BIG BLUE FANS FOR
ANALYSIS OF THE GAME OF BASKETBALL
INDIVIDUAL EFFICIENCY ANALYSIS
Those who have read my efficiency based analyses of Kentucky basketball teams over the last several years know that my entire focus has been on the efficiencies that the entire team produced. I have essentially ignored the individual players in those analyses. However, during the 2009-10 season, I initiated an examination of efficiencies for individual players. I began this new project because I am now satisfied that the team-based system has stabilized and the major elements of that system are in place and functioning. In addition, I was curious to learn whether the general wisdom about the relative contributions of Wall, Cousins, Patterson, et al had an objective basis. I have continued the individual efficiency tracking since, and now through the 2014 season. I have updated the historical data to reflect the additional data.
Efficiency measures the points produced or allowed for each possession of the game, on average, over the course of a season. The difference between offensive and defensive efficiency is the Net Game Efficiency, NGE, also measured as points per possession. In 2009-10, the average NCAA D1 basketball team posted an efficiency of 1.011 points per possession [ppp] with an NGE = 0.000 ppp. For a team, it is a relatively easy concept to grasp. However, for individual players, how should variable playing time, steals, turnovers, blocks, rebounds [defensive and offensive], shots, etc. be factored into an individual analysis is the big issue. This system for individual efficiency is new, and so long as it continues to show promise, I will continue to modify and develop the concept in the years ahead.
At the offensive end of the court, I started with a Raw Scoring Efficiency for each player based on points scored per scoring opportunity. A scoring opportunity is a shot taken or a free throw opportunity [1, 2, or 3 free throw attempts]. For Kentucky's players, Stevenson had the highest Scoring Efficiency, 1.319 ppp, and other than Krebs, Harris had the lowest, 0.871 ppp. However, we all understand that an offensive rebound continues a possession after a missed shot, to some extent obviating the impact of the missed shot, and a turnover ends a possession, eliminating an opportunity to take a shot. Therefore, I adjusted the scoring opportunities by crediting a player for an offensive rebound and debiting a player for a turnover. Finally, many scores result from the assist of a teammate. It seems unfair to give the scoring player full credit for the score when he benefits from the sacrifice of his teammate for the benefit of the team. Therefore, I adjusted the points scored by crediting the assist maker a portion of basket, and debiting the assist recipient those same points.
The adjusted offensive efficiencies showed a different outcome than the raw values. Patterson, Stevenson, and Cousins led the team at 1.349, 1,339, and 1,228 ppp respectively. The lowest offensive outputs, other than Krebs, were Harrellson, Dodson, and Bledsoe at 0.831, 0.874, and 0.953 ppp respectively. The remaining players fell in a range of 1.03 to 1.10 ppp, all above the NCAA D1 team average.
On the defensive end, I started the analysis from a subjective view that defensive productivity is much more a result of team effort than offense, so each player started with a presumption that he contributed equally to the overall team defensive efficiency of 0.914 ppp. However, blocked shots and steals are clear factors that distinguish one player from another on the defensive end of the court. When the individual defensive efficiency is adjusted for the contributions of these two additional factors, the results show the most efficient defensive players were Cousins, Wall, and Patterson with efficiencies of 0..879, 0.883, and 0.891 ppp respectively. The weakest defensive contributors, other than Krebs were Hood, Harrellson, and Harris at 0.952, 0.933, and 0.927 ppp respectively. The other players on the roster fell within a narrow range of 0.92 and 0.90 ppp. All UK individual defensive efficiencies were better than the NCAA D1 team average.
NET GAME EFFICIENCY [NGE]:
The best measure of efficiency for a team is the Net Game Efficiency [NGE]. I believe that a similar measure, Net Game Efficiency [NGE] will provide the best measure for individual players' seasons. The following Table summarized the results of this 2009-10 individual efficiency analysis, sorted from highest to lowest NGE.
TABLE I -2011
TABLE I -2012
TABLE I -2013
TABLE I -2014
The team NGE for 2009-10 was 0.197 points per possession. Four members of the 2009-10 roster exceeded that team average, Patterson [0.458 ppp], Stevenson [0.416 ppp], Cousins [0.349 ppp], and Wall [0.213 ppp]. Two players ended the season with negative NGE values, Dodson [-0.049 ppp] and Harrellson [-0.102 ppp]. Of course, it is no surprise that Patterson, Cousins, and Wall are now hot commodities in the NBA draft. It is a little sobering that Harrellson and Dodson are 2 of the 5 returning players for the 2009-10 roster. In 2011, the average NGE dropped to 0.172 ppp, and the Championship team posted a very strong 0.251 ppp on its 38-2 journey to the title. In 2013, the team finished very poorly following Nerlens Noel's season ending injury, and limped to the finish line with an average NGE of 0.137 ppp. In 2014, the team struggled throughout most of the regular season, and especially during the SEC season, losing 10 games leading into the post season. In the Post Season, this team transformed itself and managed to advance to the championship game before falling to UConn. The Cats finished at 0.122 ppp for the season despite their strong post season run.
