Associação de indicadores do futebol com os resultados das partidas da Premier League 2015/2016

Carregando...
Imagem de Miniatura

Data

2018-02-19

Autores

Lima, Eric Matheus Rocha [UNESP]

Título da Revista

ISSN da Revista

Título de Volume

Editor

Universidade Estadual Paulista (Unesp)

Resumo

Além de ser uma modalidade esportiva que envolve relações de cooperação e oposição, o futebol também é um ambiente de rica disponibilidade e aplicação de informações acerca do que ocorre nas partidas. Ao encontro disso, o presente trabalho centra inquietações em busca do entendimento das ocorrências do jogo de futebol e da relação destas com o resultado final das partidas. Dessa forma, o objetivo do estudo consiste em apontar indicadores e variáveis que apresentam maior influência nos resultados das partidas de futebol. A relevância do estudo pauta-se no fato de permitir a análise de vários indicadores (posse de bola, passes, finalizações, entre outros) da modalidade ao mesmo tempo, permitindo a identificação de alternativas que aumentem as chances de êxito. O estudo se caracteriza como uma pesquisa quantitativa, de caráter descritivo. Para atingir os objetivos, será utilizada como amostra os dados estatísticos referentes às 380 partidas da temporada 2015/2016 da Premier League, o campeonato Inglês, dados estes coletados junto ao website do jornal Inglês “Daily Mail”. Também foram coletados dados referentes aos valores de mercado dos jogadores das equipes, obtidos no website do “Transfermarkt”, especializado em determinação financeira de elencos. Como ferramentas auxiliares para o processo de análise dos dados coletados serão utilizados alguns recursos computacionais, mais especificamente, a planilha eletrônica e o aplicativo Weka, um programa que implementa vários algoritmos da área de mineração de dados. Como resultados iniciais das análises, os números mostram que melhor conversão do total de finalizações em gols leva a uma chance superior de vitórias à conversão das finalizações certas em gols. Ao comparar apenas o total de finalizações e o total de finalizações certas, acertar mais a meta demonstrou maior probabilidade de êxito. Ter mais posse de bola e/ou realizar/acertar mais passes não aumenta a possibilidade de vitórias, a não ser que estes indicadores estejam relacionados a uma superioridade nas finalizações certas, mesmo que não levem ao gol. No entanto, estas mesmas finalizações certas mostram-se mais benéficas que a posse de bola e/ou os passes, para vencer as partidas. Além disso, ter maior percentual de passes longos e finalizar mais vezes na meta apresentou 71% de chances de triunfo. Deve-se observar também que realizar mais cruzamentos e escanteios não se traduz em grande probabilidade de vitórias. Por fim, apresentar elevado aproveitamento de gols nas finalizações certas, mesmo recebendo mais cartões amarelos, demonstrou 70% de probabilidade de vitórias. Esta dissertação oferece uma inovação metodológica que permite a identificação de alternativas aplicáveis para aumentar as chances de vitória, acrescenta conhecimento ao entendimento de que não há uma única fórmula para o sucesso e pode auxiliar na atuação dos treinadores de futebol por subsidiar alternativas que melhor se ajustem aos recursos disponíveis.
Besides being a sport that involves cooperation and opposition relations, football is also an environment of rich informations availability and application about what occurs in the matches. To that, the present work focuses concerns looking for the understanding of the games occurrences and their relation with the final match results. In this way, the study’s aim consists in pointing indicators and variables that show higher influence on football matches results’ occurrence. The relevance of the study is based on allowing the analysis of many indicators (ball possession, passes, shots, among others) of the sport at the same time, allowing the alternatives’ identification to increase chances of success. The study is characterized as a quantitative research, of described character. To reach the aim proposed, will be used as the sample the statistic data from the total of 380 Premier League, (English Football League) matches in the season 2015/2016, data obtained through accesses to Daily Mail’s website (english periodic). The data related to the teams’ market value were collected in the Transfermarkt’ website, specialized in squads financial determination. As auxiliary instruments for the collected data analysis’ process, some computational resources will be used, specifically, a spreadsheet and the Weka app, a program that implements many algorithms of data mining area. As initial analysis’ results, numbers show that better conversion of the total shots in goals takes to a higher chance of winning in relation to a better conversion of shots on target in goals. Comparing only total shots and total shots on target, reach the target with more frequency has shown higher likelihood of success. Having more ball possession and/or making more accuracy passes do not increase victory’s possibility, unless those indicators are related to shots on goal superiority, even if they do not take the team to score a goal. However, those shots on target showed themselves as more beneficial than ball possession and/or passes in order to win the matches. Besides, having higher percentage of long passes and shooting on target more times presented 71% triumph chances. It also must be observed that making more crosses and corners do not bring big victories likelihood. Ultimately, showing high exploitation of goals in shots on target, even receiving more yellow cards, demonstrated 70% victories likelihood. This dissertation offers a methodological innovation that allows the identification of useful alternatives to increase winning chances, adds knowledge to the understanding that there is not a unique formula to have success and can help on football coach’s actions for subsidizing possibilities that better fit to the available resources.

Descrição

Palavras-chave

Futebol, Premier League, Análise de dados, Desempenho humano, Football, Data analysis, Human performance

Como citar