A Statistical Analysis of the NBA Draft
Breaking down 30+ years of the NBA’s top-14 draft selections
The long-awaited NBA Draft is now here. As the NBA Draft commences tonight in the Barclay’s Center, many of the League’s worst franchises will entrust the future success of their billion dollar organizations to young, eager teenage prospects.
For players, coaches and management alike, the Draft is a perfect opportunity to re-write history. Given the proper scouting, this spectacle gives every city the ability to target their teams needs and build a strong future team core. For some, tonight’s Draft will provide a much-needed fresh start and hopefully continued success for years to come. For others who get tonight wrong, they will likely be sitting back at the bottom once again next year in June.
On May 14th, the NBA held its annual Draft Lottery, where the top-14 draft picks were assigned to the League’s non-playoff teams (excluding trades). Each team’s chance at winning the Draft Lottery is associated with their regular season W-L record. In other words, the worse you play, the higher your chances are at getting the top pick. With luck on their side, the New Orleans Pelicans won this year’s lottery, despite posting the 7th-worst record throughout the regular season.
However, tonight’s drama will not start when New Orleans picks. Unlike in years past, this year’s draft contains a clear best player in Duke University’s dunking phenom: Zion Williamson. In fact, I would go as far as to say that every credible ‘Mock Draft’ since March would have Zion getting picked at number one. Because Zion is such a lock tonight, the draft chaos will begin at number two, when the Memphis Grizzlies will be on the clock.
For months, Grizzlies scouts have been watching film and hosting work-outs to attempt to gauge their prospects’ athleticism, character, potential, IQ, etc., to determine who they believe to be the (2nd) best player in the draft class. They are projected to either select Ja Morant from Murray State or RJ Barrett from Duke University. However, scouting can only get them so far; it is impossible to truly know if Ja will be a better player than RJ ten years from now. This begs the question, “Can we measure the value of a lottery pick, regardless of who is selected?”
This article uses a statistical approach to accurately calculate the value of each draft pick in the top-14. The statistic primarily examined was ‘Win-Share-per-48-mins’ (WS48), an advanced statistic which measures the proportion of wins added by a player to his team — scaled out to a 48-minute game. This statistic is considered a catch-all statistic because of its ability to incorporate “player, team, and league-wide statistics” in its formula.
The data in the table below displays the average WS48 per player, grouped into rows representing each draft pick in chronological order. The data used in this statistical analysis is NBA Draft data from Basketball Reference, from 1985 to the present.
Evidently, on average, 1st overall picks have the highest win share, relative to other lottery picks. Yet, how are we supposed to understand the magnitude in which having one pick is better than another?
Taking this analysis one step further, the data from the table above was translated into a scatter plot to further display the strength of each draft pick in relation to the others, and is illustrated below. This plot has ‘Pick’ on the x-axis and ‘Win Share’ on the y-axis. A dashed regression line is present as well, which displays the general downward trend as you move down in the draft. The regression formula is displayed underneath the scatterplot.
In addition, the WS48 data calculated in the first table serves as the expected value for all players, based on pick. For instance, Karl Malone’s career WS48 value was .205. Malone, the 13th overall pick in the 1985 NBA Draft, exceeded the WS48 of the average 13th overall pick, .0799, by roughly .1251. As indicated in the chart below, Karl Malone was the second most valuable pick among all players in the data.
The table below shows the top-25 most valuable NBA picks, and their aggregate game statistics from 1985 to 2015. Players drafted from 2015 until the present are excluded from the table in order to ensure that each player has played at least four seasons in the NBA, guaranteeing a large enough sample size of games played.
Similarly, the top-25 least valuable NBA picks are shown in the table below. Players drafted from 2015 until the present are excluded from the table in order to ensure a large enough sample size of games played, like above. However, if players are shown to have played very few games, they are still included in the dataset because lack of play negatively affects teams as well. For example, Len Bias was picked 2nd overall in the 1986 NBA Draft but died of cardiac arrhythmia two days after his selection. His career WS48 = 0 because he never played in an NBA game. Therefore, when compared to the average WS48 of 2nd overall picks, .0983, his ‘Player Value’ equals -.0983, representing his lack of value to his team.
Historically, certain teams have a knack for drafting well, and others for drafting poorly. The diverging bar graph shown below indicates the level of drafting success for each team in the NBA since 1985. The vertical axis is the three-letter abbreviation for each NBA team. There are 37 teams on the vertical axis in order to include teams that have moved cities. The horizontal axis represents the ‘Difference’ between the sum of WS48 values for each of their picks minus the sum of expected WS48 values for the average player at each pick.
Seeing Cleveland below the halfway threshold surprised me a bit, considering that they drafted Lebron James, the eighth best player all time in terms of Player Value, according to the Most Valuable Picks Table, and one of the league’s best current point guards in Kyrie Irving. But it actually makes sense because the team also had a stretch lasting a few seasons where they made a pick in the top-5, indicating that the team was not performing well, so their players WS48 would be low as a result. They also made one of the worst number one picks of all time in Anthony Bennett during that stretch, who’s WS48 ranks 6th-worst of all time at -.1187, thereby eclipsing Lebron’s .1033.
I also noticed that a team’s average WS48 is not a direct measurement of their success in the NBA. Take the Boston Celtics for example, who are one of the lowest teams on the graph but have won five Eastern Conference championships and two NBA championships since 1985. Hopefully this graph can provide reassurance for those fans that believe getting the draft right is the only way to win a championship.
Yet, for those who believe in the draft, even with all the scouting in the world, there is no way to predict how a player’s career will pan out. Any player can outperform their expectations on any given day, leaving fans on the edge of their seats, uncertain of their team’s fate. That is the beauty of the NBA.