PLAYER RANKINGS FOR 87 UK PLAYERS; 1996-2010:
I have applied the same analysis to the rosters of the 1996 through 2009 UK teams. I do not present the corresponding equivalents of Table I here for these other 14 UK teams due to space limitations, but I have compiled the results into a ranking in Table Il on the next page. Eight-nine players appear in the year ending box scores over the subject 15 seasons. The data provided in Table II represents each of these players' full UK career except for those players that appear on a 1996, 1997, or 1998 roster and played in 1995 or earlier. I could not include any data prior to 1996 because the statistical archives for those prior years do not include offensive rebounding data, an essential component of the individual and team efficiency analyses.
Table II includes 3 separate methods of ranking individual players. Initially, I simply averaged the year ending individual NGE values for each player. However, this method produced some unexpected outcomes primarily because a few players had 1 season out of 3 or 4 that were so anomalous to their overall UK career that distorted the final result when that season received equal weight with the other 2 or 3 seasons for those few players. For example, Jodie Meeks played 3 seasons, and his second season was marked by serious injury. Despite his injury, Meeks tried to play in several games that season. As a result, his NGE value for 2008 was much poorer than his other two UK seasons.
Therefore, I added two additional methods of ranking, one weighted by the number of games played in each active season and the other weighted by the number of minutes played in each active season. Table II contains all three ranking systems, and is sorted based on the ranking that the game weighting system produces.
COMPARISON WITH PRIOR UK TEAMS:
While the results of the 2010 analysis of individual efficiencies appear reasonable, it is only one team, one year. This is not sufficient to justify any confidence in the methodology. Therefore, I have performed the same analysis of individual efficiency for each of the UK teams between 1996 and 2010, the last 15 UK teams spanning four coaches. Space does not permit the inclusion of the equivalent of Table I for each of these other 14 UK teams. Therefore, I have summarized each team by tabulating the distribution of individual NGE for each of the 15 teams as the efficiency profile for each team.
I have also included the final offensive, defensive and Net Game Efficiencies for each respective team, and further condensed the team profiles into a Cummulative Summary for the number of players on the team with positive NGE values and negative NGE values. I direct your initial attention to the team data in the center portion of Table III. Notice that the team NGE for 1996 and 1997 were 0.282 ppp and 0.260 ppp respectively. No UK team since has had a NGE above 0.200 ppp.
Last year's team came closest [0.197 ppp] and the 1998 Championship team was next closest at 0.191 ppp, which the 2003 team also matched. Other than the temporary rise in 2003, the entire period between 1999 and 2008 is marked by a gradual yet steady decline in team NGE values, culminating in a 0.046 ppp in 2008 in Gillispie's first year at UK. While two years does not establish a trend in the same way that the 10 years of 1999-2008 did, it appears that the UK NGE values are on the rebound from the low point of 2008 with 0.105 ppp in 2009 and 0.197 ppp in 2010.
Table III provides a summary of the individual player season NGE for 1996 through 2010 for comparison purposes. The team profiles presented in these distributions do not include players that did not play sufficient games or minutes to significantly impact the team outcomes. The criteria for inclusion is a player had to play at least 100 minutes AND appear in over 50% of all games during the entire season.
The 2009 team presents an interesting case study. That team truly had 2 very different “seasons” within the season. Through the first 20 games, that team was 16-4; 5-0 in the SEC including three very difficult early SEC wins on the road. Through 20 games, the team was performing better than any UK team since Pitino left with an offensive efficiency of 1.068 ppp and a defensive efficiency of 0.873 ppp for a NGE of 0.195 ppp. However, subsequent to game 20, the 2008-09 UK team collapsed as turmoil engulfed its coach, players, and athletic department. Over the last 16 games, the team's offensive efficiency fell to 1.022 ppp and the defensive efficiency rose to an astounding 1.036 ppp for a NGE of -0.014 ppp and a 6-10 record down the stretch. The team efficiencies in Table III represent the average of these two disparate mini-seasons for 2009.
