+34 96 633 71 35

1 Jun 2012

Team success in basketball. Explanations for the united states of america’s dominance at the beijing olympic games (2008)

Posted By

Anecdotally, the fast pace at which the USA men’s basketball team placed at the 2008 Olympics was the main reason for their dominance, although there is no way of quantifying what a fast pace is or how it contributed to point differentials.

Autor(es): Lago, C
Entidades(es): Universidad de Vigo
Congreso: VII Congreso Nacional de Ciencias del deporte y educación Física
Pontevedra 5, 6 y 7 de Mayo de 2011
ISBN: 978-84-614-9945-8
Palabras claves: Statistics, team sports, match analysis

Resumen explanations for the united states of america’s dominance at the beijing (2008)

Anecdotally, the fast pace at which the USA men’s basketball team placed at the 2008 Olympics was the main reason for their dominance, although there is no way of quantifying what a fast pace is or how it contributed to point differentials. The aim of this study was to examine the game-related statistics that discriminate between fast- and slow-paced games, as well as to identify key performance factors relating to point differentials. We analysed game-related statistics for each quarter of the eight games played by the USA using a k-means cluster analysis to classify game pace using ball possessions per game quarter. I then tested for differences in game statistics between slow- and fast-paced game quarters using analysis of variance and discriminant analysis. How differences in game-related statistics affected point differentials was examined using linear regression. The largest structure coefficient between game paces for the USA was for recovered balls (0.33, P<0.001). The biggest contributors to the point differences in games were recovered balls (16.9,P<0.001) and field goals (22.2,P<0.001).I conclude that when the USA play a fast-paced game, they are able to recover more balls from opponents that they then turn into effective field-goal shooting.

1. Introduction

Preparing Basketball teams to perform at the highest level of competition is a very complex process, much dependent upon the characteristics from the available players which are decisive to configure the team strategy. Traditionally, the NBA and the USA national teams have been very dominant worldwide. Over the last years, with the increase of foreign players in the NBA it could be expected that the differences between USA basketball and the rest of the world would decrease. However, in Beijing Olympics (2008) the USA team has won the gold medal with 8 wins averaging a 27.9±11.8 points difference (range 11-49). It was suggested by basketball analysts that this success was based on a faster game pace. In fact, the USA team averaged 81.1 ball possessions per game against 70.7 from all tournament teams.[1] This faster game pace is probably supported by a superior defensive assertiveness carried by high-level conditioned players. The defensive performances are very difficult to measure objectively, because they can either have direct consequences (such as a steal or blocked shot) or, like most of the time happens, indirect consequences (such as an opponent turnover or a low probability field-goal attempt).[2] Therefore, the sum of team ball steals and blocked shots (gaining possession) with the opponent team turnovers provide a fair estimate from recovered balls caused by defensive assertiveness. It is believed that defensive assertiveness is used to change game pace, that is, a team may accelerate the game by increasing defensive assertiveness.[3] However, there is no evidence on how playing faster or slower affects game performance. Also, there are no references characterizing this highest level of basketball performance.
Generally, basketball game performance depends upon shooting field-goals and securing defensive rebounds.[5,6,7] In more specific game contexts such as closely contested games, fouls and free-throws exhibit higher importance.[7] Other game-related statistics such as offensive rebounds, turnovers, steals or assists, are not reported consistently as discriminators between winning and losing teams. On the other hand, when the criterion is not game outcome (winners against losers) but season-long success (best teams against worst teams) these results change. Within this topic, Ibáñez et al.[8] have related basketball season-long success with best performances in assists, steals and blocks, denoting the importance from overall passing skills and better outside and inside defensive intensity. In fact, it is common to find empirical references to how important are the offensive performances to win games and how important are the defensive performance to win championships.[9] This is probably due to a greater stability in defensive performances,[10] probably because they are less influenced by environmental factors (such as game location). More recently, it has been suggested that the best breakdown of offensive and defensive performances can be obtained by analyzing four factors by this order of importance: (1) effective field goal percentage, (2) offensive rebounding percentage, (3) turnovers per ball possession, and (4) free-throw rate. [10, 11]
Thus, the aim of this study was to identify the game-related statisticsthat discriminate between faster and slower paced games played by the USA team and their opponents. Additionally, we aimed to identify how the four performance factors are related with USA’s team game quarter outcome.

