End-of-Season Team Notebook

8 May, 1996

Chicago Bulls Domination

Danny Ainge was saying the other night that he thought the Bulls had the best defense in the NBA. By points allowed, they were only third, but by points per 100 possessions, they were the best at 99.8; only Seattle was within one point of that. Not surprisingly, the Bulls were also the best offensive team in basketball, scoring 113 points per 100 possessions; no team was within one point of that, with Utah second at 111.7. I've never seen a case where one team was so dominant, perhaps in any sport. And, of course, since the Bulls won so many games so easily, putting scrubs in early in the fourth quarter or just not pushing it, these averages do not reflect how good the Bulls really are.

          Off. Rtg   Rank Def. Rtg   Rank
ATL        107.2      12   105.9      15
BOS        105.0      16   108.4      22
CHA        109.4       4   110.0      28
CHI        113.0       1    99.8       1
CLV        107.9      10   104.9      11
DAL        104.4      20   109.4      25
DEN        103.3      24   106.2      17
DET        106.3      15   103.5       7
GSW        106.6      14   108.1      21
HOU        107.6      11   105.8      14
IND        108.9       6   105.3      13
LAC        104.8      17   108.6      24
LAL        109.3       5   104.5       9
MIA        103.9      22   102.4       6
MIL        103.9      23   109.6      26
MIN        102.3      25   107.9      20
NJN        100.2      28   104.7      10
NYK        104.5      19   102.0       4
ORL        111.1       3   105.2      12
PHI        100.5      27   111.1      29
PHO        108.8       7   108.5      23
POR        104.3      21   101.8       3
SAC        104.8      18   107.6      19
SAS        108.7       8   102.1       5
SEA        108.7       9   100.6       2
TOR        102.0      26   109.9      27
UTA        111.7       2   104.5       8
VAN         96.0      29   106.7      18
WAS        107.1      13   106.0      16

Isn't it a little surprising that Denver's team defense is bad when they have one-time Defensive Player of the Year, Dikembe Mutombo? This isn't the first time their team defense has been bad. It was bad last year, too, making me seriously question the reputation Mutombo has gained.

On the other hand, Gary Payton certainly deserves some credit for his defense, though perhaps not Defensive Player of the Year. Scottie Pippen, Michael Jordan, Hakeem Olajuwon all rank at least as high. And how 'bout a Blazer (ranked 3rd in defense)? But which one? Maybe PJ Carlesimo should have gotten some credit for putting together an unsung defense without any defensive stars...

It amazes me nearly every year that Utah's three-man offense of Stockton, Malone, and Hornacek can really be that good. There really is so little support for these guys that it is a tremendous credit to them that defenses have such a hard time stopping them.

Consistency Measures

I have been emphasizing the importance of consistency in evaluating players and teams (see Basketball's Bell Curve, What Strategies Are Risky?, Shooters vs Dunkers: Their Effects on Winning , etc.). In short, consistency is good when you are good on average. Consistency is bad when you are bad on average. On the other hand, being inconsistent if you are a bad team on average helps you win more. I have evaluated some preliminary consistency numbers for the NBA teams. Specifically, I have calculated the standard deviation of points scored and allowed for all teams. I have also calculated the covariance of points scored vs. points allowed and come up with the overall estimate of consistency for each team. These are shown below, with comments to follow:

                                         Cov(Off.,
           Off. SD   Rank  Def. SD   Rank    Def.)   Rank Consistency  Rank
IND          9.8       3     9.2       1    33.3      24    10.7       1
MIL         10.3       6    10.6       8    45.2      13    11.3       2
LAC         10.1       4    10.5       7    41.8      17    11.3       3
CHI          9.6       1    10.0       5    31.2      25    11.4       4
VAN          9.6       2    11.1      15    42.4      15    11.4       5
DAL         13.4      28    11.3      20    85.5       1    11.7       6
UTA         10.3       7    12.1      24    56.1       8    11.8       7
POR         11.4      19    11.1      16    55.4       9    11.9       8
HOU         11.3      17     9.8       2    39.7      20    12.0       9
NJN         10.9      11    13.2      29    74.2       4    12.0      10
DET         10.2       5    11.0      14    39.3      21    12.1      11
GSW         12.2      23    10.9      13    60.6       6    12.1      12
PHO         11.4      18    11.2      17    50.0      11    12.5      13
CHA         10.5       9    11.3      19    39.9      19    12.6      14
LAL         10.5      10     9.9       4    23.8      27    12.7      15
DEN         12.5      25    12.7      28    78.2       2    12.7      16
PHI         11.1      15    11.8      21    49.6      12    12.8      17
TOR         10.9      12    12.3      25    51.5      10    12.9      18
ATL         11.0      13    12.7      27    57.0       7    13.0      19
SAS         11.1      16    10.7       9    34.3      23    13.0      20
SEA         12.5      26    10.0       6    41.9      16    13.1      21
WAS         11.5      21    11.2      18    39.0      22    13.4      22
CLV         12.1      22    10.8      10    41.0      18    13.5      23
NYK         12.3      24    12.4      26    61.1       5    13.5      24
ORL         11.5      20     9.8       3    20.3      28    13.7      25
MIN         10.4       8    10.9      12    17.8      29    13.8      26
MIA         14.1      29    11.9      22    74.2       3    13.9      27
BOS         13.0      27    10.8      11    44.2      14    14.0      28
SAC         11.0      14    12.0      23    30.0      26    14.3      29

Again, there is Chicago near the top. Not only were they on average amazing, but they were consistently dominant.

