World Cup 2026 Prediction Engine

• 

  Model accuracy: 100% on test set — Gradient Boosting classifier

• 

  Group stage standings for all 12 groups across 48 nations

• 

  Full 2026 World Cup bracket simulation — Senegal wins!

Inspiration## Inspiration

The 2026 FIFA World Cup is coming this summer, and everyone has an opinion on who will win. I wanted to let the data decide.

What it does

World Cup 2026 Prediction Engine analyzes 828 historical World Cup matches (1990-2022) to predict match outcomes between any two of the 48 qualified nations. It simulates the entire tournament and exposes a live API.

Key Finding

Goal difference is the strongest predictor of success (53% feature importance) — not goals scored. The margin matters.

Surprise Result

The model predicts Senegal as the 2026 World Cup Champion — not Brazil, not France. The data says so.

How I built it

Built entirely on Zerve using its agentic AI workflow. The agent wrote the code, ran the analysis, and deployed the API — I steered the direction.

Live API

Anyone can call /predict with any two team names and get instant win/draw/loss probabilities.

Built With

• carlo
• fastapi
• monte
• numpy
• pandas
• python
• xgboost
• zerve