This study aims to solve the problems of strong subjectivity in evaluation and delayed iteration of solutions in the technical training of football players, and constructs an intelligent decision-making theoretical model through systematic literature analysis. The PRISMA framework was used to screen 32 core documents, and the three-level coding of grounded theory was used to refine the “data-knowledge” dual-driven architecture. The model integrates multimodal data perception, hybrid algorithm decision-making and two-way feedback mechanism, and innovatively proposes the “compound trigger-double loop iteration” paradigm. Empirical comparison shows that compared with the traditional model, the new framework has a 63% lower misjudgment rate (p < 0.01) and a 6.2-fold increase in decision-making response speed. Expert validity tests confirm its logical completeness (4.7/5) and practical feasibility (4.1/5). The study provides a theoretical framework and implementation path for the intelligent transformation of the football youth training system.