Water polo originated in the late 1800s and is one of the oldest team sports of the modern Olympic Games. Nevertheless, every two years, the Federation Internationale De Natation (FINA) structures another planetarium water polo competition: the World Championships. In 2009, the 13^th edition of this international competition has been organized in Rome (Italy), counting the participation of sixteen national teams coming from every continent. In this Championship, matches have been played by two teams, consisting of 6 field players and a goalkeeper, for four 8-minute clock-time (i.e., excluding breaks in play) quarters, and in a court measuring 30m X 20m. Teams had to play a single action for a maximum of 30 seconds of clock time. Moreover, from the moment that a defender commits an exclusion foul, the latter has to stay out of the play for 20-second clock-time and to go to a delimited corner area located back to the goal line and close to his team bench (FINA, 2010), allowing the offence team to play with a player numerical advantage.

At present, water polo has been mainly analyzed in terms of physiological characteristic (Lozovina et al., 2004; Pavlik et al., 2005; Platanou and Geladas, 2006; Sardella et al., 1992; Smith, 1998; Tan et al., 2009; Tsekouras et al., 2005) and swimming capability (Falk et al., 2004; Melchiorri et al., 2009; Mujika et al., 2006; Platanou, 2006). Moreover, the situational nature of water polo makes difficult the game analyses in terms of replication (Lupo et al., 2010). Nevertheless, technical and tactical studies of this sport have been provided in terms of team (Hughes et al., 2006; Lupo et al., 2009; 2011) and play role (Lozovina et al., 2004; Lupo et al., 2007; 2008) efficacy, rules evolution (Platanou et al., 2007), different competition levels (Lupo et al., 2010), and influence of match outcome (Argudo Iturriaga et al., 2009; Escalante et al., 2011; Lupo et al., 2011; Smith, 2004; Vila et al., 2011). In particular, despite the latter aspect has been recently high-lighted for other team sports like basketball (Csataljay et al., 2009; Gómez et al., 2008; Sampaio and Janeira, 2003; Sampaio et al., 2010) and rugby (Vaz et al., 2010), only few water polo studies (Escalante et al., 2011; Vila et al., 2011) have been investigated for the present men’s inter-national rule (FINA, 2010). However, there is a lack of studies considering a specific margin of victory (i.e., a number of goal difference in the final score between winning and losing teams) as a discriminating factor to analyze the technical and tactical aspects of a water polo match.

Thus, the present study aimed to analyze the men’s water polo matches played during the 13^th edition of the World Championships (Roma, 2009) by comparing technical and tactical playing aspects of winning and losing teams playing in matches with different margins of victory, that is, closed games, 1-3 goals; and unbalanced games, >3 goals.

The Review Board of the University of Rome Foro Italico approved this study to investigate the game aspects of the water polo matches of the men’s World Championships.

Each team scoring of a single match was inserted in one of the four categories related to the established margin of victory. For example, not more than three goals winning team, W; not more than three goals losing team, L; more than three goals winning team, MW; and more than three goals losing team, ML. Only the match characterized by a winning and losing team at the end of 4^th quarter has been considered for the study, whereas all matches ending draw were excluded from the experimental sample. Therefore, 42 of the 48 (6 matches were draw at end of the 4^th quarter) men’s water polo matches of the 13^th FINA World Championships (i.e., Rome 2009) were considered for the study.

Considering the championship participation, 9 teams came from Europe (Germany, Hungary, Montenegro, Croatia, Serbia, Spain, Macedonia, Romania, Italy), 2 from Asia (China, Kazakhstan), 2 from North (USA, Canada), 1 from South (Brazil) America, 1 from Africa (South Africa), and 1 from Oceania (Australia). In relation to the four margins of victory, W, L, MW, and ML category counted 20, 20, 22, and 22 teams playing scores, respectively; while, the specific margins of victory of the closed and unbalanced games were 1.8 ± 0.8 (range: 1-3) and 8.1 ± 3.2 (range: 4-17) goals, respectively.

Although only based on established water polo habits and coaches’ statements (and not on published data), the elite men’s water polo players observed in the present study are usually involved in a minimum of six to a maximum of twelve 120/180-min training sessions per week.

