The research conducted by Luis Sánchez, Ph.D., an engineer, professor, and researcher at ESPOL’s Faculty of Geosciences Engineering (FICT), was published in the scientific journal Array, which is indexed in the Q1 quartile—one of the highest levels of impact and international visibility in scientific communication. The study proposes a reproducible methodology for pre-match forecasting of secondary variables in professional soccer, using matches from Spain’s LaLiga as a case study.
The paper, titled “A leakage-aware workflow for pre-match forecasting of secondary soccer markets: A LaLiga case study,” addresses one of the main challenges of sports analytics based on artificial intelligence and machine learning: ensuring that predictive models use only information available before the start of a sporting event. To this end, it incorporates mechanisms for temporal control, sequential validation, and reconciliation of data from multiple sources, with the aim of avoiding biases that could affect the reliability of the results.
The research analyzes variables such as total shots, shots on goal, corner kicks, cards, and fouls—indicators that provide a deeper understanding of teams’ tactical and performance-related behavior beyond the final score. The study evaluated 760 LaLiga matches played between February 2024 and March 2026 using a chronological validation protocol that replicates real-world sports forecasting conditions.
The results show systematic improvements over traditional heuristic models across all analyzed variables. Furthermore, the methodology demonstrated competitive performance compared to benchmark parametric models in markets such as shots, shots on goal, and corner kicks, while maintaining a prudent scientific stance by acknowledging that predictive performance continues to depend on the specific characteristics of each variable studied.
Beyond predictive performance, the main contribution of this work lies in the development of a transparent and reproducible methodological framework for applied research in data science. The study demonstrates that aspects such as the temporal integrity of the data, the correct identification of entities, and consistency in validation processes are fundamental to developing reliable models in complex and dynamic environments such as professional sports.
This publication strengthens ESPOL’s scientific output in the areas of artificial intelligence, data science, and sports analytics, and contributes to the development of methodologies that can be applied to other prediction problems in various fields where data quality and reliability are critical.
DOI: https://doi.org/10.1016/j.array.2026.101039
Sustainable Development Goals (SDGs)
SDG 9: Industry, Innovation, and Infrastructure, through the development of innovative and reproducible methodologies for data-driven predictive analysis.
