A machine learning project that predicts the results of the 2026 FIFA World Cup — including group stage standings, knockout bracket, and overall winner probabilities.
- Predicts the outcome of any match between two international teams
- Simulates the full 2026 World Cup group stage and knockout bracket
- Runs 10,000 Monte Carlo simulations to generate win probabilities for every team
- Displays results in an interactive Streamlit dashboard
- 44,000+ international match results (1990–2024)
- Historical FIFA World Rankings
- Features engineered from rolling form, head-to-head records, and ranking gaps
- Three models trained and compared: Logistic Regression, Random Forest, XGBoost
- XGBoost selected as the final model (best accuracy on held-out test set)
- Predicts match outcome probabilities: home win / draw / away win
- Group stage: round-robin, top 2 teams qualify
- Knockout rounds: Round of 24 → QF → SF → Final
- Monte Carlo: 10,000 full tournament simulations to produce win %
- Python, Pandas, NumPy
- Scikit-learn, XGBoost
- Streamlit
- Google Colab (development environment)
git clone https://github.com/Diiv152/fifa-predictor.git
cd fifa-predictor
pip install -r requirements.txt
streamlit run app.pyBuilt by Diivyansh as a self-directed ML learning project.