Authors: Benjamin Barrish,  Benjamin Pilnick, Ryan Shanker

benjaminbarrish2027@u.northwestern.edu, benjaminpilnick2026@u.northwestern.edu, ryanshanker2024@u.northwestern.edu

Introduction

The NBA draft is an exciting event where players from around the world come together and share a once-in-a-lifetime opportunity. The draft consists of two rounds, 30 picks each, split between lottery picks (1-14) and the rest of the draft (15+). Lottery picks are one of the best ways to change a franchise. For one to get such a pick, there is a lottery system where the NBA chooses your team name out of the bowl–where the probability of winning is determined by a team’s record in the previous regular season. 

If the “worst” teams are getting the first pick, does this mean we expect their performance (Value Over Replacement Player) to be greater than later picks? Quantifying or comparing players from different time periods is difficult, but VORP hopes to remove bias on the era and help determine which draft picks are truly the best. This analysis will explore the values of different picks in the first round of the NBA draft dating back to 1983. 

Value Over Replacement (VORP) will be used, and it is defined by sports reference as “a box score estimates of the points per 100 TEAM possessions–or per game–that a player contributed above a replacement-level (-2.0) player, translated to an average team and prorated to an 82-game season”. This definition may seem confusing, but we will attempt to explain it. In the first part of the definition, box score estimates or commonly known as Box score plus minus (BPM), is a prediction based upon the stats of the game or season that tells how many points you added to the team when you were on the court versus when you were not. It adjusts based upon the typical position stats. For example, centers typically get more rebounds and blocks per game vs. guards, thus the model weighs them more heavily compared to say points. 

In context, say LeBron James has a BPM of +10, this means that in a game, Lebron James is expected to contribute 10 more points, compared to when he’s off the court. BPM is an advanced statistic used by analysts, but it has its limitations. For example, it does not consider the number of minutes played. Thus, a player who doesn’t play as many minutes as another player will have a lower box plus minus. 

VORP removes this limitation by directly calculating team minutes and total minutes played. Take Nikola Jokic. Jokic had a single-season VORP of +9.8, which we can compare to his replacement, substitute DeMarcus “Boogie” Cousins. For simplicity, assume that Cousins has a VORP of -2, which is the average BPM of a bench player. Comparing Jokic and Cousins we see that per 100 possessions or per game, Jokic added 9.8 points more compared to Boogie Cousins. This would mean that the Denver Nuggets are significantly worse when Jokic is subbed off. Note that VORP accumulates as you play more seasons. 

Data

The data in this article was scraped from Basketball Reference using Python and BeautifulSoup. Career counting statistics, as well as some advanced statistics, were collected for each player drafted in the first 30 picks since 1983. The counting statistics collected were points per game, rebounds per game, assists per game, steals per game, turnovers per game, games played, minutes played, and effective field goal percentage (a statistic that adjusts field goal percentage to account for threes being worth more points). The advanced statistics collected were BPM, VORP, PER, win shares, and win shares per 48 minutes. PER stands for player efficiency rating, which measures a player’s per-minute performance

adjusted to team pace. A league-average PER is 15.00. Win shares is a statistic that divides up a team’s number of wins in a given season to the players based on those players’ impact on winning. The win shares per 48 minutes statistic helps allow for win share comparisons between players who played a disproportionate amount of games or minutes. BPM and VORP are both explained above. 

Analysis

First and foremost, VORP is a seasonal stat and accumulates over time. Thus, players with longer careers will have higher values and BPM per 100 possessions. However, the point of this analysis is to not simply compare players but look at picks and reward players for their longevity. In this analysis we looked at all first-round picks and then lottery picks (1-14). After calculating the averages of a pick’s VORP, we graphed the corresponding pick and their VORP. The graph below–comparing VORP and Pick number–shows a negative correlation between pick number and VORP with a correlation coefficient of -0.79. Later draft picks add less to their team per 100 possessions than say earlier draft picks. A negative correlation does not prove that any such pick is statistically different from the other, thus we analyzed individual lottery picks. When looking at the graph for lottery picks, we observe a similar pattern–decreasing VORP. 

