Age Matters according to NBA Stats
Taking a look at the National Basketball Association, NBA, season 2017–18 wouldn’t it be interesting to find out whether rookies performed better than veterans?
We’ll use SAS software to analysis our findings but first let’s set the parameters for rookies and veterans. Let’s classify a rookie as a player with less than 2 seasons experience, which in this case was 276 players, whereas a veteran will have 10 years plus experience, 161 players fall into this category.
In this study 3 distinct datasets from the NBA season 2017–18 were utilised which comprised of:
· End of season rank for each team
· Player performance statistics for players with 0–2 years NBA experience, Rookie.
· Player performance statistics for players with 10 or more years NBA experience, Veteran.
An additional dataset was constructed which combined solely of Rookie and Veterans.
Let’s set an objective here:
Objective:
To substantiate the correlation of NBA team ranking with player age contribution.
Pearson correlation coefficients for points, PTS, against minutes played, MP, indicated that Rookies were more likely to score.
Although Rookies had a higher Pearson correlation for points per minutes played Veterans had more time in play. The mean in-play time for Rookies was 32% less than Veterans.
Distribution of minutes played:
This is further reflected in the distribution of games with Veterans having greater involvement at the close of season.
Closer analysis of the data reveals that:
14 Veteran players over 1000 points accounted for 20578 points with a mean of 1470 points
14 Rookie players over 1000 points accounted for 15534 points with a mean of 1194 points
263 Veteran players with less 1000 points accounting for 37532 with a mean of 142 points
147 Rookie players with less 1000 points accounting for 37520 with a mean of 255 points
Although 14 Veterans accrued an extra 5044 points compared to the top 14 Rookies the main difference in correlation can be found in the mean distribution of points for players who scored less than 1000. The sum total of points is very similar in this category however the Veterans achieved this total with 116 fewer players. As discussed greater time on court for Veterans would have contributed to this total.
If Rookies had a higher correlation to score an analysis of player distribution by franchise revealed that higher ranked teams, at the end of the season, contained fewer active Rookies but more active Veterans.
Number of Rookies and Veterans by team rank Dataset obtained via SAS program.
To explore contributory factors that would account greater game time for Veterans I utilised the Win Share, WS, statistic from www.basketball-reference.com/about/ws.html which uses the following calculation and includes a player’s offensive contribution.
Win Share formula:
Calculate points produced for each player.
Calculate offensive possessions for each player.
Calculate marginal offense for each player. Marginal offense is equal to (points produced) — 0.92 * (league points per possession) * (offensive possessions). Note that this formula may produce a negative result for some players.
Calculate marginal points per win. Marginal points per win reduces to 0.32 * (league points per game) * ((team pace) / (league pace)).
Credit Offensive Win Shares to the players. Offensive Win Shares are credited using the following formula: (marginal offense) / (marginal points per win).
www.basketball-reference.com/about/ws.html
Using WS in conjunction with the points scored confirms that Rookies are more likely to score but Veterans have a higher win share ratio:
As indicated win share can contain a negative value as displayed in the box plot below. Although the Rookies have a number of outliers at the 75 percentile, or Q3, for Veterans in conjunction with outliers indicate a clear difference in WS between both groups.
The figure is further reflected in an analysis of covariance:
To compare a team’s overall WS rating the following SAS query was formulated:
The series plot indicates the distribution of Win shares by rank and player classification:
The top ranked WS teams are heavily weighted towards a Veteran’s WS contribution.
There were exceptions in the 2017–18 season namely Philadelphia whose rookie players excelled but could not sustain the same levels of performance in the play-offs.
Results:
The game time afforded to Rookies was 32% less than Veterans.
The mean distribution of points for players who scored less than 1000 indicated a similar total number of points for both categories however the Veterans achieved this total with 44% or 116 fewer players than the Rookies.
Veterans performed better in areas of assists, steals, blocks and turnovers.
The top ranked NBA teams at the end of the season contained a higher proportion of veteran players.
Conclusions:
Investigation into a basketball player’s performance during the season 2017–18 by experience revealed that their overall ability was not defined by the number of points posted on the scoreboard, as reflected by the higher points correlation for Rookies, rather evidence indicates that Veteran basketball players performed better in all facets of play particularly in areas of assists, steals, blocks and turnovers. This is reflected in both the game time afforded to Veterans and the ratio of Veterans over Rookies in league standings at the end of the 2017–18 season.
Bibliography
ILLOWSKY, B and DEAN, S. (2018) Introductory Statistics. OpenStax. Rice University.
CODY, R. (2015) An Introduction to SAS, University Edition. SAS Institute.
ROHATGI, V K. and SALEH, E. (2015) An Introduction to Probability and Statistics. Third Edition. John Wiley & Sons.
DER, G and EVERITT, B. (2015) Essential Statistics Using SAS, University Edition. SAS Institute.
JARMAN, K. (2015) Beyond Basic Statistics: Tips, Tricks, and Techniques Every Data Analyst Should Know. John Wiley & Sons.
