Comparing probabilistic predictive models applied to football

Autor(es)

Diniz, Marcio Alves ^[1] ; Izbicki, Rafael ^[1] ; Lopes, Danilo ^[1] ; Salasar, Luis Ernesto ^[1]

Afiliação do(s) autor(es)

^[1] Univ Fed Sao Carlos, Stat Dept, Sao Carlos, SP - Brazil

• Tipo de documento: Artigo Científico
• Fonte: Journal of the Operational Research Society; v. 70, n. 5, p. 770-782, MAY 4 2019.
• Citações Web of Science: 0

Resumo

We propose two Bayesian multinomial-Dirichlet models to predict the final outcome of football (soccer) matches and compare them to three well-known models regarding their predictive power. All the models predicted the full-time results of 1710 matches of the first division of the Brazilian football championship and the comparison used three proper scoring rules, the proportion of errors and a calibration assessment. We also provide a goodness of fit measure. Our results show that multinomial-Dirichlet models are not only competitive with standard approaches, but they are also well calibrated and present reasonable goodness of fit. (AU)

• Processo FAPESP: 14/25302-2 - Uma abordagem flexível para a estimação de uma densidade condicional em problemas com alta dimensionalidade
• Beneficiário: Rafael Izbicki
• Modalidade de apoio: Auxílio à Pesquisa - Regular
• Processo FAPESP: 17/03363-8 - Interpretabilidade e eficiência em testes de hipótese
• Beneficiário: Rafael Izbicki
• Modalidade de apoio: Auxílio à Pesquisa - Regular