An intelligent predictive model for low-scoring competitive games

The aim of this thesis is to design an intelligent predictive model for low-scoring competitive games. More specifically, this model integrates classical mathematical approaches with advanced computational intelligence techniques to leverage historical datasets for more accurate predictions. The thesis is divided into four main parts: The first section provides an in-depth introduction to the theoretical background, covering essential concepts in game theory and reviewing the most prominent models in the existing literature. This part lays the foundation to better understanding the challenges and methodologies used in forecasting low-scoring outcomes. The second part focuses on the creation and the refinement of the dataset. It details the process of data collection, the criteria for selecting relevant features and the methodologies used to ensure the dataset is robust and representative. The third section is dedicated to the development of the predictive model. It covers the introduction of new features, the selection of the most appropriate model and the optimization of its parameters; some of which are taken from the literature of existing models, while others are optimized by original research and experimentation. The final section applies the model to various datasets, evaluating its performance using a range of metrics. The results are then analyzed, comparing the model’s predictions to actual outcomes. The thesis concludes with a discussion on the model’s effectiveness, potential limitations, and suggestions for future improvements and extensions.