why VORP. The simple answer is that it does justice to the role players. While most other metrics provide immense value to box score statistics and the “eye test” will make the flashiest players stand out most, VORP looks at, and appreciates, the players that affect winning.
Let’s take Mikal Bridges for example, on the surface his statline of 13.5 pts, 4.1 reb, 2.3 ast, 1.1 stl, 0.9 blk seems pretty average, but ask any Suns fan and they will tell you that he was instrumental in the team finishing with the 2nd best record and making an impressive finals run. What about his VORP? His VORP of 3.0 was ranked 21st in the league, ahead of multiple all stars like Paul George, Donovan Mitchell, Jaylen Brown, etc. While on this topic, any guesses on who led the league in VORP? It was the MVP: Nikola Jokic.
Okay great, so we love VORP, we love everything about it. Now how do you go about predicting what kind of VORP a player will have in the league, when.. you know, they haven’t played in the league yet? I’m glad you asked! We can finally start talking about what makes a great draft pick.
First, we look at the categories the players belong to, in this case: their position, and the level at which they played. As far as level is concerned, we see in the charts above that international players tend to be high-risk high-reward prospects as they have the highest ceiling and the lowest floor, and while information on high school players and G-League prospects are limited, we can also tell that Freshmen tend to perform well in the NBA. This could be due to the fact that freshmen are younger and get more time to get accustomed to the league.
When it comes to positions, we see that centers are a run-away winner. They have the highest ceiling and the highest floor but what if you are a team that already has two great big men? Adding another one would be a waste, especially if you need someone who can cover another position. To combat this, we will now move from a player’s categories to their quantities, i.e, their stats.
In the above figure, each cell represents the correlation coefficient of a given stat, in a given position, with the VORP. To simplify, if the ‘ast’ row of the ‘C’ column reads ‘0.16’, it means that with 1 unit increase in assists, VORP tends to increase by 0.16. When it comes to box score stats, we see that points and assists are actually the worst predictors of a player’s value, instead it is rebounds which tends to play a massive role as a predictor. This could be because it helps estimate the player’s ability to cope with the physical adjustment most players struggle with. Moreover, we see that point guards who can rebound well do especially well, despite that not being part of the regular job description - eg: Luka Doncic, who had the second most rebounds for a PG in the league last season.
In fact, this trend appears to persist across the board - centers that are good at passing and stealing the ball (rather than scoring and blocking) do better in the league. Rebounding and stealing seems to be important for every position, which also makes sense as the game has evolved to a point where the attempt is to make every possession as efficient as possible and these are amazing ways of taking possessions away from your opponent and for yourself. When it comes to advanced stats, the box plus/minus stats and free throws attempted show value in for all positions, while for PGs specifically, we see value in three point shots attempted and the efficiency in which they’re scored.
Finally when looking at which aspects are the most negatively correlated with VORP, we arrive at the pick number they were selected in and their age, which makes sense too. This tells us that in general, teams do a fairly good job of selecting players higher up in the draft and that younger players tend to improve more over the course of their career.
Great. With all of that out of the way, we can finally approach the 2021 NBA Draft!
So, who does my model think the top 3 players in the 2021 NBA draft class are?
Cade Cunningham - Detroit Pistons [Predicted VORP - 2.92].
Now this might come as a surprise to most, with me being one of them but the #1 overall pick is projected to be the 3rd best player from this draft. As discussed earlier, he fits into the Luka Doncic mold of point guards that can score and rebound exceptionally well. However, unlike Doncic, his playmaking can fail him at times, averaging only 3.5 assists per 36 mins, against Doncic’s 6.6. However, there is faith that Cade will outperform his playmaking massively once in the NBA playing alongside players who match his mettle.
Evan Mobley - Cleveland Cavaliers [Predicted VORP - 4.58].
Evan Mobley is one of the most complete big-men to come out of the college system in years, arguably the best rounded center since Joel Embiid. He has shown brilliance in scoring, drawing fouls, being a solid rim protector and even playmaking out of the post. The only drawback on his portfolio seems to be his below average rebounding rate. This gets him the #2 spot on this list.
Finally, the #1 option - considering 2 out of the top 3 picks are up above, along with the fact that I said Houston Rockets are the winners of the draft, I’m sure we all know who reigns supreme. Without further ado, let me introduce to you…
Alperen Şengün - Houston Rockets [Predicted VORP - 6.76]
I apologize for the fake-out, and to be fair Jalen Green was also a very high ranking prospect (#5 with 2.7 VORP) but the 16th overall pick, Sengun is set to take the league by storm* (*if things go right). His profile, seems to fit perfectly with all the discoveries we’ve made so far about the recipe for an ideal draft prospect. He is a center - the best position, an international prospect - the highest ceiling, his one drawback is blocks - the one value that tends to affect VORP least for centers, and his strongest skills are scoring, assisting and stealing the ball, heck, he was also the 4th youngest player in the draft. All signs point to him being a long time star in the league, and given the current young core of the Rockets, it seems like he will have ample opportunity and support to be able to do so.
The full list can be found below:
Finally, the winning and losing teams:
Winners:
Houston Rockets - No surprises here, they got arguably the best player in the draft, they have another top 5 player in Jalen Green and their other 2 draft picks (Joshua Primo and Usman Garuba) all rank in the top top 25 prospects of this draft. They are the clear winners.
Orlando Magic - The Magic drafted only 2 players but both players (Jalen Suggs and Franz Wagner) are 2 of the top 10 projected players in this draft class and fit perfectly with their young core, they come out of this draft looking stronger in their rebuild stance.
Cleveland Cavaliers - Many questions will be asked about the Cavs drafting Mobley and extending Jarett Allen, but they emerged with the clear #2 player in this draft class, who could very well be one of the hybrid big men that can fit in at PF as well.
Losers:
Portland Trailblazers - The Blazers traded away a future second to buy into the draft and did so by drafting one of the lowest ranking players in the class.
Milwaukee Bucks - The Bucks traded away the 31st pick to draft 2 players in the 50s, and neither of these picks are as good as if they had just drafted someone like Queta, McBride or Butler at 31.
Any team that didn’t buy into the second round - There were players with significant upside being dealt in the second round - Juhann Begarin, Sharife Cooper, Jared Butler, BJ Boston are all great prospects that were available in the 40s and 50s.
The full performance by teams is attached below:
As a closing remark, I would like to point out that everything mentioned here is a prediction and not a certainty. Most machine learning models are imperfect and if making these predictions was as easy as I’ve made it out to be, especially on a zero-budget, NBA front office members would be out of jobs by yesterday, and there’s a reason they aren’t. So, here’s to hoping that this paragraph is enough to absolve me of any responsibility for being wrong about my predictions, and to hoping that the Dallas Mavericks win the NBA title this year.