ANALYSIS OF THE BRAZILIAN CHAMPIONSHIP FIRST DIVISION PERFORMANCE BETWEEN 2003 AND 2014

ANÁLISE DO DESEMPENHO DO CAMPEONATO BRASILEIRO DE PRIMEIRA DIVISÃO ENTRE 2003 E 2014

ABSTRACT

The aim of this paper is to evaluate the performance of teams in the first division of brazilian soccer, based on data of the brazilian championship, between 2003 and 2014. The multivariate technique used for the analysis was the construction of a measure of dissimilarity and similarity as well as the analysis of some indicators. The main results show that there was a large concentration of high performances in clubs of south and southeast regions. Furthermore, it was observed that the São Paulo Futebol Clube was the team with better performance in the period.

Keywords:
Performance; Analysis; Similarity; Dissimilarity; Soccer; Brazilian championship.

RESUMO

O objetivo do presente artigo é avaliar o desempenho dos times da primeira divisão do futebol brasileiro, com base em dados do campeonato brasileiro, no período entre 2003 e 2014. A técnica multivariada utilizada para a análise foi a construção de uma medida de dissimilaridade e de similaridade, bem como a análise de alguns indicadores. Como principais resultados observou-se uma grande concentração de desempenhos acima da média em clubes das regiões sul e sudeste. Além disso, observou-se que o São Paulo Futebol Clube foi o time com melhor desempenho no período de pontos corridos.

Palavras-chave:
Desempenho; Análise; Similaridade; Dissimilaridade; Futebol; Campeonato brasileiro.

Introduction

Football has a special magic for most of the people. In Brazil, the magic of the ball reaches dimensions which are difficult to explain¹. The Brazilian championship is one of the most balanced in the world, with many teams in theoretical conditions (technical) to fight for the title, according to what can be seen in the results of the researched years. In the last years, more intensively, there was a further movement of football towards the gym and the science. In this way, more and more researchers dedicate themselves to study football, in its multiple facets².

Many studies have been carried out in order to try and assess or try to foresee the team performances in competitions. Among them we can mention the papers by Anon et al³, Marcelino Sampaio and Mesquita⁴, De Araujo et al⁵, as being focused on the use of statistical methods for predictions. The paper by Almeida, Oliveira and Silva⁶, has brought an interesting analysis about teams hosting the matches in series A (first division) and B (second division), having found a clear advantage in hosting the match with greater impact in series B of championship. Likewise, the papers by Hass⁷ and Gómes e Mendo⁸ also aimed at assessing the impact of hosting the match and the effective support of the crowd in the result. It is a relevant approach, since in certain conditions, the hosting can be a complicating factor, mainly if the team does not have a good relationship with the crowd at the moment of the game. More specific factors, such as possession and shots, for instance, can explain the result of a team^9-11. However, the paper by Carvalho, Scaglia e Costa¹², provides and interesting analysis of tactical performance and final results (tie, victory and defeat), finding significant differences and possible justifications between tactical performance and final result.

The performance forecast could help sponsors in decision taking, such as choosing the team to invest their money in. It could also influence players decisions concerning the choice of teams with potential to reach first positions. The aim of this article is to assess first division teams of Brazilian football performances, based on results obtained in first division Brazilian championships, within 2003 and 2014, through indicators that can mirror the teams performances.

Methods

Data used in this paper were obtained from the site Futdados¹³ and transferred to the software SPSS 21 Pro. Work was carried out with data from 2003 to 2014, since the current qualifying system came into being in 2003 in Brazilian championship. Calculations were performed only for first division championship. The multivariate technique used in the analysis was the development of a dissimilarity and similarity measure, as well as some indicators analysis.

The available variables for analysis were the number of points, matches, victories, ties, defeats, goals scored for and against, goal difference and number of times the team has participated in the championship (in the first division) over the period considered.

