This study proposes a comprehensive framework for Olympic medal prediction based on nation stratification. Countries are grouped into superpowers, strong, medium, and small tiers, each modeled with tailored methods. Random Forest effectively predicts superpower outcomes, while AHP and TOPSIS guide small nations in niche event investment. Causal inference techniques like PSM and DID assess the impact of elite coaching programs. Results show strategic planning can significantly enhance medal performance, especially for host and smaller nations. The framework supports adaptive, data-driven Olympic strategies for nations of varying scales.