Handball Offensive Analysis: Comparative Evolution of Linear Tendency Lines between Finalist Teams in the EHF CL

Abstract

(1) Background: This study aims to analyze the evolution of the finalist teams in five years of the EHF Champions League and compare winning teams with defeated teams, presenting their evolution along the competition from the quarterfinals to the final; (2) Methods: A total of 35 games were analyzed, corresponding to four games per team over five seasons. Tendency lines were created according to the 13 offensive indicators of winning teams, divided into three macro categories: game actions, goals, and effectiveness of shots; (3) Results: The results indicated 21 linear tendency lines valid for the study. Winning teams presented five increasing lines (ascending line) and 16 decreasing lines (negative slope) during the competition. Assists, goals from the 6 m, and effectiveness at 6 m shots presented an increasing tendency line for winning teams, and the effectiveness of shots, the effectiveness of the organized attack, and the effectiveness of shots from 9–6 m presented a decreasing tendency line; (4) Conclusions: winning teams have more constancy during the competition. Due to the expected increased difficulty in successive matches, tendency lines tend to present decreasing slopes as teams advance toward the final. Nevertheless, winning teams need to adapt to the increasing challenge, describing growth in some key indicators, and being more regular than other teams.

1. Introduction

The complexity of the game and the constant difficulty in evaluating the performance of athletes contribute to understanding the dynamics and relationships between players of any given team [1,2,3]. In this sense, performance analysis in handball aims to provide an opportunity to predict the result of the game, and to be well-prepared for future competitive scenarios [4].

Knowledge of the indicators that describe performance in team sports is a determining factor for coaches’ decision making when planning training and during matches [5,6,7]. Accordingly, the analysis of handball matches during the main competitions is becoming an essential factor influencing the focus of training programs and the content of training units [5,8].

To better recognize the outcome of a match, it is essential to focus on a type of analysis that allows an understanding of the relationships between the factors that lead to victory [9,10]. To find a model, Debanne and Laffaye [11] showed that it is possible to use relevant variables to predict the winner and loser of a handball match and their difference in goals by applying a multiple regression model, using data from the German and French championships.

On the other hand, “longitudinal studies” are used to recognize changes in the game, using the same sample in all data collections [12,13]. For instance, in handball match analysis, researchers use longitudinal studies to create a pattern or evolution of teams over the years [14,15,16,17]; determining the evolution of games’ actions after a certain period can emphasize the development tendency of the handball match [18].

Moreover, several other studies in handball try to identify why teams are successful, making the comparison between winners and losers [19,20,21,22,23]. Hence, the tendency lines aim to show how winning teams evolve during competition within the determined success indicators, whether constant, growing, or descending.

Ferreira Filho, Sousa [24] observed and evaluated handball matches from 1986 to 1995, in a total of 19 games, to thereby diagnose the level of technical–tactical performance of handball teams and compare the performance levels of teams in world championships and Olympic games, and found that there was an evolution in the performance parameters. Vuleta, Sporiš [22] also suggest in their research that it is necessary to follow the tendency of a handball match and its constant changes in certain variables.

In the pursuit of identifying the evolution of teams, Alexandru and Acsinte [18] make a comparison with the evolution of teams in the European championship in two consecutive years, with the primary variable of the study being the number of goals scored. This allowed knowing the evolution of each team, the tendency to improve the game, the way each team adapted to the particularities of its players, the international conception of games, and the way of training.

Given the above, the tendency line method in sports was used by Mendes, Fuentes [25] to measure the variability in tennis players’ services. In this sense, a linear tendency line is a straight line of best fit used with simple linear data sets. In accordance with these requirements, Skarbalius, Pukėnas [21] has noticed an evolution in the variability of goals scored by teams in the European championship of handball between the years 2002 and 2010. Ferrari, Vaz [26] analyzed the French men’s teams as world champions in seniors and juniors. The results showed the team’s progression during the championship from the quarterfinal to the final match within the offensive variables of the handball match.

