clock menu more-arrow no yes

Filed under:

ESPN's Draft Projection Model highlights risk/reward of NBA Draft

New, comments

Perfect examples of statistical warriors are Washington's Marquese Chriss and Dejounte Murray, who have gone from iffy freshman seasons into potential draft night gold.

Ron Chenoy-USA TODAY Sports

The NBA Draft is coming up in less than two weeks, and the silly season is fully underway throughout the basketball bouncing country.

While the NBA Finals holds some intrigue for fans, it's the upcoming draft that holds all the intrigue for front offices whose entire careers are on the line. Picking the gem among the dirty rocks is often a product of luck.

For the Phoenix Suns, coming off a 23-59 season and holding four of the top 34 draft picks, having some good luck on draft night is likely the difference between keeping and losing their jobs over the next 1-2 years.

The old saying is that good luck is the product of opportunity and preparation. Every GM gets the opportunity on draft night, so the big variable is preparation.

And preparation can take a number of forms: film watching, individual workouts, group workouts, interviews and statistical modeling, to name a few.

Lately, we have seen a previously unknown player (Washington's Marquese Chriss) rise up the draft boards from late lottery to potential #3 overall pick, depending on the expert you believe. Center prospect Cheick Diallo hardly saw the floor at Kansas, but could rise all the way into the lottery by draft night.

By the same token, some players rated highly all year are starting to drop ever so slightly down the mock drafts. California small forward Jaylen Brown, Utah's Jakob Poeltl and Marquette's Henry Ellenson have all dipped to some degree. Brown is still top-10, but both Poeltl and Ellenson appear to be destined for late lottery selections.

Yet none of these players has appeared in a competitive game of five-and-five in at least two months.

That's the hallmark of silly season, otherwise dubbed as the pre-draft process. Since the college season ended, NBA front offices have been relegated to film and scripted workouts for every player in the draft pool. When that happens, measurements and athletic ability begin to take on more importance than basketball IQ in some circles.

This is also the time that number-crunchers go crazy, somehow devising systems that project college stats into the NBA world to give us a chances of success vs. failure for each prospect.

On the one hand, I find it very interesting to hear that player A's resume stacks up against these other players with similar resumes, and while he has a 10% of becoming an All-Star, he also has a 40% of busting out. On the other hand, every player is different. So the projection models can only go so far, and have a wide error rate.

But on the whole, you get a feel for which players - based on their college experience and their physical and athletic profile - are "safer" than others.

According to one projection model, 7'1" center Jakob Poeltl is one of the safest picks on the board. He's most likely to have a long, steady career, with only a small chance of All-Star status (12%) and a small chance of busting out completely (16%). That's because productive, young seven-footers in college generally do the same, to a lesser degree, in the NBA.

On the other hand, 6'10" forward Ben Simmons - the projected #1 overall pick - has the widest range of All-Star potential (25%) versus embarrassment (35%) of anyone in the draft pool. According to this model, he's even scarier than Marquese Chriss (21% and 31%, respectively).

ESPN admits openly that the projection model is imperfect. They show the last five year's worth of top prospects - according to the modeling system - and admit that only half of them reached their potential. But then again, when you say there's a 30% a player becomes an All-Star and he doesn't, aren't you still right?

This year, the model shows some very interesting outliers. An important note: the system projects these guys in NBA uniforms through each year of their career, but the results ARE ONLY YEARS 2-5, where they are most likely under the drafting team's control.

...predicting a player's statistical plus-minus (SPM) in years two through five in the league, which is the time frame players are under team control for below-market prices without unnecessarily penalizing them for outlier rookie seasons.

The model uses box score stats (adjusted for level of competition faced and pace), scouts' rankings and player information -- such as age, height, weight and position from 2001 to 2011 -- to project out future classes. The resulting output is twofold: a player's draft grade, which is based on his total projected SPM, and his percentage chance to play at the level of an All-Star, starter, bench player and bust in his first five seasons.

The players with the highest draft grades have the most expected SPM during years two through five, but they might not necessarily have the highest ceiling. In comparison, the buckets of All-Star, starter, bench player and bust give a clearer picture of the tradeoff between risk and reward for each prospect.

First of all, Ben Simmons is only the 4th best prospect, according to this. Jakob Poeltl and Kriss Dunn come in ahead of him, along with Brandon Ingram.

Cheick Diallo, ranked in the bottom half of mock drafts, projects to be better statistically than Jaylen Brown, who commonly shows up in the Top 5 of those same mocks.

You might also be surprised to see Dejounte Murray - a second round projection - show up at 20 on the list, ahead of fellow point guard prospects Tyler Ulis (30) and Demetrius Jackson (36) and Wade Baldwin (41). The teammate of Marquese Chriss at Washington statistically should be drafted ahead of three players ranked 15-20 on most big boards. Yet, scouts and executives likely don't do any such thing.

The top unknown ranking is Chinanu Onuaku, coming at #13 overall in terms of projected impact in years 2-5 in the league. We all knew that, right? Onuaku is 38th on DX's Big Board.

How is this kind of projection possible? Most of these guys barely played a year of college, and some of them barely played at all. Diallo and Chriss, for example, are wading in a very, very shallow stat pool.

I really can't tell you. All that is proprietary information not shared, but likely is based on mounds and mounds of data, using statistics from prior draft classes and trying to "profile" this year's prospects against those players. So it's got some hint of reality to it, but a whole lot of guesswork wrapped into this scary projection that is more likely than not to be wrong.

Modeling like this only adds one more kernel of corn to the popcorn bag of information GMs devour all spring leading up the naming of a single player at a time on draft night.

It's not like this information is important, right? I mean, a GM's job is only on the line and all.