Preparing volleyball teams to perform at the highest standard of competition is a complex process dependent on the interactions of technical, tactical, and fitness characteristics of available players. The nature of a strategic game such as volleyball is complex, dynamic, and nonlinear (7,8,12); therefore, the performance indicators used in the modeling process should be able to capture at least some of these properties. The available research is mainly performed under static conditions, by describing how the final outcome is affected by tactical responses (16,17,28), technical requirements (15,18,22), or physical determinants (30).
The effect of skill performance in the final outcome was already described whether at the set level (14,17,18) or at the game level (15,23,24,33). Independently of the applied methods, the results show that performance in attack is highly related to success in volleyball (15,17,33). Nevertheless, it is important to consider that volleyball is played in independent sets with different match temporal moments. The initial set is the first moment of confrontation between opponent teams, and the final set is where the match outcome is to be decided. It is clear that both of these moments carry unique physiological (28,30) and psychological features (13), which can determine performance, particularly in attack and serve actions.
For example, each specific match period could have different effects on the players' behaviors, especially in controlling physical and psychological stress. O'Donoghue (21) and Pollard and Pollard (27) showed that expending additional physical and mental effort on the important points in a tennis match, whereas relaxing on the less important points may increase the chances of winning the match. The effect of the hypothetical variation in players' efforts in 1 particular set, in a best-of-n set match, on the probability of winning the set or match was also studied within a theoretical mathematical modeling approach (2,3). The results seem to demonstrate that it is best to alternate between periods of high and low energy expenditure, rather than looking for more stability (3). Therefore, it may be possible that variability can enhance performance (2) and that the initial (4,10) and the final periods of the match (1) can assume greater importance to the final outcome of the game.
Studies have focused on the initial periods of the matches meant to characterize the influence of early success in a competition as a predictor of final outcome. Early success models (4) argue that early points scored in a match or early sets wins in a multiset match result in increased momentum, which leads to match victory (5). However, further investigations using different methodologies found different results because they showed that momentum could not be inferred from early success (32). Research focused on the final periods of the matches is more consensual, and all report that the last periods of the matches had a determinant role in the final outcome (1,20).
In Volleyball, there still exist limited information concerning the performance variation in offensive game actions, along these critical game periods. If differences exist among game periods, these may provide insights into optimal design of specific training programs to prepare for the competitions. Additionally, the knowledge of these results may also add new insights into the process of physical conditioning, by influencing coaches' decision-making process in player substitutions and using different risk strategies along the game. Thus, the main aim of this study was to examine the variation of attack and serve performances in the beginning and end of the initial and final sets of volleyball matches, according to the quality of opposition.
Experimental Approach to the Problem
The purpose of the study was examining attack and serve efficacy along different game periods, by different quality of opposition teams. The data were recorded during play at top-level competition from FIVB (Fédération Internationale de Volleyball) Men's World Cup 2007.
The study sample consisted of 600 serves and 1,128 attacks performed during FIVB (Fédération Internationale de Volleyball) Men's World Cup 2007, from 144 players of 12 nationalities (Argentina, Australia, Brazil, Bulgaria, Egypt, Japan, Korea, Puerto Rico, Russia Spain, Tunisia, and U.S.A). The protocol of the study was approved by the Ethics Committee at the Center of Research, Education, Innovation and Intervention in Sport of University of Porto and according to the revised Declaration of Helsinki.
All the matches were recorded with a camera positioned 15 m above the volleyball court and 20 m back from the edge of the field, parallel to the baseline. Four specialized teams each 1 with 2 trained operators performed the data gathering. Each team analyzed a minimum of 20 sets and a maximum of 32 sets.
