By Ryan Tabibian '25
Introduction
Before the 2000s, NBA teams drafted players with the expectation that they would offer significant contributions immediately. It was only in the early 2000s when phrases such as ‘he’s raw’ and ‘needs developmental time’ became ubiquitous in pre-draft analysis. These ideas were emphasized by the formal creation of the NBA’s first summer league in 2002. This developmental process is perhaps most prevalent in the current construction of the Golden State Warriors. Along with the Warriors’s current core of Steph Curry, Draymond Green, and Klay Thompson, they have young players such as James Wiseman, Jonathan Kuminga, and Moses Moody, who are seen as longer-term projects in need of ‘developmental minutes’ according to many pundits.
Recently, Warriors head coach Steve Kerr admitted that his approach to Wiseman’s minutes may have interfered with his development last year and said that he "thinks it's important not to automatically assume that minutes equal development...development also includes observation." However, some NBA head coaches correlate rookie development with amount of playing time even if they are not particularly efficient/effective because of the presumed value of experience gained through lots of minutes.
Statistical Methodology
I use win shares per 48 minutes (WS/48) as an indicator of efficiency. Win shares are a measure of how many wins a player adds to his team. WS/48 is a rate stat that calculates how many wins a player is adding to their team over 48 minutes (the length of a full NBA game). WS/48 is better than simple win shares as it measures the efficiency at which a player produces wins. Thus, if one player compiles win shares by playing many minutes we can still compare his rate stats to another player who might not have played as many minutes.
I compared a player’s WS/48 to their total minutes played (MP) and usage percentage (USG%) in their rookie year. USG% is a metric used to calculate what percentage of team plays a player was involved in while on the floor. This differs from MP as some rookies may spend lots of minutes on the floor but are not necessarily used directly by their team. USG% quantifies a player’s involvement in his team’s plays to a greater degree of accuracy than MP.
I sampled players drafted after 2006 (the year the ‘one-and-done rule’ began: players had to play at least one year in college before becoming draft eligible) that played at least 51 games and averaged at least 15 minutes per game (MP/G) in their first three NBA regular seasons. I compared their MP and USG% in their rookie year to the change in WS/48 from their first year to their second year and their first year to their third year. Essentially, this allows me to see the improvement or decline of efficiency between their rookie, second, and third years in the league.
I found the difference in WS/48 between years 1 and 2 by subtracting a player’s year 1 WS/48 from their year 2 WS/48 and used the same process for their years 1 and 3 difference. The first graph shows the difference in WS/48 compared to MP in the players rookie year. The second graph is similar but compares USG% instead of MP.
Analysis
Both graphs show a small positive correlation between difference in WS/48 and MP or USG% highlighted by their regression lines. But this small positive regression coefficient (slope) does not indicate a statistically significant difference in improvement of WS/48 for a player with maximal MP or USG% versus one with minimal MP or USG%. In other words, based on the regression lines of the graphs a player on the right-most end of the regression line is not producing a statistically significant improvement of WS/48 versus one on the left-most end. Additionally, the data in both graphs are weakly related to the regression lines. Furthermore, the extremely low R-squared values in both graphs highlight the low predictability of improvement or decline given any amount of MP or USG% for a rookie.
Conclusion
The NBA’s previous bias towards developing rookies through a baptism by fire is not correlated with future improvement in the analysis. Steve Kerr’s claim that observation in one’s rookie year—and not minutes or on-court usage—can correlate with successful improvement in efficiency in future years bears study.
Confounding factors include that, clearly, not all rookies are created equal. Some rookies come into the league playing at an all-star level already—such as Kevin Durant and Blake Griffin— and thus improvement in their WS/48 is difficult. Other rookies come into the league at older ages because they spend more time in the NCAA or playing in the Euro League. So, "development" is not necessary during their rookie year and instead they are immediate contributors. Overall, these players that are already playing well/efficiently will inherently be given more minutes and this could skew the data as they also have less room for improvement in years 2 and 3. Additionally, low MP and USG% in one’s rookie year also seems to not be correlated with future improvement. Lastly, due to the minimum criteria of 51 games and 15 MP/G in the sampled data, injuries may have affected the data.
Further Investigation
If we were able to rigorously eliminate some of these confounding factors that produce rookie outliers, would the same conclusion still hold? Kerr has a good development record and has rarely drafted one of these rookie outliers during his tenure as the Golden State Warriors head coach. Perhaps eliminating these outliers would yield data that more clearly favors Kerr’s approach of developing rookies (think Jordan Poole’s or Kevon Looney’s development curve).
Another possibility is that metrics other than MP or USG% should be studied. Comparing field goals attempted or assist percentage to the difference in WS/48 may yield different conclusions. Thus, data may indicate that coaches should instead encourage players to hone specific skills statistically correlated with future improvement.