2. Methods

Sample and Procedures
Archival data were obtained from the official FIBA play-by-play records for the Beijing Olympics (2008). There were eight games played by the USA team against the following opponents: China (preliminary round, 31 point win), Angola (preliminary round, 21 point win), Greece (preliminary round, 23 point win), Spain (preliminary round, 37 point win), Germany (preliminary round, 49 point win), Australia (quarter-final, 31 point win), Argentina (semi-final, 20 point win) and Spain (final, 11 point win). The play-by-play game-related statistics were accumulated by game quarters (n=64) and included free-throw, two and three point field-goals (both missed and made), defensive and offensive rebounds, assists (pass that contributes directly to a field goal), committed fouls, blocked shots, steals and turnovers. Afterwards, the recovered balls were calculated by adding steals, blocked shots (that ended in gaining possession) and opponents’ turnovers. Team ball possessions and accumulated score differences at the beginning of each game quarter were also calculated and registered. Ball possessions were defined as the period of play between when one team gains control of the ball and when the other team gains control of the ball [10] and were calculated using the equation: Ball possessions = (field-goals attempted) – (offensive rebounds) + (turnovers) – 0.4 x (free throws attempted).
According to the available literature, [10,11] the effective field goal percentage was calculated by the equation: Effective Field Goal Percentage = (Field Goals Made + 0.5 x Three Point Field Goals Made) / Field Goals Attempted. Accordingly, the offensive rebounding percentage was calculated by the equation: Offensive Rebounding Percentage = Offensive Rebounds / (Offensive Rebounds + Opponents Defensive Rebounds). The recovered balls per ball possession were calculated by the equation: Recovered Balls per Ball Possession= (Steals + Blocked Shots + Opponents Turnovers) / Ball Possessions. Finally, the free throw rate was calculated by the equation: Free Throw Rate = Free Throws Made / Field Goal Attempted.
For analysis purposes, it were used the differences between the confronting teams in these four performance factors. All data was gathered by FIBA professional technicians, however, two games were used to test data reliability and results were very satisfactory (coefficients above 0.92).

3. Data Analysis

Stage 1. The game-related statistics that discriminate between faster and slower paced game quarters
A k-means cluster analysis was performed in order to identify a cut-off value of ball possessions in order to classify the game quarters. The results allowed identifying cluster 1 (faster game quarters) with ball possessions averaging 20.9±0.9 (n=21, range 19.3-22.7) and cluster 2 (slower game quarters) averaged 17.0±1.3 (n=11, range 15.1-18.8) ball possessions.
One-way independent measures Anova was used to test for univariate differences between slower and faster game quarters. All game-related statistics were normalized to quarter ball possessions. Additionally, a descriptive discriminant analysis was performed in order to determine which of the obtained variables were more useful in predicting game pace. The discriminant analysis is considered to be robust with these derived rate variables.[12] The interpretation of the obtained discriminant function was based on examination of the structure coefficients greater than |0.30|, meaning that variables with higher absolute values had a powerful contribution to discriminate between groups.[13] Validation of discriminant models was conducted using the leave-one-out method of cross-validation.[12] Cross-validation analysis was used in order to understand the usefulness of discriminant functions when classifying new data. This method involves generating the discriminant function on all but one of the participants (n-1) and then testing for group membership on that participant. The process is repeated for each participant (n times) and the percentage of correct classifications generated through averaging for the n trials.