In order to facilitate analysis, four performance indicators were created: the relation between the number of wins and the total of matches (D1), the relation between the number of ties and the total of matches (D2), the relation between the number of defeats and the total of matches (D3) and the utilization rate (D4), defined as the complement of total of defeats divided by the sum of ties and wins. The creation of the indicators came up to standardize data, thus the number of editions of each team, the number of players, among other aspects, would not be influenced by a greater or lesser participation in terms of number of championships in the first division of Brazilian championship. Therefore, based on table 1 the four indicators were created to calculate the similarity and dissimilarity measure 41 clubs were included in the samples altogether.

Dissimilarity Measure

The Euclidian distance was used as dissimilarity measure. The Euclidian distance for individuals i and j, using p parameters, is obtained by the formula ¹⁴:

In the present paper, i and j are the taken clubs. It is noted that in this case the template reaches a differentiated formula, as we are dealing with multiple analysis (41 clubs were considered altogether). Concerning the parameters used, they were four: D1, D2, D3 and D4, defined previously.

Similarity Measure

As for the similarity, it can be measured by Pearson correlation coefficient, which can be calculated by the formula ¹⁵:

where

and

are the measured values of both variables, and

are the arithmetic means of both variables

In the case of the current study, it is a correlation measure among the variables considered, for the considered parameters (D1, D2, D3 and D4). The analysis of table 1 gives provides an idea of the concentration of teams, with better performance, in the South (26,83% of the total) and Southeast (43,90%). Both regions make 71,73% of the teams which disputed the first division of Brazilian championship in the period considered. However, the central - west region (7,32%), northeast (19.51%) and North (2,44%) had a quite smaller participation.

Table 1
Championship data

Table 2
Indicators of Effectiveness and Ineffectiveness

Results and Discussions

The analysis of Table 1 shows that 6 out of 41 clubs analysed: São Paulo, Cruzeiro, Internacional, Santos, Fluminense and Flamengo have participated in 13 editions of Brazilian championship through qualifying system, considered. On the other hand, Corinthians, Grêmio, Atlético Mineiro and Atlético Paranaense have taken part in 12 editions. Another aspect that calls attention is that 71% of the clubs considered in the analysis belong to the South and Southeast regions of Brazil. In the case of considering clubs which have had 5 or more participations, in the first division, in the period analyzed, only Goiás, Vitória, Sport and Bahia would represent regions other than South / Southeast. Therefore, a major concentration, in terms of power, of clubs from the South and Southeast can be inferred, concerning the first division Brazilian championship.

Table 2 is interesting for all the analyses. First of all it represents the four indicators which were used to calculate the similarity and dissimilarity. Then, it makes the analyses fairer, as they are indexes and do not take into consideration the number of participation of each club, but the effectiveness in each analyzed aspect. the index D1 measures the relation between the number of wins and the total of matches. The index D2 measures the relation between the number of ties and the total of matches. The index D3 measures the relation between the number of defeats and the total of matches. And finally, index D4 is an exploitation measure, considering the total of wins and ties in relation to the number of matches.

Once points for wins and ties are scored the exploitation rate gives an idea of the effectiveness of the club in the search for points. It is clean that the victory must be the major objective, as it scores more points than ties (3 points against 1 point), but ultimately, depending on the situation, a tie can be considered a reasonable result. From the effectiveness point of view, in terms of results, the total amount of points (TP) can be obtained through the victory (V), tie (E) or defeat (D). Obviously, the defeat would be the less effective result, in terms of score for the championship, but it is a result that invariably occurs, even with teams that might become champions. As victory scores 3 points, we assume that the victory is the higher effectiveness index, the tie would be the second best effectiveness index and defeat would be the worst. This ways in terms of effectiveness, we would have an index of 1 for victory, 1/3 for tie and 0 for defeat.