Finally, Gryko, Bodasiński [27] assessed the tendency of teams in two competitive years of the world championship (2013 and 2015). The results indicated that the analysis of this cycle revealed changes in the characteristics of the offensive game and that they should have an impact on the specific training so that teams’ game actions are closer to the opponent’s goal area, where multiplayer actions are fundamental to the successful performance in the game, as evidenced by the tendency observed in handball over the last few years [14,28].

Therefore, the main aim of the study was to analyze the successful performances in the EHF CL over 5 years within each competitive stage (the quarterfinal, the semifinal, and the final [26,29]). We evaluate their development during the competition using tendency lines according to the key indicators of competitive success in a win–lose comparison.

2. Materials and Methods

For this study we used a total of 35 matches, and analyzed the key indicators of the quarterfinals (4 matches), the semifinals (2 matches), and the final (1 match) of the EHF Champions League finalists in five seasons (n = 35), 2012–2013, 2013–2014, 2014–2015, 2015–2016, and 2016–2017, considering the averages of the finalists and making a comparison between winning and losing teams [26,29].

The variables used in this study were stipulated according to [29]. They determined the variables that differentiate the winners from the losers during the same seasons mentioned above. The effectiveness of the shots and the offensive actions were considered, and technical or tactical errors were excluded. The authors identified 13 key indicators for the match variables of winning teams and divided them into three categories: the offensive actions, goals, and effectiveness of shots.

Matches were analyzed using VideObserver^® (VO SPORTS, Portugal) and hereafter, the data were exported to a spreadsheet in Microsoft Excel 2016 for Windows^®. Then, a third-order exponential function [25] was applied, aiming to determine the linear trends of teams according to the variables considered as determinants for the success of teams in the analyzed games. In this way, the slope of the line was observed, defined as variable m (matrix), which indicated whether the slope was increasing (ascending line) or decreasing (negative slope), or if was constant, thus allowing the analysis of the linear tendency performance of the respective teams [21,26].

To better understand the results, the value of the coefficient of determination (R²) was used to indicate, as a percentage, how well the model can explain the observed values, using Pearson’s correlation to measure the degree [5,30,31].

For the reliability of the observation of the evaluated matches, intra- and interobserver agreements were used in all criteria, as was stipulated by the index of Kappa Cohen show in

Table 1. Kappa values for intraobserver and interobserver.

	Intraobserver		Interobserver
Variable	Kappa	CI (95%)	Kappa	CI (95%)
Assists	0.85	0.83–0.88	0.82	0.80–0.85
Goals	0.99	0.98–0.99	0.98	0.98–0.99
Game Action	0.84	0.82–0.86	0.88	0.86–0.90
Effectiveness of Shooting	0.83	0.81–0.85	0.82	0.80–0.84

[32,33].

This study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of FCDEF.UC (Scientific Committee of FCDEF.UC, date: 22 April 2021, Project nr. 137905NON).

3. Results

Following the previously mentioned study, the results were divided according to the offensive actions, goals, and effectiveness of the finalist teams during the five seasons, and within those analyzed periods, there was never the same winning team, showing the difficulty encountered by the teams in the EHF CL. Teams were divided by country: two German teams (2012–2013 and 2013–2014), one Spanish team (2014–2015), one Polish team (2015–2016), and one Macedonian team (2016–2017). Teams that finished second were from Spain (2012–2013), Germany (2013–2014), France (2016–2017), and the same team from Hungary finished second twice (2014–2015 and 2015–2016).

3.1. Offensive Actions and Goals

Assists were the only variable that presented significant values for the two teams. They both had a strong correlation (75%) in the evolution of the championship and both presented values in a growing line. Furthermore, the winning teams scored goals in increasing amounts at 6 m, showing a very strong correlation (95%).

The defeated teams show a decreasing tendency line in a total of goals (m = −0.9 very strong correlation 99%) and goals in collective actions type-I (m = −1.2 strong correlation 87%); winning teams do not have any negative lines and the goal in positioned attack also shows no tendency (

Table 2. Descriptive analysis of offensive actions performed in matches.