All the men’s water polo matches of the 13^th World Championships were recorded by means of a video camera (JVC GR-DVL 107, Yokohama, Japan) positioned at a side of the pool, at the level of the midfield line, at a height of 12m and at a distance of 10m from the pool. Subsequently, the video recordings were replayed by a Video Home System (SONY SLV-E1000VC, Tokyo, Japan) to facilitate the handling of the playing images (i.e., still, replay, slow motion picture, etc.). Therefore, a notational analysis has been executed according to the following technical and tactical indicators (Table 1), which were considered in relation to even, counterattack, power-play, and transition playing situation. In particular an even situation is characterized by a number of offensive players related to the ball position which is never larger than that of the defence, within the offensive half-court; a counterattack refers to playing situations where, relatively to the ball position, the number of offensive players is larger than that of the defence, determining, therefore, at the moment of the end of the action, a real numerical advantage for the offensive players; a power-play situation originates following an exclusion foul committed by a defensive player who has to be out from the play for 20-s clock time; a transition situation is a swimming play phase performed further to a defensive action and before of the offensive arrangement of the most advanced player within the offensive half-court. 1)

Occurrence of actions, that is, the ratio between the number of even, counterattack, power-play and transition actions, with respect to the total number of actions of match. In particular, an action was defined from the moment that a player gained possession of the ball until possession was lost to the opposing team or re-obtained after shot, or other play event coincident with the resetting of the 30s action-time);

To avoid any inter-observer variability, a single experienced observer (with more than 300 analyzed water polo matches) scored all matches. However, before the study, the observer scored a random sample of 8 matches twice, where each observation was separated by 2 weeks, and quantified in terms of reliability (Kappa coefficients). The results reported agreement coefficients never below than 0.97.

Means, standard deviations and ranges (i.e., minimum and maximum) were calculated for each dependent variable. Statistical analyses were conducted using a SPSS package (version 17.00, Institute Inc., Cary, NC) and the criterion for significance was set at p ≤ 0.05.

The analysis of variance (ANOVA) approach was applied for the 1^st, 2^nd, 3^rd, 4^th, and 5^th indicator, where the playing situations (i.e., even, counterattack, power-play, and transition) were considered as dependent variables and the margin of victory (i.e., W, L, MW, and ML) as independent ones. Although for the other indicators, the independent variable remained the same, the dependent variables were the offensive roles which ended the even actions (i.e., centre forward versus perimeter player), the offensive even arrangements (i.e., 6 versus 6 or 5 versus 5, 4 versus 4 or 3 versus 3, and 2 versus 2 or 1 versus 1), the offensive roles which ended the power-play actions in relation to the offensive arrangements (i.e., centre forward and perimeter players, during the 4:2 and 3:3 arrangement), the defensive even arrangements (i.e., "pressing", "zone 2", "zone M", and "zone 2-3-4"), and the defensive power-play arrangements (i.e., "cluster", "anticipating"). Then, for each obtained significant difference in terms of margin of victory (the independent variable), Bonferroni’s post-hoc test was applied. Nevertheless, to provide meaningful analysis for comparisons, the Cohen’s effect sizes (ESs) were also calculated (Cohen, 1988). In particular, an ES ≤ 0.2, was considered trivial; from 0.3 to 0.6, small; from 0.7 to 1.2, moderate; and >1.2, large.

The percentage of occurrence of actions (Figure 5) reported differences for even (p = 0.002, ES range: 0.3-0.4), counterattack (p < 0.001, ES range: 0.6-0.7), and power-play (p = 0.001, ES = 0.4) situation. In particular, for even actions, differences emerged between ML and W (p = 0.024, ES = 0.3), and MW (p = 0.002, ES = 0.4); for counterattacks, between W and L (p = 0.001, ES = 0.6), L and MW (p < 0.001, ES = 0.7), and MW and ML (p < 0.001, ES = 0.6); and, for power-play actions, between W and MW (p = 0.035, ES = 0.4), and ML (p = 0.016, ES = 0.3), and L with respect to MW (p = 0.004, ES = 0.4) and ML (p = 0.002, ES = 0.4). For transitions, no difference has been reported.