There were 600 rallies selectively sampled from a total of 5,117 rallies observed. To ensure equal representation from the 12 teams in competition, a minimum of 3 and a maximum of 5 matches of each team were sampled. A 2 Step Cluster analysis (Distance Measure: Log-likelihood; Clustering Criterion: Schwarz's Bayesian Criterion) was used to group teams into competitive levels. The number of clusters was fixed in 2, and the variables used were points at the end of the competition (each win gives 2 points and each loss gives 1 point), ratio of total number of points won and lost, total ratio of sets won and lost, and the percentage of sets won. The first cluster was named Higher Level (HIGH) and included the first 4 ranked teams (Brazil, Russia, Bulgaria, and the U.S.A, corresponding to 4 matches and 16 sets), and the second cluster was named Lower Level (LOW) and included the last 5 ranked teams (Australia, Japan, Egypt, Korea, and Tunisia, corresponding to 3 matches and 9 sets). The quality of opposition considered was Higher Level vs. Higher Level (HIGH × HIGH, n = 240 rallies) and Lower Level vs. Lower Level (LOW × LOW, n = 360 rallies).
Match period was analyzed by the first and last 15 rallies of the initial and final game sets. The initial set was always set number 1; however, the final set was number 3 in matches ending 3–0, set number 4 in matches ending 3–1, and set number 5 in matches ending with 3–2. The identification of the ‘last 15 rallies’ was made in a retrospective way, that is, the rallies were taken by counting backward 15 rallies from the last rally played in the set. Efficacy of serve was assessed in a gradual 5-point scale in which 1 represent a mistake and 5 represent an ace (total reception failure), as described by Rocha and Barbanti (29). Efficacy of attack was assessed in a gradual 6-point scale in which 1 represent a mistake and 6 represent a point, as described by Palao et al. (22).
To guarantee reliability of the observations, intraobserver and interobserver agreements were assessed via the percentage error method (9,11). After a 3-week period, to prevent for any learning effect, each observation team reanalyzed one random set. For interobserver reliability testing, each observation team analyzed 1 set already analyzed by another team. The reliability values obtained were <5% error for all variables recorded.
A double 2-point Moving Average (6) of serve efficacy and attack efficacy was plotted by match period for each quality of opposition groups (HIGH × HIGH and LOW × LOW). These moving averages, or time-invariant smoothing linear filters, were applied sequentially by adding one observation at a time, assuming that the time series components change through time in a stochastic manner (6). Matlab software (Mathworks, Natick, MA, USA) was used to calculate Approximate Entropy (ApEn) values for the serve efficacy and attack efficacy in all match period and quality of opposition groups. The ApEn quantifies the amount of randomness in a time series. The algorithm essentially determines the probability that short sequences of consecutive data points repeat, at least approximately, throughout a longer temporal sequence of points. Input parameters for the ApEn calculation were (a) a series length (m) of 2 data points, (b) a tolerance window (r) normalized to 0.5 times the SD of individual time series, and (c) a lag value of 5 (25). Expressing the average probability in logarithmic form (and taking the inverse), ApEn generates a unit-less real number that ranges from 0 to 2 (26). Zero values correspond to time series where the sequences of data points are perfectly repeatable. A sine wave, for example, oscillates continuously in a repeatable and predictable fashion. Values of 2 correspond to time series for which any repeating sequences of points occur by chance alone.
HIGH × HIGH Matches
Figure 1 present the results of serve efficacy variations in the first and last 15 rallies from the initial and final sets. Because serve efficacy was accessed in a gradual 5-point scale in which 1 represents a mistake and 5 represent an ace, the higher the values, the higher the serve efficacy of both teams. The results identify that teams served better at the beginning of the initial set (Figure 1), that is, there was a decrease in the serve efficacy from the first 15 rallies to the last 15 rallies. In the final set, there was an inverse tendency, that is, the teams served better at the end of the set. The serve efficacy in the last 15 rallies oscillated in a similar way in initial and final sets. A substantial decrease was identified from the ninth to the seventh rallies, followed by a faster increase until achieving a similar efficacy.
