Previsão de resultados de jogos de voleibol utilizando regressão logística

Abstract

The aim of the present study was to find logistic regression models to predict the probability of a volleyball team winning a match in terms of factors such as match location (home, away or neutral), position in the standings, score difference in the sets and winning two sets first were studied. Matches from two distinct competitions were assessed: 132 from Superliga Nacional de Voleibol, which comprises teams form clubs and 82 from the Volleyball World League 2016, comprising international squads. Regarding the matches among clubs, two cases were analysed: the victory probability of a team which won two sets first and the victory probability when a team is playing home games. In the third case, the victory probability of a team which won two sets first was also studied, but in this case, only games which took place in a neutral venue regarding international squads were taken into account. The results showed that the difference in points in the first set won was not significant in any of the three analysed cases. However, the difference in points in the second set won came across as a significant factor in the second case. Evidences were also found that holding a better place in the standings than the opponent influences in the winning probability of the team clubs. However, the same factor is not significant in international squads matches. It was observed that winning two sets first increases the probability of winning a match and that the probability is very similar when these two sets were won by the home team(84%) or the away team(83%). It was possible to conclude that the variables utilised in this study are not enough to clearly determine the odds of winning a match suggesting that other factors should be added to future studies. However, the results related to winning two sets first and the difference in the standings between the teams might be seen as valuable information to take decisions before or during a match.