As we get closer and closer to the draft, you’ll start to hear more about players who “model” well, or players “graded out” well…things like that. Most people don’t know exactly that that means. So let’s take a deeper look at what goes into a draft model for a team. (Keep in mind that every team has their own proprietary models/formulas/thoughts…so this may not accurately reflect every team)
First off, why do we use these models? Great question, the answer is simple. Accurately predicting someone to be successful at a job when they have no actual experience in that job field is not easy. Yes, we know that all of these future draft picks have played basketball before. Many of them for years and years at a high level. But no player in this year’s draft has played an NBA game yet. So, to help the selection process, teams put together complex formulas based on historical data to help project the level of impact each player in the draft will have. Like most things in life, this is not fool proof, but it certainly helps. Daryl Morey has been the subject of many articles about draft models. He has been known to go in depth about certain factors and how many iterations his group needed until they got something they were comfortable with.
Long story short, teams look at successful NBA players. Then, if those players went to college, they take all of the information they can find about those players and do a bunch of math things to them to see what categories tend to translate to success. As the model gains more and more information about all players, it can separate the players who are most likely to be successful in the NBA from those who are great college players but may struggle when they get to the next level.
It’s important to note, that while this is a scientific way of drafting, science can take longer than most fans are willing to give. Memphis saw this in 2014 as their draft model scores had Jordan Adams extremely high and the Grizzlies selected him at #22 in the 1st round. Thanks, John Hollinger! For those keeping track at home, Adams played in 32 NBA games, total. He did manage to score 101 points in those 32 games though. Other picks from Hollinger’s draft model include: Jamaal Franklin, Janis Timma, Jarell Martin (HOME RUN), and Wade Baldwin. Not exactly the hit rate they were hoping for…Don’t worry Grizz fans, the other Shooting Guards selected right after Jordan Adams won’t make you puke. Rodney Hood (23), Bogdan Bogdanovic (27), and Joe Harris (33) were all taken after Adams. So were Shabazz Napier and Clint Capela…yikes.
The model will learn more and more as each season passes, but it’s up to the human element to decide how all of the information gets weighed. There are a bunch of different categories that go into each player’s draft score, including: Scoring, Shooting, Rebounding, Defending, Efficiency, Testing…there are certainly more, but like we said, each team is different.
And the inputs are weighted differently for each position as well. Shooting metrics for guards may be more important than it is for bigs…but that may be changing as we speak! As we’ve talked about before, there are certain statistics that have been teased out to find they are a better indicator for NBA success than the logical explanation. NCAA 3PT% does not predict NBA 3PT% as accurately as NCAA FT% does. This seems counter-intuitive, but the numbers aren’t wrong. And as we’ve highlighted in other articles, it’s the reason why some big guys in the NBA have been able to easily transition to shooting from distance.
A word to the wise for those who are trying to create their own model in order to impress the world. If Alec Peters, or Alan Williams end up as the top ranked player in your model, something is off. If Kevin Durant or James Harden does, you’re probably headed in the right direction.
So why do teams have handfuls of employees to work on these things if they can lead you down the wrong path? That answer, too, is simple. It’s a way of creating another scouting voice. One that may pick up something that the guys on the road didn’t see. It catches every game and every stat in those games. Which is something that no scout or personnel executive can do. The key is to find the balance. If you lean too heavily to one side, you can fall into traps. Jordan Adams was a statistical monster, but had a bad body and didn’t make it into the NBA because of that. But there is a chance that Ian Clark gets overlooked because he went to Belmont…and a draft model may have scored him well. And because of that, the decision makers had to take a look at him to see if he was for real.
As a leader, it’s important to know what your limitations, and the boundaries of your team’s talents. Using analytical models can enhance a scouting groups abilities, but like most things, shouldn’t be left up to just the machines. It’s also important to throw a caution flag at the other end of the spectrum, teams that don’t look at historical data. If you aren’t learning from everyone’s mistakes, it’s difficult to improve. “History doesn’t repeat itself, but it often rhymes.” -Mark Twain
Knowing where your blind spot is, and trying to shine some light on it is a good thing. It’s a way to get better. And this is why the best teams in all sectors of business (not just the NBA) analyze as much information as they usefully can.