Importance of Consistency
Obviously, teams do not play at the same level of efficiency on the offensive and defensive ends each time they play. FN1 For example, a team that has an average offensive efficiency of 0.9 points per possession will perform below and above this average about 50 percent of the time. The variance and skew of this distribution are measures of a team's consistency.
Fans often lament that the only thing wrong with their team this year is a lack of, or poor consistency. However, such laments are often invalid. Fans tend to remember those games in which their team played at the top of their game, and lament when the team does not perform at that elevated level every game. Fans rarely confront the average condition of their team's game, and the statistical variations that occur around that average level of performance.
Here are the numbers on consistency for recent Kentucky teams plus the 1996 championship team. As the data illustrates, Kentucky's consistency season to season falls within relatively narrow ranges for offense and defense.
Off Eff 
Def Eff 
Off. Stdev 
Def Stdev 
NORMALIZED 
NET GAME 
WINNING 
POWER 

STD 
DEV 
EFF 
PERCENT 
RATING 

2005  0.916 
0.777 
0.121 
0.124 
13.2% 
16.0% 
0.137 
85.7% 
118.0% 
2004  0.930 
0.808 
0.114 
0.100 
12.3% 
12.4% 
0.122 
85.3% 
115.1% 
2003  0.936 
0.809 
0.107 
0.136 
11.4% 
16.8% 
0.127 
86.8% 
115.7% 
2002  0.889 
0.806 
0.115 
0.126 
12.9% 
15.6% 
0.083 
70.6% 
110.3% 
2001  0.928 
0.814 
0.131 
0.135 
14.1% 
16.6% 
0.115 
72.2% 
114.1% 
2000  0.838 
0.775 
0.114 
0.133 
13.6% 
17.2% 
0.063 
71.4% 
108.2% 
1996  0.960 
0.747 
0.116 
0.128 
12.1% 
17.2% 
0.213 
94.7% 
128.5% 
Average  0.914 
0.791 
0.117 
0.126 
12.8% 
16.0% 
0.123 
81.0% 
115.7% 
Maximum  0.960 
0.814 
0.131 
0.136 
14.1% 
17.2% 
0.213 
94.7% 
128.5% 
Minimum  0.838 
0.747 
0.107 
0.100 
11.4% 
12.4% 
0.063 
70.6% 
108.2% 
Std. Dev.  0.040 
0.025 
0.007 
0.012 
0.9% 
1.7% 
0.048 
9.5% 
6.6% 
For the years 1996 and 2000 through 2005 season to date, Kentucky's offensive efficiency has averaged 0.915 ppp with a normalized average standard deviation of 12.1%. On defense, Kentucky has averaged 0.782 ppp with an average normalized standard deviation of 16.8%. These average standard deviations indicate that in recent years, Kentucky has been more consistent on offense than defense. In the years included in this tabulation, Kentucky's offensive consistency for the season has been better than its defensive season consistency in every year. Of course, that difference is marginal for the 2004 season, but statistically significant for the other years.
The 2005 team however is 1/3 through the regular season, and is estabishing new records for offensive consistency [8.5%] and defensive inconsistency [22.1%]. Ignoring the partial results of the 2005 partial season, the offense efficiency standard deviation has varied between 11.4% and 14.1% while their defensive efficiency standard deviations have varied between 12.4% and 17.2%. The 2002 team and the 2000 team posted very similar results, results favorably comparable to the average standard deviations. Regardless, during those seasons, many fans attributed the overall poorer performance of those teams to inconsistency. However, the fan frustration was caused by overall diminished quality of play, not inconsistency. Consider the following statements:
• The 2002 or the 2000 team is more inconsistent than the 1996 championship team
• The 2002 team or the 2000 team is more inconsistent than the 2001 team.
These misperceptions are simply exposed by the data.
The 2005 Kentucky team has posted an impressive first 11 games [1/3 of typical season] at 102. The following table shows the distribution of the Net Game Efficiencies for all posible combinations of offensive and defensive performances over ranges of plus and minus one standard deviation unit from the means. For offense, the range is 0.997 to 0.827 ppp. For defense, the range is 0.574 to 0.863 ppp. Kentucky should win whenever the Net Game Efficiency is greater than zero, and only 2 of the 81 possible combinations produce a negative Net Game Efficiency.
Mean  0.9122  0.7183  
Std. Dev.  0.0852  0.1442  
Mean +/ Std Dev 
1 
0.75 
0.5 
0.25 
0 
0.25 
0.5 
0.75 
1 

Off. Eff. 
0.9974 
0.9761 
0.9548 
0.9335 
0.9122 
0.8909 
0.8696 
0.8483 
0.8270 

Def. Eff. 
N 
E 
T 
G 
A 
M 
E 

0.5742 
0.4232 
0.4019 
0.3806 
0.3593 
0.3380 
0.3167 
0.2954 
0.2741 
0.2528 

0.6102 
0.3871 
0.3658 
0.3445 
0.3233 
0.3020 
0.2807 
0.2594 
0.2381 
0.2168 

0.6462 
0.3511 
0.3298 
0.3085 
0.2872 
0.2659 
0.2446 
0.2233 
0.2020 
0.1807 

0.6823 
0.3151 
0.2938 
0.2725 
0.2512 
0.2299 
0.2086 
0.1873 
0.1660 
0.1447 

0.7183 
0.2790 
0.2577 
0.2364 
0.2151 
0.1938 
0.1725 
0.1512 
0.1299 
0.1086 

0.7544 
0.2430 
0.2217 
0.2004 
0.1791 
0.1578 
0.1365 
0.1152 
0.0939 
0.0726 

0.7904 
0.2069 
0.1856 
0.1643 
0.1430 
0.1217 
0.1004 
0.0791 
0.0578 
0.0365 

0.8265 
0.1709 
0.1496 
0.1283 
0.1070 
0.0857 
0.0644 
0.0431 
0.0218 
0.0005 

0.8625 
0.1348 
0.1135 
0.0922 
0.0709 
0.0497 
0.0284 
0.0071 
0.0142 
0.0355 
Please note that the standard deviation for defense is nearly twice as large as offense. The offensive standard deviation is lower than observed by this writer over the last 10 years, when values seemed to be 0.11. Similarly, the defensive standard deviation is slightly higher than in recent years. I have no explanation for why 2005 would be different from other years. However, the data is clear; the 2005 Kentucky team is playing with greater than usual consistency at the offensive end and less consistency than usual on the defensive end. Perhaps this relates to the much easier early season scheduling during November and December 2004.
If Kentucky maintains this relationship of average efficiencies and standard deviations for them, Kentucky will be extremely difficult to defeat. About 30 percent of game performances should fall outside one standard deviation from the means. This means that on 15 percent of the games, the offensive efficiency will be poorer than the mean less one standard deviation, and similarly for the mean plus one standard deviation unit. This relationship applies to offense and defense. The odds are quite low that Kenucky would have an offensive performance poorer that the mean less one standard deviation in the same game as the defensive efficiency is greater than the mean plus one standard deviation unit. The current 2005 data indicates that Kentucky should win 22 percent of its games by 20 or more points, win 44 percent of its games by 10 to 19 points, and win 31 percent of its games by 1 to 9 points. Kentucky should lose about 3 percent of its games.
Can Kentucky maintain this early performance levels on offense and defense? That is what consistency is really all about.
FN1 Some games will produce better offensive efficiencies and poorer defensive efficiencies. Other games will produce better offensive efficiencies and better defensive efficiencies. Yet other games will have poorer than average offensive performance with poorer, better or average defensive performance.
Copyright 200405 Richard Cheeks
All Rights Reserved