Variables		Quarterfinals ± DP	Semifinals ± DP	Finals ± DP	m	R²
Assists	Winners	8.10 ± 3.30	10.20 ± 1.33	10.20 ± 1.72	1.05	0.75 **
	Losers	7.71 ± 3.01	8.20 ± 2.56	8.20 ± 4.07	0.24	0.75 **
Total Goals	Winners	29 ± 3.98	33.40 ± 5.89	30.20 ± 4.92	0.6	0.06
	Losers	30 ± 4.44	29.20 ± 1.60	28.20 ± 5.49	−0.9	0.99 ***
Goals Action Type-I	Winners	22 ± 3.48	28.60 ± 6.15	25.40 ± 3.61	1.7	0.26
	Losers	25 ± 3.85	23 ± 1.26	22.60 ± 4.50	−1.2	0.87 **
Goals Attack Positioned	Winners	20.60 ± 3.57	25.20 ± 3.97	21.40 ± 6.05	0.4	0.02
	Losers	20.40 ± 4.20	17.80 ± 3.49	20.20 ± 4.92	−0.11	0.00
Goals 6 m	Winners	16.20 ± 3.27	18 ± 2.97	18.80 ± 4.96	1.3	0.95 ***
	Losers	14.50 ± 5.10	17.20 ± 2.93	13.40 ± 4.59	−0.5	0.09

Note. M = Matrix; ** strong correlation; *** very strong correlation.

).

Tendency lines for the assists made by teams during the competition were ascending for both winning and losing teams, but the line of the winning team compared to the defeated team shows a steeper inclination, which demonstrates a greater evolution during the competition of the winning teams (Figure A1).

In the second group of analyzed variables, both winning and losing teams had very close numbers in the quarterfinals, but in the development of the competition, winning teams always showed a more significant evolution than the defeated teams, mainly in regard to the total number of goals and the goals scored from 6 m. As to goals from Type-I collective actions and positional attacks, teams were shown to be more balanced and had closer lines of tendency (Figure A2).

3.2. Effectiveness of Shooting

Winning teams presented an ascending line of tendency in the effectiveness of type-I collective actions (m = 1.37 strong correlation 73%) and efficacy in shots from the 6 m (m = 1.50 moderate correlation 51%); losing teams do not show any increasing tendency line in shooting effectiveness (

Table 3. Descriptive analysis of the effectiveness of shooting performed in matches.

Variables		Quarterfinals ± DP	Semifinals ± DP	Finals ± DP	m	R²
Action Type-I	Winners	66.53 ± 6.38	69.21 ± 4.84	69.17 ± 6.39	1.37	0.73 **
	Losers	72.67 ± 6.08	66.29 ± 2.32	63.63 ± 8.48	−4.51	0.94 ***
Action Type-III	Winners	68.08 ± 15.75	67.70 ± 17.19	45.25 ± 19.60	−11.41	0.88 **
	Losers	65.64 ± 21.69	67.70 ± 14.87	57.76 ± 8.66	−3.94	0.56 *
Goals Attack Positioned	Winners	63.61 ± 8.33	63.51 ± 4.77	62.12 ± 8.93	−0.74	0.80 **
	Losers	67.48 ± 8.82	58.09 ± 7.75	57.88 ± 9.60	−4.80	0.76 **
Goals Fast Attack	Winners	79.75 ± 17.34	77.90 ± 14.91	74.05 ± 16.12	−2.85	0.96 ***
	Losers	75.03 ± 13.13	70.89 ± 17.13	69.43 ± 9.82	−2.77	0.92 ***
Goals	Winners	66.39 ± 4.72	66.32 ± 3.43	65.43 ± 6.24	−0.47	0.80 **
	Losers	70.71 ± 5.71	65.57 ± 3.06	61.15 ± 7.02	−4.78	0.99 ***
9 m	Winners	47.43 ± 9.71	46.91 ± 12.02	35.67 ± 7.70	−5.88	0.78 **
	Losers	55.49 ± 4.23	52.42 ± 8.86	51.44 ± 14.49	−2.02	0.91 ***
9–6 m	Winners	55.23 ± 13.95	52.74 ± 13.56	52.61 ± 8.78	−1.30	0.78 **
	Losers	65.24 ± 12.41	54.43 ± 13.45	50.94 ± 13.28	−7.14	0.91 ***
6 m	Winners	72.74 ± 5.63	71.67 ± 10.71	75.67 ± 7.45	1.50	0.51 *
	Losers	76.28 ± 8.78	79.42 ± 7.52	66.43 ± 8.34	−4.67	0.52 *