Now, I direct your attention to the top ½ of Table III and the distribution of players by their individual NGE values for each cited season. This distribution ranges from a high of 0.4 ppp or higher and a low of -0.4 ppp or lower, with 0.10 ppp increments between these extremes. The column for the 2010 team is the distribution of the results for 2010 in Table I. The distributions of the 1996 through 2009 teams appear in those respective columns of Table III. At the bottom 1/3 of Table III is a recasting of the basic team profiles, showing Cumulative totals for players with positive and negative NGE values.
Again, note how the profiles for the 1996 and 1997 teams are top heavy with 2 players above 0.4 ppp and 0 and 1 player with negative NGE values in 1996 and 1997 respectively. Further, note how the number of players at the top ranges declines and the number of players in the negative ranges increases from 1999 through 2008. However, 2009 and 2010 shows a trend toward the top-heavy distributions of 1996 and 1997.
It appears from the analysis that Coach Gillispie began the difficult task of reversing Tubby Smith's trend of a deteriorating UK basketball roster and deteriorating play. Furthermore, it seems clear that Coach Calipari continued the work of returning UK to national prominence in 2010, and based on recruiting results to date for 2011, 2012 and beyond, Coach Calipari is well on his way to returning UK to the top of the national college basketball scene.
COMPARISON WITH PRIOR NCAA CHAMPIONSHIP TEAMS:
Table IV provides a summary of the individual player season NGE for 1996 through 2010 for 12 of the 15 National Champions over that 15 year period for comparison purposes. This Table results from the same analysis described above. No data is available for the 1997 Arizona, 1999 UConn, and 2002 Maryland teams to perform the analysis. Therefore, those columns are empty.
I did add 4 columns to Table IV to present the mean and ranges of results achieved by these 12 champions. As noted above, please note that these Championship teams posted team NGE values that averaged 0.226 ppp, and that the 1996 UK 0.282 ppp is the Gold Standard among these NCAA championship teams. However, the 2008 Kansas team's 0.270 ppp is clearly in that same rarified air. Furthermore, notice that the Championship teams have individual player profiles that are top heavy, with very few players in the negative NGE ranges. The average NCAA Champion has had almost 2 players at 0.4 ppp or higher, and less than 1 player with a negative NGE values. Except for the 1998 UK team that won the Championship that had 3 players in the negative range, of the other 11 NCAA Champions, six had only 1 and five had none.
These 12 Champions finished with team NGE values ranging from 0.133 ppp [Syracuse 2003] to 0.282 ppp [UK 1996] with a mean of 0.226. Ten of the twelve Champions had a season NGE above 0.206 ppp, and the mean of the 12 is 0.226 ppp. While the 2010 UK result of 0.197 ppp was sufficient to win a Championship if circumstances broke their way, the 0.197 ppp NGE value does not quite measure up to the compelling criteria of a NGE value of 0.206 or higher established by this data.
UK has some more work to do to return the program to true final four and national competitiveness.
COMPARISON OF UK COACHES WITH NCAA CHAMPIONSHIP TEAMS:
Table IV provides a summary average individual team NGE profiles, by Coach, for UK from 1996 through 2010, and for 12 of the 15 National Champions. While I would like to complete the Pitino profile by adding his 1992 through 1995 teams into this analysis, it is clear that the two Pitino teams included in this analysis establish the parameters of greatness when compared to the average NCAA Champions during this 15-year period. It is also clear that the average team profiles of the Tubby Smith teams do not compare favorably with the minimum criteria [Team NGE > 0.206 ppp; at least 2 players in the 0.4 or higher range, and no more than one player with a negative individual NGE value] for legitimate NCAA Championship teams.
It is equally clear that Gillispie inherited a depleted roster yet he began to reverse the decline in the first 20 games of his second season prior to the complete collapse that will forever define his brief UK tenure. Finally, based on the 2010 results and the recruiting success to date for 2011 and 2012, Calipari appears well positioned to finish the job at UK, and in my opinion, in relatively short order.
This analysis of individual player efficiencies is clearly a new work in process, and I would not want anyone to misrepresent or overstate the significance of this early attempt to quantify the individual contributions to a team activity. Much more work will be required before any reliance upon this type of analysis of individual players can be reasonably justified. However, in my opinion, the use of the individual outcomes from this analysis to describe a team profile and evaluate that team relative to other teams in the same year, and prior years appears to have a solid foundation. I am encouraged by the results of this early analysis, even with the possible anomalies that occur within the individual player rankings. Therefore, I will continue this work into the 2010-11 season. I welcome comments and suggestions from those interested in this type of analysis.
CHECK OUT THESE OTHER ANALYTICAL WRITINGS