Stage 2. How the four performance factors relate to game quarter outcome
Linear regression models were used to identify the effects of the independent variables on game quarter outcome (points scored-points allowed) in the whole game (Model 1), in the first half game quarters (Model 2) and in the second half game quarters (Model 3). When estimating both models, evidence of heteroskedasticity in residuals and multicollinearity among regressors were not found. Moreover, the RESET test by Ramsey [14] did not reveal specification problems. When interpreting the statistical results, positive or negative coefficients indicate a greater or lower propensity to increase/decrease game quarter outcome (QO), respectively. Four independent variables were included in the model: Effective Field Goal Percentage (FG), Offensive Rebounding Percentage (OR), Recovered Balls per Ball Possession (RB), and Free Throw Rate (FT). The model is as follows:

Contenido disponible en el CD Colección Congresos nº 15

Differences in accumulated score differences at the beginning of each quarter and in game quarter outcome were tested with non-parametric Wilcoxon matched pairs test.
The statistical analyses were performed using SPSS software release 16.0 (SPSS Inc. Chicago, Illinois, USA) and STATA for Windows version 10.0 (Stata Corp. LP. Texas, USA). The level of statistical significance was set at P < 0.05.

4. Results

Stage 1. The game-related statistics that discriminate between faster and slower paced game quarters
The Table 1 presents the descriptive results obtained in faster and slower paced game quarters for USA team and for the Opponents. For the USA team, univariate Anova allowed indentifying differences in 2 point field goals made, free throws missed and recovered balls, with higher values identified in faster games (P < 0.05). The differences in the Opponents were only observed in committed fouls (P < 0.05). In the discriminant analysis, a c2=34.5 was obtained for the USA team (P < 0.01) and a c2=19.0 for the Opponents, however, this result failed to reach statistical significance (P > 0.05). The structure coefficients quantify the potential of each game-related statistic to maximize differences between means amongst slower and faster paced game quarters. The larger the magnitude of the coefficients, the greater the contribution of that game-related statistic to the discriminant function. For the USA team, the discriminant function reflected an emphasis on recovered balls.

Table 1 – Descriptive results and discriminant function structure coefficients (SC) from the game-related statistics for slower and faster games (mean ± S.D.).

Contenido disponible en el CD Colección Congresos nº 15

The leave-one-out test summarizes the ability of the discriminant functions to correctly classify the game quarters in their respective groups (slower of faster). This analysis provided an overall percentage of successful classification of 93.8% for the USA team.

Stage 2. How the four factors relate to game quarter outcome
The effects of the independent variables on game quarter outcome are displayed in Table 2. In the whole match (Model 1) the game quarter outcome was explained by the four factors. However, their order of importance was changed and the recovered balls appear as the second most important factor in explaining differences in game quarter scores. For each recovered ball per possession more than the opponent, the USA team increases by 16.85 points the game quarter outcome. The intercept was not statistically significant. The linear regression model explained about 71% of the variance in game quarter outcome.
The game quarter outcome in the first half game quarters (Model 2) was explained by the four independent variables included in the model. The importance of each recovered ball per possession more than the opponent increased to 18.49 points the game quarter outcome (P < 0.01). The coefficient of determination was 0.87.
Finally, the game quarter outcome in the second half game quarters (Model 2) was explained only by superiority in field goal percentages. Each additional recovered ball increases by 16.27 points the game quarter outcome (P < 0.01). The coefficient of determination was 0.43.

Table 2. The influence of the four performance factors on USA’s game quarter outcome. Results from the three regression models (S.E.).

Contenido disponible en el CD Colección Congresos nº 15

From the first half game quarters to the second half game quarters there were statistically significant differences in accumulated score differences at the beginning of each quarter (7.3±8.8 vs. 25.5±10.3, P < 0.01) but not in game quarter final outcome (8.5±5.7 vs. 7.4±5.1, P > 0.05).

