Forecasting English League Football Results: A Fuzzy Inference System approach

Many papers and articles have been written about forecasting the outcome of sporting events, various methods have been used, the predominant methods in use are statistical, with a smaller proportion using soft computing techniques (ie Neural Nets, Fuzzy Logic, adaptive networks, and Genetic algorithms). This book explores the creation of Fuzzy Inference Systems to forecast English Football league outcomes using Clustering techniques. It attempts to achieve at least the results obtained by Statistical techniques and Fuzzy Inference systems to date, both of which achieved 64% forecast performance. The training and test datasets were taken from 22 seasons of fixtures (1984 to 2005). A specific genetic optimisation procedure called 'radii competition' is used to achieve the best forecast performance. The data includes, 12 factors about each team, including historical and current facts to build a profile of each team. This profile enables the program to set relative strengths. The book succeeds in its goal by exceeding the performance achieved by previous statistical and fuzzy models by 10%, maximum forecast performance achieved: 74.21%.