I Built 11 Models to Predict the 2026 World Cup. They Crown Four Different Champions.

Eleven different models were built to predict the 2026 World Cup champion, using a shared dataset of 358 historical matches. The models included rating systems like Elo and PageRank, goal-based models like Poisson, and classifiers such as logistic regression, XGBoost, and a neural network. Each model was fed into a common tournament simulator to generate thousands of outcomes. The primary finding is that the models disagree, crowning four different champions and demonstrating the importance of understanding model variance rather than trusting a single answer.