5. Discussion

The aim of this study was to identify the game-related statistics that discriminate between faster and slower paced games played by the USA team and their opponents. Also, we aimed to identify how the advantages in the four performance factors are related with USA’s team game quarter outcome. We reasoned that USA’s success was based on a faster game pace, probably originated by a superior defensive assertiveness. Consequently, this will be shown in an increased importance in recovered balls.
As suspected, the multivariate analysis allowed identifying only the recovered balls as the variable with enough power to be considered a robust variable to discriminate between slower and fast paced games. The recovered balls are a measure of defensive performance dependent upon players’ assertiveness and fitness level.[3, 15] This is required to all team players because, for example, here are included the guards’ ability to steal the ball and the centres ability to block the field goals.[16] More indirectly, this is also a measure of forcing the opponents to lose the ball. That is, effective team defensive communication can lead to increased defensive pressure in all passing spots, and ultimately to bad passes and/or passes that are easier to steal.[17] Also, it has been suggested that less effective teams use ball dribbling more frequently, [18] which can increase the probability of a steal. Consequently, this style of play increases the opportunities for fast breaks and, therefore, accelerates the game.
There is no evidence in the available literature on how different are the performances in slower of faster games. As reasoned, when the USA team play faster recovers more balls (2.64±1.43 vs. 5.24±2.47, P < 0.05) and this is probably reflected in the increase of fast breaks and quick decision offensive patters, taking advantage of an unbalanced defense (because has lost ball possession). The opportunities to score field goals originated by this reasons should be much more near the basket and, obviously, with higher probability of success. Therefore, it was expected that results from two point field goals attempted remain equal and results from two point field goals made increase (6.09±2.12 vs. 8.00±2.10, P < 0.05). Globally, it is suggested that differences in recovered balls (+2.6 per quarter) reflected in 1.91 points per quarter.
There were also differences in free-throws missed (1.27±1.49 vs. 2.62±1.56, P < 0.05). With the increase in game pace, it is probable that players could be in a more assertive state and also more fatigued. These two conditions could be affecting them at foul line. These wasted opportunities to score via free-throw had the origin in opponents committed fouls. In fact, these were the only differences identified between the opponents slow and fast paced games and lead to understand why playing at faster pace was so disadvantageous for them. In fact, to USA’s opponents, accelerating the game only resulted in an increase of committed fouls. Basketball experts acknowledge that one of the keys to excellent team defense is to exert pressure on the opposition without fouling.[9] Team fouls are directly associated with opponents’ opportunities to score from the free-throw line and with the threat of losing players by the disqualification rule. It is a fact that teams that foul less exhibit higher defensive performances.[9,10] However, it is unknown if this is originated by their defensive efficiency, by their opponent’s offensive inefficiency or by both.
Winning to the opponents in the four performance factors explained much from the USA’s game quarter score differences. Whole game results showed, not surprisingly, that having a better effective field goal percentage than your opponent is the most important factor to win the game quarter. However, results also showed that recovered balls per possession were the second most important factor. That is, recovering more balls than the opponents in defense is more important than maintaining ball possession in offense through rebounding. This was probably the key to success, i.e., an assertive defense forces the opponents to take riskier decisions and, ultimately, this increases the probability to recover balls.
To a much lesser extent having a better free-throw rate than the opponent was somehow important. This variable expresses the ability to get to the foul line and score free throws. Available research recognizes the importance of free-throw shooting because the majority of close games are decided, ultimately, at the free throw line. [7, 19,20] For example, in the last 5 minutes of close games, Kozar et al. [7] showed that winning teams scored more than two-thirds of the final score via free throws. However, because all of USA’s games were decided by wide margins (27.9±11.8 points difference, range 11-49), the free throws have lost importance.
This observed general trend was substantially different when analyzed the first half and second half game quarters. In the first half, all four performance factors had higher coefficients to account for 87% of the variance. On the contrary, in the second half game quarters, only superiority in effective field goal percentage was a significant factor and the model explained 43% of the variance. Integrating all results, it seems that the USA team played a stronger defense in the first half and has a result the differences at the beginning of the second half were already too high to be a major concern (25.5±10.3 points). Because of this, it is probable that players were unable to maintain the same intensity. The Olympics’ schedule is very concentrated and this enhances the importance of an adequate recovery. Therefore, it would be more likely that winning by large margins would cause a much less intense play.
In conclusion, the results provide enough evidence to justify USA’s basketball success. These differences are attributable to players’ characteristics, team strategy and game direction. Moreover, results suggest that games final scores can be underestimating the differences between the USA’s teams and their opponents.