Figure 2 presents the results of attack efficacy variations in the first and last 15 rallies of the initial and final set. Because attack efficacy was accessed in a gradual 6-point scale in which 1 represents a mistake and 6 represent a point, the higher the values, the higher the attack efficacy of both teams. The results identify that teams progressively increased the attack efficacy until the 15th rally in the initial set (Figure 2). The attack efficacy decreased until the fourth to the fifth rally and increased from here to the sixth to the seventh rally. In the final sets, the teams started with same efficacy levels until the 15th rally. The variation of attack efficacy over the last 15 rallies was similar in both sets (initial and final).
LOW × LOW Matches
The results showed very few oscillations on serve efficacy throughout the 4 analyzed match periods (Figure 3).
The attack efficacy increased progressively throughout all the first 15 rallies of the initial set (Figure 4). In the last 15 rallies, the attack efficacy had similar variation either in initial and final sets. There was a decrease in attack efficacy ranging from the ninth to the fourth rallies from the end of the set, because the lowest value was reached by the seventh rallies from the end of the set.
The purpose of this study was to examine the variation of attack and serve performance in the beginning and in the end of the initial and final sets of volleyball matches, according to the quality of opposition. Approximate Entropy values were used to identify randomness in attack and serve performances data. The results suggested that volleyball matches presented different profiles depending on the match period (i.e., the first or last rallies of the set; in the initial or final sets of the match). Some of these different profiles were observed in both HIGH × HIGH and LOW × LOW matches, and others were only observed in a particular quality of opposition group.
When the sets are moving toward its end, the attack (HIGH × HIGH and LOW × LOW) and serve performances (HIGH × HIGH) decreased. After this reduction, the performance increased to the same levels achieved so far and sometimes to even higher levels. Previous research showed that not all points have the same importance to win the matches and the best players expended additional physical and mental effort on the important points while they relaxed on the unimportant points (2,3). Our results seem to suggest that players may perceive that the most important points are played at the end of the set; thus, they manage to spare their best effort for the decisive rallies. Coaches and players acknowledge the importance of the fact that ‘leaving it all on the field’ throughout all the match should be carefully interpreted (3). Therefore, it might be advantageous to increase the attention and even measure technical, tactical, and conditioning variables in these specific conditions. This way, it might be possible to increase holistic preparation for the last set rallies.
Comparisons of serve performance in the initial set showed that higher-level teams (HIGH × HIGH) had higher serve efficacy and higher ApEn in the first 15 rallies. The serve is a closed game action, where there is no direct interference with any other player. Therefore, these standard conditions ensure that only situational variables could be affecting its efficacy (31). Moreover, the increased serve performances at the beginning of the matches could be justified by the need to have some advantage, even taking high risks, when playing against high quality opponents (16). Commonly in volleyball within this opposition level, a weak service allows for a well-organized attack by the opponent team (19). This could somehow explain why high-level players use different serve strategies depending on the period of the set. Thus, in the first 15 rallies, the players took more risks because they had nothing to lose (reflected in a high serve performance), but in the last rallies, they have refrained the risks (reflected in a lower serve performance). Consequently, the serve efficacy was more predictable as confirmed by the lower values of ApEn.
The results obtained identified better attack performances in the beginning of the set (i.e., first 15 rallies). Because the attack is game action most correlated with the final outcome of volleyball matches (15,17,33), the results showed that players, independently of their level, managed the risk according to their attack skills, to increase the score advantage as soon as possible. However, the hypothetical advantage about the likelihood of winning the match achieved in the beginning of the sets offers some controversy. Some studies showed a significant advantage when winning the first points or sets of a match, but other studies could not identified any advantages (for a review see ).