Table 3 shows the effectiveness in terms of victory for number of matches. In column 1 we have the original classification, based on the number of points the club has in the 13 editions of the considered Brazilian championship. In column 2 we have the classification based on the percentage of wins in relation to the number of matches. In the 3 first positions there is no change in the columns order, but from position 4 on some changes are noticed. From the championship point of view, as whole, the second column is extremely important as it shows the teams effectiveness in terms of victory that, as previously observed, is where the highest score for the championship occurs. Evidently, the team that reaches the greatest consistency in terms of victory and, at the same time, the lowest number of defeats. If we look at the 10 first clubs, concerning the total number of points, through original classification (column 1), we can see that only Flamengo had an expressive fall in positions when comparing to column 2. In the case of Flamengo it was a fall of 7 positions. Grêmio and Corinthians had the larger positive variations, gaining 3 and 2 positions, respectively.

Table 3
Percentage of victories

When the percentage of ties is analyzed in relation to the total number of matches, in Table 4, we can see that, on the whole, the smaller clubs play more aiming the tie or show a lower potential for victory. Looking at the table we can see that among the 10 teams which have drawn the most, only Flamengo and Corinthians are placed, in the original classification, among the 10 teams with highest scores in Brazilian championship, within the considered period. From the point of view of match strategy, perhaps the option of tie, as tatics, implies in giving up the victory, avoiding the risk of defeat. However, depending on the aim of the club, according to its specificities, this can be a great strategy, in order to bitter use the available resources.

Table 4
Percentage of tié

The analysis of Table 5 makes the importance of trying to avoid defeat clear if we observe the relation between the original classification, with the teams that scored more points, and the column with the teams that had the highest number of defeats, there is an inverse relationship in terms of position. In this aspect São Paulo turns up to be more effective, followed by Corinthians. Since defeat does not score points, losing less matches is one of the requisites to persue a comfortable position in a championship for points, where regularity is very important, especially when the teams are very well technically balanced.

Table 5
Percentage of defeat

Table 6 measures the exploitation rate of each team. As defined previously, the index measures the complement for the division of the number of defeats (D) by the total of ties (E) and wins (V). This way, the exploitation rate is given by 1 - [ D / (E + V)].

As expected, the exploitation rate is one of the most effective rates to measure the club performance in a championship for points. The analysis of table 6 calls attention for the position of Corinthians, which jumps from the 6^th position in the original classification, considering the total of points, to the 2^nd position, when the exploitation rate is considered. Cruzeiro, on the other hand, shows an opposite pattern, having a sharp fall since it changes the 2^nd position in the original classification for the 7^th place, based on the exploitation rate.

Table 6
Percentage of exploitation

Similarity and Dissimilarity measures

The ultimate goal of this study is to analyze the performance of series A teams, in Brazilian championship, based on multivariate analysis measures. The measures chosen were the similarity and dissimilarity analysis. They are classification and grouping techniques. As dissimilarity measure Euclidian distance will be used and Pearson correlation coefficient will be used as similarity measure. Therefore, the bigger the dissimilarity measure, the further from the best performance the team will be. On the other hand, the bigger the similarity measure, the closer to the best performance the team will be.

Table 7 analysis shows some differences in classifications between similarity and dissimilarity measures. This may occur since they are techniques which use different methodology for calculating. However, no one can expect big changes when assessing both results. In all the analysis developed in this study, São Paulo Futebol Clube (SPFC) ranked first position, thus both the similarity and the dissimilarity measures were elaborated using SPFC as reference. It is important to emphasize that it was not the authors choice, but instead a result of the analysis concerning the position of the team.

Table 7
Measure of similarity and dissimilarity

Conclusions

The present article has developed an analysis of the teams performance in the first division of Brazilian football. Therefore, base on the available data regarding their performances, some additional indicators where created, aiming to have a relative comparison of their performances. The indicators created have helped with the analysis of effectiveness and ineffectiveness of the teams, in a objective way, throughout the period considered. Based on the same indicators, similarity, measured by Person linear correlation, as well as dissimilarity measured by Euclidian distance, were analyzed. On the whole, analysis have showed there is a concentration, in the South and Southeast regions, of clubs with better performance. Furthermore, we could observe the Brazilian championship in first division, is very disputed, with more than 10 teams with very similar performance (measured, for instance, by similarity). An interesting practical application of the present study would be to detect possible strategies to be adopted in order to adjust the team to the expected performance. This paper can also present an interesting insight regarding possible strategies concerning the team behavior (play for a draw or try the victory running the risk of not scoring points), according to what was verified based on the exploitation rate.

