The NBA draft is always a crapshoot at best, as teams attempt to guess which players will go on to be stars, and then hope they'll get lucky enough to find a diamond in the rough.
As a result, tons of poor picks are made every year. So what if each team could re-do a first-round pick and select someone else?
Im not worried about who the team would select instead, just which picks they would want back.
To avoid subjectivity, I'm using a formula that I developed in the past to evaluate draft picks. You can find the full explanation here.
First of all, what exactly can teams count on when they make a draft pick? Well, unless a team releases or trades a player, they are under the team's control after the draft for up to four years, the maximum length of a rookie contract. So really, when looking at draft steals, we should focus almost exclusively on the first four years of a player's career.
As a result, I looked at the first four years of Win Shares data for a player, as provided by basketball-reference.com. Win Shares are an advanced basketball metric calculated so that one Win Share is exactly equal to one win provided by that player to his team's cause. It's the combination of Offensive Win Shares and Defensive Win Shares, a full breakdown of which can be found on this page, called "Calculating Win Shares."
Starting with the year 1990, when the NBA Draft first introduced the current lottery system, I looked at each and every single player drafted into The Association, tracking their draft position and the amount of Win Shares they produced in their first four seasons in the league. It is important to note that I only evaluated data through the 2007 draft because the players taken in 2008-2011 have not yet played out their first four seasons in the league.
After I had data for all 1,028 players drafted from 1990-2007, I took the average number of Four-Year Win Shares for each draft position and plotted them on a scatterplot (which you can see in the embedded picture with draft position along the x-axis and Four-Year Win Shares along the y-axis).
Using a best-fit logistical regression, I found the following formula: Four-Year Win Shares = -5.836* ln (draft position) +24.537.
For the statistically inclined out there, that equation has a coefficient of determination (r^2) of 0.91024. For the non-statistically inclined, the equation fits extremely well.
Using this formula, we can plug in a number for draft position and have the formula show how many Four-Year Win Shares a player drafted there should be expected to produce. For example, the first overall pick of a draft should produce 24.537 Win Shares while the 30th overall pick should produce 4.688.
With that data firmly established, we can tell exactly how much players have exceeded or failed to live up to the expectations associated with the slot in which they were drafted. That can be done by subtracting the expected win shares based on the draft position from the actual number of Four-Year Win Shares that players produced. If the difference is positive, the player exceeded expectations by that much and was a bit of a steal. If the difference is negative, the player failed to live up to the expectations and was a bit of a bust.
Now let's go back to the Monta Ellis example.
Ellis was drafted 40th overall, so he should have been expected to produce 3.08 Four-Year Win Shares. The shooting guard actually produced 13.7 over the first four years of his career, meaning that the Golden State Warriors "stole" 10.69 Four-Year Win Shares when they drafted him. This was still a great pick, there's no denying that. It's just not quite as great as quite a few players drafted ahead of him.
It's important to realize exactly what we're looking at. As some of you may have realized, even No. 1 picks may be considered steals.
The formula has changed slightly because I now have data from 1982-2008, but the methodology still remains valid.
It is also important to note that only players drafted between 2003-2008 were considered eligible here as those drafted between 2009-2012 have not played out the first four seasons of their careers.
The player with the biggest negative difference between expectations and actual production will be selected for each of the 30 teams.
Read on to find out who they were.