Note: m= matrix; * moderate correlation; ** strong correlation; *** very strong correlation.

).

There was a large majority of negative tendency lines in Table 2. The effectiveness of winning teams decreases for collective actions type-III (m = −11.41 strong correlation 88%), organized attack (m = −0.74 strong correlation 80%), and fast attack (m = −2.85 very strong correlation 96%), in goals (m = −0.47 strong correlation 80%), in shots at 9 m (m = −5.88 strong correlation 78%), and shots between 9 and 6 m (m = −1.30 strong correlation 78%).

Defeated teams exhibited a negative tendency line for their efficiency in all analyzed variables: in collective actions type-I (m = −4.51 very strong correlation 94%), in collective actions type-III (m = −3.94 moderate correlation 56%), effectiveness in organized attack (m = −4.80 strong correlation 76%) and fast attack (m = −2.77 very strong correlation 92%), in the effectiveness of goals (m = −4.78 very strong correlation 99%), shots at 9 m (m = −2.02 very strong correlation 91%), shots between 9 and 6 m (m = −7.14 strong correlation 78%), and shots at 6 m (m = −4.67 moderate correlation 52%).

The effectiveness of teams has always been very close, forming two very similar lines; however, the winning team only demonstrates a growing line in the effectiveness of shots from collective actions type-I. As to the effectiveness of shooting in organized attack and fast attack, winning teams presented a tendency line that was superior to defeated teams, and even when there was a negative slope in the line, it was less declivity than the tendency line of defeated teams. The opposite is shown in the effectiveness of the collective actions type-III, where the slope of the winning teams’ line is much steeper in comparison to that of defeated teams (Figure A3).

The effectiveness of completed shots displayed the most dispersion in tendency lines, where teams presented the most significant discrepancies between them. Defeated teams exhibited a higher average in all analyzed variables in the quarterfinals of the EHF CL, but along the competition, their effectiveness of shots tended to decrease and to exhibit a much higher slope than winning teams for the effectiveness of shots and goals, the effectiveness of shots between 9 and 6 m, and shots at 6 m. Only in the effectiveness of the shots from the 9 m did defeated teams present better results than winning teams (Figure A4).

When we evaluated all of the tendency lines in our study, the results indicated that 21 of those lines were statistically valid according to Pearson’s correlation. Five of those lines presented an ascending slope, with the team presenting an evolution during the championship in this final phase of the competition, and 16 presented a negative team evolution due to a higher difficulty found in the final phase of the competition.

4. Discussion

Linear tendency lines allow an understanding of the evolution of the finalist teams during the championship and for a perspective to be drawn on where teams improve or worsen according to key performance indicators [28]. The evaluation of the assists appears to be a critical factor in the growth of teams during the competition, showing a more significant interaction between players and the demand for decision making in creating easier shots to be performed. Therefore, the assist is considered the final pass for the accomplishment of the shot when an opponent cannot intervene between the shooter and the goalkeeper. The tendency line of this variable shows the superiority of winning teams over the defeated teams in the growth of the straight line (m = 1.05 winners; m = 0.24 losers), and the values of the means, confirming the studies of Bajgorić, Rogulj [34], are such that successful teams make a greater number of assists compared to less successful ones, due to having a more collective and less individualized play.