One of the strangest entries near the top of this chart is Dallas. From game to game, their offense and their defense were terribly inconsistent, but they were very consistent in playing to the level of their competition. In other words, they always played just bad enough to lose.

Given the flux in Houston's personnel through the course of the season, I would have expected a much more inconsistent team. Overall, they were the ninth most consistent team in the league, somewhat different from their past two championship seasons when their "high variance" style really helped them win as underdogs in the playoffs.

Teams that rank lowest in the Covariance(Off., Def.) category have the smallest value and, hence, play to the level of their competition the least. Generally, these teams are either very good or very young. Numbers 24-29 on the list fit this pattern fairly well: Indiana, Chicago, Sacramento, Los Angeles Lakers, Orlando, Minnesota. I tend to find that young teams with a small correlation between their offense and defense have a ways to go before they improve, which doesn't bode well for Sacramento and Minnesota.

The four most inconsistent teams underwent some major personnel changes this season, but the fifth most inconsistent team, Orlando, stayed pretty much the same. The most consistent team, Indiana, had very little change from last season and that change was from a spot-up long range shooter named Byron Scott to a spot-up long range shooter named Ricky Pierce.

Did They Win Enough or Too Much?

Generally, a team's winning percentage is a fairly accurate reflection of the points scored and allowed by that team. What this implies is that we can determine what winning percentage a team should have based on their scores throughout the season. I have two methods for doing this, the Pythagorean Method and the Correlated Gaussian Method, both of which are explained in The Fundamentals for Analyzing Basketball. The Pythagorean Method uses only the overall points scored and allowed by a team to estimate winning percentage. The Correlated Gaussian Method uses the consistency numbers presented above along with points scored and allowed to estimate winning percentage. Below, I compare the results of the two methods to teams' actual winning percentages:

                            Pythagorean Corr. Gauss
            W     L    WIN%      Win%      Win%
ATL         46    36   0.561     0.551     0.537
BOS         33    49   0.402     0.370     0.404
CHA         41    41   0.500     0.477     0.482
CHI         72    10   0.878     0.885     0.859
CLV         47    35   0.573     0.617     0.576
DAL         26    56   0.317     0.316     0.337
DEN         35    47   0.427     0.389     0.416
DET         46    36   0.561     0.608     0.582
GSW         36    46   0.439     0.442     0.452
HOU         48    34   0.585     0.570     0.558
IND         52    30   0.634     0.634     0.619
LAC         29    53   0.354     0.357     0.376
LAL         53    29   0.646     0.675     0.637
MIA         42    40   0.512     0.561     0.541
MIL         25    57   0.305     0.291     0.320
MIN         26    56   0.317     0.293     0.349
NJN         30    52   0.366     0.325     0.362
NYK         47    35   0.573     0.598     0.568
ORL         60    22   0.732     0.711     0.658
PHI         18    64   0.220     0.160     0.217
PHO         41    41   0.500     0.513     0.511
POR         44    38   0.537     0.598     0.578
SAC         39    43   0.476     0.391     0.425
SAS         59    23   0.720     0.739     0.686
SEA         64    18   0.780     0.782     0.724
TOR         21    61   0.256     0.227     0.281
UTA         55    27   0.671     0.749     0.711
VAN         15    67   0.183     0.150     0.191
WAS         39    43   0.476     0.543     0.532

Generally, there is enough noise in an 82 game season to explain differences between actual records and estimated records of about four games, or a difference in winning percentage of about 0.050. If a larger difference between actual and estimated winning percentage is observed, this indicates that the team may not be as good or as bad as their record. For example, the Washington Bullets' estimated winning percentage (using either method) was substantially higher than their actual winning percentage. This indicates that the talent on the team is better than their record implied. This was because the Bullets went only 4-8 in games decided by 3 points or less, games that often come down to luck. Presumably, the Bullets will not continue to be unlucky.

Other teams deviating significantly from their expected winning percentages include Orlando, Seattle, and Sacramento. Unlike Washington, however, all three teams won more than was expected, implying that they may not win as much next season. It's hard to say how much to trust this measure, but probably not very much. Really the only team that is guaranteed to get worse next season is Chicago -- because they can't get any better.... On season record alone, Chicago would have a 58% chance of beating the Laker team that had the 69-13 record.