The quality of opposition interacted with performance in serve and attack when taking in consideration a specific match period. As expected, these results suggest that teams used different offensive strategies according to their opponents. Although high-level teams (HIGH × HIGH) had better serve performances in the last 15 rallies of the final sets, the lower level teams (LOW × LOW) obtained the same performance during the entire match period. The serve is likely to be affected by external situations, at least from a psychological standpoint (31). Therefore, it seems remarkable that players of high-level teams not only did not decrease the serve performance in the last points, but they were able to increase their performances. These findings were also presented by Barnett et al. (2), who modeled the probability of winning a set in a best-of-n set match. The results showed that the best players have advantage by varying their effort and strategies, taking risks over the match. In contrast, weaker players have not taken so many risks in serving strategies, as was confirmed in this study.
This study emphasizes the need of having an holistic approach to sports performance, by considering the interaction between technical, tactical, and fitness dimensions. In particular, the results might have helped to reveal the need to explore the dynamics of the strength and conditioning process by using game tactical and technical determinants. Analysis of attack and serve performances allowed a characterization of the initial and final periods of the matches. The fact that volleyball is not a time limited sport creates several difficulties when planning the conditioning drills according to competition specificities. Our results identify the need for players to perceive the most important points at the end of the set and to manage their effort throughout the match attempting to reach this period in optimal condition. This was particularly evident when it was identified that high-level teams had better performances in the last rallies of the final sets and the lower level teams obtained identical performances during the entire match period. Therefore, it might be beneficial to coaches to stress the need to perform at the highest level, particularly at the end of the training drills that simulate competition scenarios.
This work was supported by the Portuguese Science and Technology Foundation (Ph.D. scholarship, SFRH/BD/38776/2007) and Operational Program for Science and Innovation 2010 (POCI 2010) cofinanced by Social European Found (FEDER). The authors declare that they have no conflict of interest relevant to the content of this manuscript. The results of this study do not constitute endorsement of the product by the authors or the National Strength and Conditioning Association.
1. Bar-Eli M, Tractinsky N. Criticality of game situations and decision making in basketball: An application of performance crisis perspective. Psychol Sport Exerc 1: 27–39, 2000.
2. Barnett T, Zeleznikow J, MacMahon C. Using game theory to optimize performance in a best-of-n set match. J Quant Anal Sports 6: 2010.
3. Brimberg J, Hurley W, Lior D. Allocating energy in a first-to-n match. IMA J Manage Math 15: 25–37, 2004.
4. Burke K, Houseworth S. Structural charting and perceptions of momentum in intercollegiate volleyball. J Sport Behav 18: 167–178, 1995.
5. Courneya K. Importance of game location and scoring first in college baseball. Percept Mot Skills 71: 624–626, 1990.
6. Dagum E. Time series: Seasonal adjustment. In: International Encyclopedia of the Social & Behavioral Sciences. J. S. Neil, B. B. Paul, eds. Oxford, United Kingdom: Pergamon, 2004. pp. 15739–15746.
7. Davids K, Glazier P, Araújo D, Bartlett R. Movement systems as dynamical systems: The functional role of variability and its implications for sports medicine. Sport Med 33: 245–260, 2003.
8. Glazier PS. Game, set and match? Substantive issues and future directions in performance analysis
. Sport Med 40: 625–634, 2010.
9. Hughes M, Cooper S, Nevill A. Analysis of notation data: Reliability. In: Notational Analysis
of Sport: Systems for Better Coaching and Performance in Sport (2nd ed.). M. Hughes, I. Franks, eds. Abingdon, United Kingdom: Routledge, 2004. pp. 189–204.
10. Iso-Ahola S, Mobily K. Psychological momentum: A phenomenon and an empirical (unobtrusive) validation of its influence in a competitive sport tournament. Psychol Rep 46: 391–401, 1980.
11. James N, Taylor J, Stanley S. Reliability procedures for categorical data in performance analysis
. Int J Perform Anal Sport 7: 1–11, 2007.
12. Lames M, McGarry T. On the search for reliable performance indicators in game sports. Int J Perform Anal Sport 7: 62–79, 2007.