The total goals scored tend to decrease with teams confronting opponents with a higher degree of difficulty, resulting in more balanced games between teams at an advanced stage of the competition. Ferreira Filho, Sousa [24] noted that throughout the years, there was a natural tendency for goals to decrease during the evolution of the championship, as teams were more defensively organized, causing a greater balance between them. This explains the increase in the number of goals scored from 6 m by winning teams, evidenced in the tendency line (m = 1.3 very strong correlation 95%), because there are more goal assists and players are always looking for the best option for the shot, being the closest possible to the goal (6 m), which makes the shot easier for goal conversion [35].

For Type-I collective actions, which are the offensive actions that demonstrate a beginning, a progression, and finalization, defeated teams showed lower tendency lines with fewer goals scored than winning team, which was observed to decrease during the evolution of the competition (m = −1.2 losers). This was contrary to Ferrari, Vaz [26], who reported values that showed an increase in type-I collective actions from the initial course until the end of the championship. As were the results obtained by these authors, Ferrari, Vaz [19] and Vuleta, Sporiš [22,26,36], shooting effectiveness was the category that most differentiated the winners from losers, and it is the category that shows more reliability in the tendency lines. In the efficacy of type-I collective actions, winning teams presented a positive tendency line (m = 1.37), the opposite of the defeated teams (m = −4.51 very strong correlation 94%). In the efficacy of the type-III collective actions, both teams presented a decrease in the tendency line because of a higher level of play and a more difficult offense. However, this declining slope is significantly steeper for winning teams than for the defeated teams (m = −11.41 winners, m = −3.94 losers), very similar to the study of Ferrari, Vaz [26], which also showed a larger declivity for the team that was successful in the competition.

A very similar tendency line is verified in the effectiveness of the organized attack, as presented in the graph that demonstrates two lines of negative tendencies; however, the winning team’s line (m = −0.74) shows a much shallower slope compared to the defeated team’s slope (m = −4.80), a difference of −4.06 between both teams. Concerning the effectiveness of the goals in fast attack, the finalist teams also showed a similar decrease in the tendency lines (m = −2.85 winners very strong correlation 96%; m = −2.77 losers very strong correlation 92%) showing that the average values of the effectiveness of winning teams were higher than the defeated teams.

For the effectiveness of the total goals scored, the finalists showed a decrease during the evolution of the championship (Ferreira Filho et al., 2001), but the winning teams in this study show greater offensive stability during the matches, in which the tendency line is very close to a straight line (m = −0.74), different to the defeated teams. The defeated teams start with higher effectiveness than the winning teams, but during the course of the competition, the tendency line shows an accentuated decrease in the total effectiveness of shots (m = −4.78 very strong correlation of 99%).

The 9 m efficacy variable is the only key indicator in which defeated teams have a higher average than winning teams, from the beginning of the competition to the end, even though both teams have negative values in the tendency lines (m = −5.08 winners, m = −2.02 losers very strong correlation 91%). These findings are in line with Gryko, Bodasiński [27], who, in their study, confirmed that there was a decrease in the efficiency of the shots from 9 m during the course of the competitions, mainly in comparison between the winners and losers. For the effectiveness of shots made between 9–6 m, the finalists also showed a decrease in the tendency line, but again, winning teams showed a more regular average than defeated teams (m = −1.30 winners; −7.14 losers very strong correlation 91%), showing that the comparison between teams ends up being made for regularity throughout the competition, taking into account that the average of the defeated teams starts higher in comparison with the winning team, but in the end, the winning team stays more regular until the final game.

The most significant difference in the efficiency of the shots between the analyzed teams was in the shots from 6 m. Once again, the winning teams started with lower averages than the losing teams in the quarterfinals, and during the evolution of the competition, they increase their scores, presenting a positive tendency line (m = 1.50). The opposite occurs with the defeated teams showing a higher average than the winning teams, but during the competition, they will decrease (m = −4.67). This explains, therefore, the studies of Gryko, Bodasiński [27] and Srhoj, Rogulj [37], which indicated that the effectiveness of shots from 6 m tended to increase during the final phase of the competition for the winning teams.