Table 2 shows differences in terms of margin of victory for mean duration of actions (even: p < 0.001, ES = 0.4; power-play: p < 0.001, ES range = 0.5-0.6; transition: p = 0.004, ES range = 0.4-0.5), action outcome (even: goal, p < 0.001, ES range = 0.7-0.8; exclusion, p = 0.004, ES range = 0.4-0.5; lost possession, p = 0.043, ES = 0.3; penalty, p < 0.001, ES = 0.6; power-play: goal, p < 0.001, ES range = 0.5-0.6; no goal shot, p = 0.001, ES = 0.5; lost possession, p = 0.004, ES = 0.5), and zone origin (even: Z3, p = 0.044, ES = 0.4; counterattack: Z5, p = 0.024, ES = 0.4; Z6, p = 0.002, ES = 0.5; power-play: Z4, p = 0.001, ES range = 0.4-0.6) and technical execution (even: drive shot, p = 0.002, ES = 0.5; off-the-water shot, p = 0.016, ES = 0.5; power-play: shot after 1 fake, p = 0.035, ES = 0.4) of shots, in relation to the playing situations.

No difference emerged for the occurrence of the even actions ended by means of the forward’s and perimeter players’ game events (Table 3). Nevertheless, differences emerged for the even offensive arrangements (6 versus 6 and 5 versus 5: p < 0.001, ES range = 0.4-0.6; 4 versus 4 and 3 versus 3: p < 0.001, ES range = 0.4-0.5; 2 versus 2 and 1 versus 1: p < 0.001, ES range = 0.4-0.6; Table 3). For the power-play offensive arrangements and roles occurring at the end of action (Table 3), only a difference emerged for the "3-3" offensive arrangement and perimeter player events (p = 0.009). Finally, differences in terms of margin of victory emerged for "pressing" (p < 0.001, ES range = 0.4-0.6), "zone 1-2" (p < 0.001, ES range = 0.6-0.7), and "zone 2-3-4" (p = 0.002, ES range=0.4-0.5) even defensive arrangements, as well as for "cluster" and "anticipating" (p = 0.020) power-play defensive arrangements (Table 3).

This is the first study that considered technical and tactical aspects of water polo according to the new international rule and a margin of victory. In particular, the in-troducing of the latter factor allowed to investigate the water polo game by means of a higher awareness, to minimize the effect of the situational nature that inevitably makes hard the replication of technical and tactical analyses (Lupo et al., 2010) and, coherently to other team sports (Csataljay et al., 2009; Gómez et al., 2008; Sampaio and Janeira, 2003; Sampaio et al., 2010; Vaz et al., 2010), to define the influence of match outcomes on performance. Nevertheless, in this study, different playing strategies sometimes favored high data heterogeneity (high standard deviations and ranges, and small ESs), even for teams appertaining to the same category of margin of victory.

The main finding of this study showed that the men’s water polo matches of the World Championships are mainly characterized by the divergence between closed and unbalanced games rather than the comparison between winning and losing teams. Thus, it could assume that the game aspects of the elite men’s water polo matches have to be analyzed in relation to specific margins of victory and not only considering the winning and losing outcome.

For even offensive actions, the higher ML’s occurrence with respect to the W and MW ones emerged for effect of a minor frequency of counterattacks, probably due to a low ability to promptly perform the offensive action. In terms of power-play actions, both teams playing closed games (W and L) reported higher occurrences with respect to those of unbalanced game (MW and ML), confirming the importance of this particular game phase, previously highlighted in other studies on elite men’s water polo (Lupo et al., 2007; 2010). The reduced occurrences of the transitions (range: 4-6%) limit any possible explanations on the relative data.

In line to a previous study on elite and sub-elite men’s water polo (Lupo et al., 2010), the shorter duration of the L’s even actions with respect to the W and ML ones could be interpreted in favor of a different grade of ability to maintain ball possessions and to defense. In particular, it could be speculated that W are more able to maintain the ball possession, while ML hardly find good opportunity to score a goal, spending more time to perform action. On the other hand, for power-play, MW showed a shorter duration than that of the ML, W and L, speculating that, during unbalanced games, the winning teams could obtain offensive opportunities to successfully shot (as demonstrated by the goal and "no goal shot" occurrences) after few seconds from the origin of the action.

In terms of scored goals, winning teams showed their supremacy with respect to losing teams especially during even and power-play actions. However, as easily expected, for both goals scored and "no goal shots" indicators, the main divergence emerged between MW and ML, speculating different grade of abilities to create effective offensive playing opportunities as well as to limit opponent play.