Finally, it is clear that, within the period considered and considering the points. São Paulo Futebol Clube had the best performance according to various criteria. What also calls attention is the performance of Sport Clube Corinthians Paulista, in terms of effectiveness, that can be checked both by its exploitation rate and the similarity measure, or even by the dissimilarity measure..

References

• ¹
  Valentin BR, Coelho M. Sobre as escolinhas de futebol: processo civilizador e práticas pedagógicas. Motriz Rev Edu Fis 2005;11(3):185-197.
• ²
  Reilly T, Gilbourne D. Science and football: a review of applied research in the football codes. J Sports Sci 2003;21(9):693-705.
• ³
  Añon IC, Yamanaka GK, Machado JC, Scaglia A. Performance da equipe da Espanha e seus adversários nos jogos da Copa do Mundo FIFA 2010. RBF 2013;06(1):33-44.
• ⁴
  Marcelino R, Sampaio J, Mesquita I. Investigação centrada na análise do jogo: da modelação estática à modelação dinâmica. Rev Port Cien Desp 2011;11(1):481-499.
• ⁵
  De Araujo CTP, Tavares L, Alvares LG, Neto FL, Suzuki AK. Modelagem estatística para a previsão de jogos de futebol: Uma aplicação no campeonato brasileiro de futebol 2014. Revista da Estatística UFOP 2015;4(2):12-20.
• ⁶
  Almeida LG, Oliveira ML, Silva CD. Uma análise da vantagem de jogar em casa nas duas principais divisões do futebol profissional brasileiro. Rev bras educ fís esporte 2011;25(1):49-54.
• ⁷
  Haas DJ. Productive Efficiency of English Football Teams - A Data Envelopment Analysis Approach. Manage Decis Econ 2003;24(5):403-410.
• ⁸
  Gómez R, Mendo H. Revisión de Indicadores de Rendimento en Fútbol. RICCAFD 2012;1(1):1-14.
• ⁹
  Lago-Ballesteros J, Lago-Peñas C. Performance in Team Sports: Identifying the Keys to Success in Soccer. J Hum Kinet 2010;25(1):85-91.
• ¹⁰
  Lago-Peñas C, Lago-Ballesteros J, Rey E. Differences in performance indicators between winning and losing teams in the UEFA Champions League. J Hum Kinet 2011;27(1):135-146.
• ¹¹
  Vázquez AV, Gayo AA, Pita HB, Fernández CA. Diseño y aplicación de una batería multidimensional de indicadores de rendimiento para evaluar la prestación competitiva en el fútbol de alto nivel. Int J Sport Sci 2011;7(23):103-112.
• ¹²
  Carvalho FM, Scaglia AJ, Costa IT. Influência do Desempenho Tático sobre o Resultado Final em Jogo Reduzido de Futebol. Rev Educ Fis UEM 2013;24(3):393-400.
• ¹³
  Futdados.com [Internet]. Pontos corridos: Campanhas acumuladas desde 2003 - Todas as equipes. [acesso em: 17 jun 2015]. Disponível em: Disponível em: http://futdados.com/pontos-corridos-campanhas-acumuladas/
  » http://futdados.com/pontos-corridos-campanhas-acumuladas/
• ¹⁴
  Dattorro J. Convex Optimization & Euclidean Distance Geometry. Palo Alto, Califórnia: Meboo Publishing; 2008.
• ¹⁵
  Hair Jr JF, Anderson RE, Tatham RL, Black WC. Multivariate Data Analysis. 5.ed. New Jersey: Prentice Hall; 1998.

Publication Dates

• Publication in this collection
  2016

History

• Received
  04 Oct 2015
• Reviewed
  15 May 2016
• Accepted
  02 June 2016