Practical applications

  • The highest level of basketball competition is sustained by individual and team defensive performances.
  • This style of play requires high level conditioned players.


  1. Dezman B, Trninic S, Dizdar D. Expert model of decision-making system for efficient orientation of basketball players to positions and roles in the game – Empirical verification. Coll Antropol 2001; 25(1):141-52.
  2. Gómez M, Tsamourtzis E, Lorenzo A. Defensive systems in basketball ball possessions. Int J Perform Anal Sport 2006; 6(1): 98-107.
  3. Gomez M, Lorenzo A, Sampaio J, et al. Game-related statistics that discriminated winning and losing teams from the Spanish men’s professional basketball teams. Collegium Antropologicum 2008; 32(2): 451-6.
  4. Ibanez S, Sampaio J, Feu S, et al. Basketball game-related statistics that discriminate between teams’ season-long success. Eur J Sport Sci 2008; 8(6):1-4.
  5. Ibanez S, Sampaio J, Saenz-Lopez P, et al. Game statistics discriminating the final outcome of Junior World Basketball Championship matches (Portugal 1999). J Hum Movement Stud 2003; 45(1):1-19.
  6. Ittenbach R, Esters I. Utility of team indices for predicting end of season ranking in two national pools. J Sport Behavior 1995; 18(3): 216-25.
  7. Karipidis A, Fotinakis P, Taxildaris K, et al. Factors characterizing a successful performance in basketball. Journal of Human Movement Studies 2001; 41(5): 385-97.
  8. Kozar B, Vaughn R, Whitfield K., et al. Importance of free throws at various stages of basketball games. Perceptual and Motor Skills 1994; 78(1): 243-8.
  9. Krause J, Pim R. Coaching Basketball, Columbus. McGraw-Hill 1994.
  10. Kubatko J, Oliver D, Pelton K, et al. A Starting Point for Analyzing Basketball Statistics. Journal of Quantitative Analysis in Sports 2007; 3(1): 1-24.
  11. Norušis, M. SPSS for windows release 6.0, Chicago. SPSS Inc. 1993.
  12. Oliver D. Basketball on paper. Rules and tools for performance analysis, Washington. Brassey´s, Inc 2004.
  13. Otto K. Defense Wins! Scholastic Coach & Athletic Director 1988; 67(10): 24-5.
  14. Pelton, K. Olympic Recap. The Final Numbers. Basketball Prospectus. http://www.basketballprospectus.com/article.php?articleid=409 (accessed October 2008).
  15. Ramsey J. Tests for specification errors in classical least-squares regression analysis. J R Stat Soc 1969; 31(B): 350-71.
  16. Sampaio J, Janeira M. Statistical analyses of basketball team performance: understanding teams’ wins and losses according to a different index of ball possessions. Int J Perform Anal Sport 2003; 3(1): 40-9.
  17. Sampaio J, Janeira M, Ibáñez S, et al. Discriminant analysis of game-related statistics between basketball guards, forwards and centres in three professional leagues. Eur J Sports Sci 2006; 6(3): 173-8.
  18. Tabachnick B, Fidell L. Using Multivariate Statistics, 5th ed. Boston: Allyn and Bacon 2007.
  19. Trninic S, Dezman B. Differences in playing efficiency structure of three types of basketball players in defence. 4th International Scientific Conference on Kinesiology, Proceedings Book 2005; 522-5.
  20. Trninic S, Dizdar D, Dezman B. Empirical verification of the weighted system of criteria for the elite basketball players quality evaluation. Coll Antropol 2000; 24(2): 443-65.

Leave a Reply

Open chat
Saludos de Alto Rendimiento:

Para información sobre los cursos y másteres ONLINE, puede contactarnos por aquí.

Asegúrate de haber completado el formulario (azul) de información del curso/máster.