Although in previous studies (Lupo et al., 2007; 2010), the best men’s water polo teams achieve high occurrence of opponent exclusions during even action in order to perform a consequent high number of power-play actions, the present data did not show any effect between winning and losing teams, but only between teams playing closed games (W and L) and MW, highlighting how this game phase (i.e., the passage from the exclusion achieved during even actions to the power-play actions) is crucial for team playing matches with small margin of victory. The lost possessions occurring during even and power-play actions, as well as the penalty achieved during even action, showed huge divergences between teams perform-ing unbalanced games, inferring likely for the interpretation of action duration data, divergent offensive abilities to maintain ball possession and defensive skills to steal the opposite ball possession.

The analysis of origin of shots did not display significant effects. However, for each game situation, winning teams showed high values of shots (despite statistically reported only for Counterattacks, between MW and L, and ML) originated from the "zone 6" (i.e., the central zone inside the five-meter area) with respect to losing ones, suggesting that the winnings teams are more able to centrally penetrate inside the five-meter area (rationally considered as the most favorable position to score a goal) and the losing ones are scarcely skilled to defend.

Consistent with the literature (Hughes et al., 2006; Lupo et al., 2010; 2011), the most performed technical execution of shots was drive shot. In particular, it could be inferred that this type of shot provides the best opportunity to end the action quickly, which is useful to limit the opponents defensive counter act.

Nevertheless, for even actions, W reported a significant low occurrence of drive shots and a higher frequency of "off-the-water shot" with respect to ML, determining, therefore, an extended selection of shot executions. Conversely, for power-play actions, the lower occurrence of the W’s shots after 1 fake with respect to the L one could be interpreted according to the latter’s necessity to mislead the goalkeeper (Lupo et al., 2010), instead of ending the action with a quick shot, which seems to be the best play solution considering the related goal and "no goal shot" data (W: goal = 44 ± 17%, "no goal shots" = 41 ± 19; L: goal = 36 ± 11, "no goal shots" = 51 ± 13).

The play events occurring at the end of even actions by centre forward and perimeter players reported quite balanced occurrences, therefore confirming the already highlighted high impact of centre forward for the game of an entire team (Lupo et al., 2007), because the offensive arrangements are mainly characterized by the presence of 5 perimeter players and only one centre forward (Figure 2).

The analysis of the even offensive arrangements showed that the teams performing closed games showed more (>90%) complete offensive arrangements (i.e., 6 or 5 offensive players) than MW (88%) and ML (84%), probably for a divergent technical and tactical, and physical abilities to quickly perform the swimming phases useful to complete an offensive arrangement, within the limit of 30 seconds clock-time of a single ball possession (FINA, 2010). Although, for the power-play offensive arrangement, the 3:3 arrangement ended by perimeter player reported a difference between MW and ML, their low occurrences limit any potential interpretation. However, consistent with previous studies (Hughes et al., 2006; Lupo et al., 2010), the 4:2 arrangement resulted the most adopted Power play offensive arrangement.

Relatively to the defensive arrangement performed during even actions, the high occurrence of the "zone 2" performed during closed games (W and L) showed how in these matches it is important to minimize the opponent centre forward play, which is crucial to obtain an opponent exclusion foul (Lupo et al., 2007; 2008). Conversely, during unbalanced games (MW and ML), the above mentioned defensive strategy was not evident, while "pressing" emerged as the most approved. Coherently to even actions, for the defensive arrangements of power-play, the teams performing the closed games tended to principally cover the opponent players located close to their goal ("cluster" defence) instead of pressing the perimeter players ("anticipating" defence; Figure 3).

Although the situational nature of water polo limit the replication of studies (Lupo et al., 2010), this technical and tactical study tends to promote the knowledge of the elite men’s water polo, promoting the identification of specific training planning (Smith, 1998) for water polo teams and roles, considering the offensive game phases as well as the defensive ones. Therefore, consistent with other studies (Hughes et al., 2006; Lozovina et al., 2004; Lupo et al., 2007; 2008; 2009; 2010; 2011), results from the current study confirms notational analysis as a useful method for better coaching (Hughes and Franks, 2004) through the interpretation of water polo technical and tactical aspects (Lupo et al., 2010), even with particular reference to specific competition levels, margins of victory, game situations and roles.

In this study, discrepancies mainly emerged between closed and unbalanced games, rather than between winning and losing teams, highlighting that the water polo matches of the men’s World Championships need to be analyzed considering successful/unsuccessful teams as well as specific margins of victory. Thus, coaches can improve their knowledge and training, considering, for closed matches, high occurrences of opponent’s exclusion fouls to perform power-play actions, while for unbalanced ones, a divergent grade of defensive skills between teams during each game situation.