Sports prediction models are a vital area of sports analytics that uses statistical modeling, data mining, and machine learning to predict results and guide decision-making in the sports industry. Any kind of coach, analyst, bettor or fan can benefit from the football, handball, cricket, tennis, basketball forecasting abilities. However, the literature considered indicates a methodology progression, and the use of traditional statistical models, such as Logistic Regression, Poisson Regression, Gaussian approximation, etc. was superseded by the use of modern artificial intelligence techniques, such as adaptive back-propagation neural networks, ensemble learning methods, and “explainable AI integrated with deep learning”. Conclusions indicate that, in distinct sports, different information is relevant: in football defensive efficiency and shooting accuracy is dominating, whereas in cricket overs and wickets influence game outcomes, in tennis game performance predicting momentum is important and in basketball the recent number of won games proves very useful. In the meantime, studies in handball are growing rapidly and include the use of a sensor-based inertial measurement unit (IMU) and computer vision techniques. This paper synthesizes contributions from 2016 to 2025, identifies methodological strengths and limitations, and proposes future directions focusing on generalizable models, multi-modal data integration, and transparent explainable systems.