13. Males J, Kerr J, Thatcher J, Bellew E. Team process and players' psychological responses to failure in a National Volleyball Team. Sport Psychol 20: 275–294, 2006.
14. Marcelino R, Mesquita I, eds. Associations between Performance Indicators and Set's Result on Male Volleyball. Presented at 5th International Scientific Conference on Kinesiology, 2008; Zagreb, Croatia.
15. Marcelino R, Mesquita I, Afonso J. The weight of terminal actions in volleyball. Contributions of the spike, serve and block for the teams' rankings in the World League'2005. Int J Perform Anal Sport 8: 1–7, 2008.
16. Marcelino R, Mesquita I, Sampaio J. Efficacy of the volleyball game actions related to the quality of opposition. Open Sports Sci J 3: 34–35, 2010.
17. Marelic N, Resetar T, Jankovic V. Discriminant analysis of the sets won and the sets lost by one team in A1 Italian Volleyball League—A case study. Kinesiology 36: 75–82, 2004.
18. Monteiro R, Mesquita I, Marcelino R. Relationship between the set outcome and the dig and attack efficacy in elite male volleyball game. Int J Perform Anal Sport 9: 294–305, 2009.
19. Moras G, Busca B, Pena J, Rodriguez S, Vallejo L, Tous-Fajardo J, Mujika I. A comparative study between serve mode and speed and its effectiveness in a high-level volleyball tournament. J Sports Med Phys Fitness 48: 31–36, 2008.
20. Navarro R, Lorenzo A, Gómez M, Sampaio J. Analysis of critical moments in the league ACB 2007-08. Rev Psicol Deporte 18(Suppl): 391–395, 2009.
21. O'Donoghue P. The most important points in grand slam singles tennis. Res Q Exerc Sport 72: 125–131, 2001.
22. Palao J, Manzanares P, Ortega E. Techniques used and efficacy of volleyball skills in relation to gender. Int J Perform Anal Sport 9: 281–293, 2009.
23. Palao J, Santos J, Ureña A. Effect of team level on skill performance in volleyball. Int J Perform Anal Sport 4: 50–60, 2004.
24. Papadimitriou K, Pashali E, Sermaki I, Mellas S, Papas M. The effect of the opponents serve on the offensive actions of Greek setters in volleyball games. Int J Perform Anal Sport 4: 23–33, 2004.
25. Pincus S. Approximate entropy
(ApEn) as a complexity measure. Chaos 5: 110–117, 1995.
26. Pincus SM. Approximate entropy
as a measure of system complexity. Proc Natl Acad Sci U S A 88: 2297–2301, 1991.
27. Pollard G, Pollard G, eds. Importances 1: The Most Important Sets in a Match, and the Most Important Points in a Game of Tennis. London, UK: Tennis Science & Technology, 3: 2007.
28. Quiroga ME, García-Manso JM, Rodríguez-Ruiz D, Sarmiento S, De Saa Y, Moreno MP. Relation between in-game role and service characteristics in elite women's volleyball. J Strength Cond Res 24: 2316–2321, 2010.
29. Rocha C, Barbanti V. An analysis of the confrontations in the first sequence of game actions in Brazilian volleyball. J Hum Mov Stud 50: 259–272, 2006.
30. Sheppard J, Gabbett T, Stanganelli L. An analysis of playing positions in elite men's volleyball: considerations for competition demands and physiologic characteristics. J Strength Cond Res 23: 1858–1866, 2009.
31. Shondell D, Reynaud C. The Volleyball Coaching Bible. Champaign, IL: Human Kinetics, 2002
32. Weinberg R, Jackson A. The effects of psychological momentum on male and female tennis players revisited. J Sport Behav 12: 167–179, 1989.
33. Zetou E, Tsigilis N. Does effectiveness of skill in complex I predict win in men's Olympic volleyball games? J Quant Anal Sports 3: 1–9, 2007. Article 3.