Capitalizing on market inefficiencies with neural networks: a case study in football betting

• Autores: Arana, Joaquín Matías
• Año de publicación: 2024
• Idioma: inglés
• Tipo de recurso: tesis de grado
• Colaborador/a o director/a de tesis: Gómez Seeber, Matías José
• Institución: Universidad de San Andrés

Fil: Arana, Joaquín Matías. Universidad de San Andrés. Departamento de Economía; Argentina.
This paper introduces a machine learning approach for profiting from sports betting markets, which are non-arbitraged inefficient markets that present opportunities for exploiting discrepancies between bookmaker odds and actual probabilities of outcomes. I will develop a Neural Network Model that accurately forecasts the events of two different football competitions: The English Premier League and The Argentine’s First Division. A betting strategy based on Markow’s portfolio theory is developed to exploit inconsistencies and maximize yield over time. The findings suggest that machine learning algorithms, like the one presented in this study, could potentially enable consistent profitability in sports betting by identifying inefficiencies in bookmaker odds.