There’s usually a bias to the 1st overall pick. We expect them to put their team into contention, or at least improve the team. But is this statistically true? To prove that 1st overall picks are indeed better than later picks, we compared the average VORP for number 1 overall picks to all 30 picks. We used a 1-sample t-test to determine if there was a significant difference in average VOPR for all picks in round 1. In a similar fashion we compared the means between 1st overall picks and lottery picks. The output of both tests is listed below. Since the p-values for both tests–0.002206 and 0.0001972– is less than 0.05–the significance threshold– there exists a difference. Even better, 1st overall picks are significantly greater than all other picks. And similarly, pick 1 has a significantly larger VORP compared to lottery picks. Thus, we can conclude that on average 1st overall picks add more points than their average replacement player per game. We cannot determine exactly by how much, simply because VORP accumulates over time. 

If 1st overall picks are always better, does that mean picking any player earlier in the draft results in better value in the long run? To investigate this question, we used a similar analysis–a two sample t-test. Samples of all 7th, 8th, 14th picks were obtained. We analyzed our null hypothesis: there is no difference between 7th and 8th picks. Note that we used a 0.05 threshold for significance testing, the same as a 95% confidence interval. According to the figure below, the p-value obtained for 7th and 8th picks was 0.1481, which means that there is no significant difference in the added value over replacement players for 7th and 8th overall picks. Similarly, 8th/14th picks had an even higher p-value of 0.8381. By the transitive property and significance test threshold mark, we can conclude that picks 7th/8th and 14th don’t differ at all. Note: Similar results were obtained for PPG. 

Case Study: 2018 NBA Draft: Luka Doncic and Trae Young Trade 

In the 2018 NBA draft, the Dallas Mavericks and Atlanta Hawks swapped picks. Atlanta selected Luka Doncic with the third overall pick and immediately sent him to Dallas for the fifth overall pick and a protected 2019 first-round pick. Atlanta used the 5th overall pick to select Trae Young, as well as using the 2019 first-round pick to select Cam Reddish. Who won the trade? Was it worth it for Dallas to trade up and draft Luka? 

We can analyze whether it was worth it to trade for Luka based on VORP. Comparing different players VORP is difficult since the number of minutes played differ. However, in this situation we are able to compare because of similar volume of game action – Luka has played 371 games and 12861 minutes, while Trae has played 397 games with 13551 minutes played. 

Since the 2019 season Luka and Trae have had similar numbers in points per game, but their values differ significantly. According to the data, Luka has an overall VORP of 30.9, while Trae Young has an overall VORP of 17.5. In fact, Trae Young has never surpassed Luka in VORP for a single season. In addition, the Hawks used their first-round pick to draft Cam Reddish, and he has a negative VORP. Overall, it was definitely worth it for the Mavericks to trade up and draft Luka Doncic instead of waiting to select a player with the fifth overall pick. 

2017 NBA Draft: Jayson Tatum and Markelle Fultz Trade 

However, this is not to say that all trades up in the draft will result in positive outcomes. In the 2017 NBA draft, the Boston Celtics traded away the first pick in the draft (which they received from the Nets four years prior) to the Philadelphia 76ers in exchange for the third pick in the same draft, as well as a future first-round pick. The 76ers used the pick to select Markelle Fultz, while the Celtics selected Jayson Tatum with the third pick in 2017 and Romeo Langford with the future first-rounder (2019, 14th overall. 

Due to Fultz’s injury issues that have plagued him throughout his career, he has played roughly a third of Tatum’s career minutes, who has never missed more than 20 games in a season. 

With that said, almost all of Tatum’s advanced stats that do not reward higher minutes played exceed Fultz’s. Tatum’s career BPM is an excellent 3.5 (ranked 48th all time), while Fultz’s career BPM is a fairly average -0.9. Tatum has also doubled Fultz in win shares per 48 minutes (0.15 to 0.071), while holding a comfortable lead over Fultz in PER (19.9 to 14.8). Even disregarding the fact that Tatum is far more dependable and available on the court than Fultz, there is no debate as to who has the greater impact when on the court. 

It is clear that, even despite trading down and Langford underperforming in the NBA (-3.4 career BPM and 0.043 win shares per 48 minutes), the Celtics won this trade by a comfortable margin. Although the first pick is a valuable asset that usually translates to a player that is significantly better than his peers in the same draft class, the responsibility is still on the organization to select that player, as no player is guaranteed to find success in the NBA. 

Conclusion

If NBA General Managers and owners desperately need to blow up their team and rebuild, coaches will face a moral dilemma of whether they want to tank or keep winning. In a perfect world where fans don’t influence the organization, our guess would be tanking, which will lead to a higher probability of obtaining the first  overall pick, and a greater impact in the long run. However, trading up in the draft will sometimes work. But at the end of the day, it’s the luck of the draw.