When comparing the winning and defeated finalist teams, only five tendency lines appear ascending. Four of them are related to the winning teams and showed that even with the greatest difficulty found in the games, the winning teams were able to evolve in four key areas during the competition, different from the teams that were defeated, who only evolved in one of the variables of success of the game.

Of the 21 tendency lines considered valid for this study, there is clear evidence that the success of the team’s offensive actions during the competition will decrease, lowering efficiency and the number of goals scored, due to the increase in difficulties encountered in the course of the competition passing from the quarterfinals to the final. Furthermore, those 16 tendency lines are descending lines, which means that teams have reduced their ability to develop the key indicators of the offensive game process due to the increased difficulty in the games and the need to develop more challenging attacking solutions. Winning teams are more regular and do not have a tendency line with such an accentuated decline, especially for the key variables of the game that differentiate the winning teams from the losers.

5. Conclusions

First of all, linear tendency lines are useful in understanding team dynamics. Therefore, in comparison with the finalists, it was verified that assists and goals, and effectiveness from 6 m are key success indicators of winning teams, being the only variables where teams maintained an ascending line throughout the final phase of the competition. Furthermore, goal effectiveness, organized attack effectiveness, and effectiveness from 9–6 m shots were represented by decreasing tendency lines, with winning teams showing lines close to a straight line, demonstrating great regularity throughout the competition in these key performance indicator areas. As they evolve in the competition, teams tend to decrease in attack effectiveness and have more difficulty in overcoming the opposing defense, with winning teams needing to adapt to the new difficulties encountered in the game in each competitive phase, and adopting regularity in these success performance indicators differentiate the winners from the losers.

Due to the expected increased difficulty in successive matches, tendency lines tend to present decreasing slopes as teams advance toward the final. It can be concluded that winning teams have more regularity during the competition and experience an increase in key factors of the game such as assists, goals at 6 m, effectiveness in Type-I actions, and shots at 6 m, thus determining decisive factors for the success of the teams. Nevertheless, winning teams need to adapt to the increasing challenges, and growth in some key indicators is described as being more regular than in others.

The limitations of this study are due to the method used for the research and the need for a future, more advanced, and extended analysis of the winning teams throughout the entire competition. A further consideration is that the data are from before the important change in rules regarding the substitutions of the 7th player; this study has defiantly less significance now than before the rule change since the game has evolved, especially at the top level. This is a clear limitation of the study that should be mentioned.

For future research, it is also necessary to look at the “Psychological Theory” to explain changes between the various stages of the competition and the performance of the teams. In this case, we consider that the “Ecological Dynamics Framework” [38,39,40] may be able to give a better explanation of how athletes’ behavior varies according to opponents’ actions, the coaches’ options, the technical and tactical actions, the context in which the competition takes place, the restrictions imposed by the game itself, and many other factors. Additionally, it is also important to analyze, in future research, the different performances and the great variability in behaviors that emerge from opponent’s actions, which can be different according to the dynamics and the technical and tactical aspects of the teams, especially in the final stages of the competition. It should be observed that the game of handball has an almost unrepeatable characteristic, where an action hardly ever happens twice in the same way (i.e., the nonlinear dynamics of the game), being also provided with a great intra and interindividual variability that may influence the behavioral dynamics of the teams. This aspect thus deserves to be further explored through a more “harmonious” approach, crossing linear statistics with nonlinear techniques (e.g., approximate entropy, dimension correlation, and Lyapunov exponent, among others). Hence, coaches must understand the purpose of their training plan and the difficulties imposed by the opponents.

A further limitation of this study that should be corrected in the future is the fact that different context variables were not used (e.g., type of defense, values of the opponent, venue, home advantage, etc.), which can offer a better interpretation of the results, increase the knowledge of the performance tendencies of the teams and the individuals, and collective behavior changes through space and time. This type of procedure would also allow us to better understand how the linear tendencies can be adjusted to different phases of the competition (e.g., previous matches and final) in “winning” and “losing” teams, which in this study can be seen as a limitation. Finally, another limitation that deserves further study concerns the fact that this study did not include any association of interdependence made due to the high level of competition, something that could have restricted the results obtained.