NBA Draft: Player upside isn't linear, but geometric according to the data

"Yes, I will give you an autograph if you can name the f(x) of my eyebrow." - Jerry Lai-US PRESSWIRE

ESPN's Bradford Doolittle used Tom Haberstroh's numbers to confirm what's Roland Beech found out 6 years ago. This ish be non-linear, homie.

Okay, what does this mean? Well, in a linear way the absolutely best player in a draft class will be the first. And the absolutely second best be the second, and so forth. That works out great in theory, because we like to think that the team that needs the best player will draft first. And the team that is second most awful will get the second best player, and so forth. But just as there are huge variations in the NBA Draft order so too do we see huge variations in the quality of the draft according to the order players are drafted. We don't need to look much farther than this past draft class where Anthony Bennett went first, and Michael Carter-Williams went ten spots after.

And you add it all up the averages should balance out season to season variation, right? Not so according to Bradford Doolittle of ESPN. (Check out the article here, be aware, it's In$ider.) The data set he uses is from 1989, and the data primarily attributed to ESPN's Tom Haberstroh. Anyway, here's a snippit:

When it comes to draft slots, all things being equal, the higher you pick, the higher the chance you're going to add a productive player. That seems obvious, right? Well, all things aren't equal.

The human element always introduces a certain amount of chaos into the proceedings, and this is the case with the NBA draft. According to the research compiled by my colleague Tom Haberstroh as part of his DRAFT Initiative, the cumulative values produced by each draft slot do not shake out in the smooth manner in which you'd expect.

Since 1989, when the NBA moved to a two-round format, the 25 No. 1 overall picks have produced an average of 8.4 Estimated Wins Added (EWA), easily the most value compiled by any draft slot. With upper-tier Hall of Fame players such as LeBron James and Shaquille O'Neal included in that calculation, this is no surprise. No. 1 picks have also exceeded their expected average EWA by 1.7 wins, the best performance of any slot. This isn't exactly breaking news, but by and large it's a great -- sometimes franchise-making -- thing to pick first, and the typical top pick carries with him at least a modicum of certainty, the 2013 draft aside (when the Cavs selected Anthony Bennett).

After that, however, things are not nearly so straightforward.

Picks No. 2 through No. 5 have produced nearly an equal amount of average value, with No. 3s slightly outperforming No. 2s over the last quarter century. There's a dip from picks No. 6 through No. 8, while No. 9s have done exceedingly well. Teams have especially bombed out when picking sixth.

- Bradford Doolittle (2014), ESPN

Yeah. Tom Haberstroh includes a line graph that shows a) the actual value (in EWA/YR) for each position, and b) a curve that approximates how good or bad things are. I'm not going to show that graph but there are a few things that pop out:

  1. The Number one pick, even with all the bad #1 picks we've had (Andrea Bargnani, etc) is still ahead of the curve over the data set (1989 till now). That's why you tank boys and girls.
  2. The Number two pick is below the curve. We know that, we have Marvin Williams.
  3. The Number three, four and five picks are WAAAY ahead of the curve. In fact, you could argue that the #5 pick is better than the #4 pick. The difference between their actual EWV vs the Expected value of the #5 pick is almost the largest positive difference

Duh, we want a Top 5 pick. Why? Because things go haywire right after the Top 5. The sixth, seventh, and eighth picks are way below the curve of their expected positive impact. There are a few bumps up around #9, #10, #24, #26, #37, #43, #45, #47 (that Paul Millsap, Mo Williams pick), and then again in the late 50s thanks to the San Antonio Spurs scouting department.

This is great info, but it's not new info -- at least not to me. While the data at that tracks the exact same thing is now 6 seasons older and un-updated, the data set is from 1989 to 2008, and covers much of the same ups and downs. Roland Beech breaks it down:

First up is a simple, what should you expect, performance-wise from a given pick number? If you have say the 5th pick in the draft, what are your chances on average of landing a star player?

Since B-R provides career games, and then per game points, rebounds, assists and minutes, I have gone with an admittedly highly simplistic look on things by just going with:

Rating = points/game + rebounds/game + assists/game

Why use this definition? It's the data I have easily on hand, which while not a good player rating system is a decent wag for these purposes. Then I group players as follows:

  • Star -- 20+ rating
  • Solid -- 15 to 19.9
  • Role Player -- 10 to14.9
  • Deep Bench -- 5 to 9.9
  • Complete Bust -- less than 5
  • DNP -- (never played in the NBA)
Keep in mind the stats are career per game averages so lower than the peak performance years of a player. Moreover, there is also some bias in that using recent years some of the current players may well spike up their career 'standing' with more years under the belt.

- Roland Beech (2008),

Yeah, if you look at his breakdown (which breaks it down into player honorifics, and not just numbers) you see the same thing. Mostly, that the #3 is usually better than the #2, and that the walls come crumbling down after #5.

You want a Top 5 pick if you are going to be in the lotto for a reason. And that reason is, for the most part since 1989 till today, that a Top 5 pick is *way* better than any other pick in the draft. Yes, we know that the Utah Jazz got *really really* lucky by drafting Hall of Famers at #16 (John Stockton) and #13 (Karl Malone) in back-to-back drafts. But that will never, ever, happen again. Scouting has advanced since the early 1980s. Don't pretend you are better than the average when you are not.

And don't pretend that the Jazz do not need a Top 5 pick from this draft, otherwise, on average, the Jazz will be left with, on average, vastly inferior expected and recorded value. The upside isn't linear, sure, sometimes a #6 could be better than a #4, but that rarely ever happens. And the expected upside from that number 1 pick is off the charts, it's a franchise changing, geometric production expectation compared to the normative curve for the rest of the draft.

And that's why you tank in this era of the NBA with this era's salary cap, and this era's rookie scale contract.

Log In Sign Up

Log In Sign Up

Please choose a new SB Nation username and password

As part of the new SB Nation launch, prior users will need to choose a permanent username, along with a new password.

Your username will be used to login to SB Nation going forward.

I already have a Vox Media account!

Verify Vox Media account

Please login to your Vox Media account. This account will be linked to your previously existing Eater account.

Please choose a new SB Nation username and password

As part of the new SB Nation launch, prior MT authors will need to choose a new username and password.

Your username will be used to login to SB Nation going forward.

Forgot password?

We'll email you a reset link.

If you signed up using a 3rd party account like Facebook or Twitter, please login with it instead.

Forgot password?

Try another email?

Almost done,

By becoming a registered user, you are also agreeing to our Terms and confirming that you have read our Privacy Policy.

Join SLC Dunk

You must be a member of SLC Dunk to participate.

We have our own Community Guidelines at SLC Dunk. You should read them.

Join SLC Dunk

You must be a member of SLC Dunk to participate.

We have our own Community Guidelines at SLC Dunk. You should read them.




Choose an available username to complete sign up.

In order to provide our users with a better overall experience, we ask for more information from Facebook when using it to login so that we can learn more about our audience and provide you with the best possible experience. We do not store specific user data and the sharing of it is not required to login with Facebook.