🏆 FIFA World Cup 2026 — Winner Prediction Model

A Machine Learning project that predicts the winner of the 2026 FIFA World Cup using historical international football data, FIFA rankings, and tournament simulations.

---

📌 Project Overview

This project builds a complete data science pipeline to predict which team is most likely to win the 2026 FIFA World Cup hosted by USA, Canada & Mexico. It uses historical match data going back to 1872, engineers meaningful features, trains multiple ML models, and simulates the entire 48-team tournament thousands of times using Monte Carlo simulation.

---

📂 Project Structure

Fifa-world-cup-2026/
│
├── worldcup_2026_prediction.ipynb   # Main Jupyter Notebook (full pipeline)
├── results.csv                      # Historical international match results
├── fifa_ranking.csv                 # FIFA world rankings data
│
├── feature_importance.png           # Feature importance chart
├── model_comparison.png             # ML model accuracy comparison
├── outcome_distribution.png         # Win/Draw/Loss distribution
├── goals_over_time.png              # Average goals per match over time
├── top_teams_wins.png               # Top teams by all-time wins
├── wc_wins_by_country.png           # World Cup wins by country
├── tournament_types.png             # Match types breakdown
└── win_probabilities.png            # Final predicted win probabilities

---

🔬 Methodology

1. Data Collection

• Historical international football results (1872–2026)
• FIFA World Rankings (1992–2026)

2. Exploratory Data Analysis (EDA)

• Match outcome distributions (Home Win / Draw / Away Win)
• Top teams by all-time wins
• Goals scored trends over decades
• World Cup performance by country

3. Feature Engineering

Feature	Description
home_win_rate	Win rate of home team in last 10 matches
away_win_rate	Win rate of away team in last 10 matches
home_goals_scored	Average goals scored by home team
home_goals_conceded	Average goals conceded by home team
away_goals_scored	Average goals scored by away team
away_goals_conceded	Average goals conceded by away team
h2h_home_win_rate	Head-to-head win rate between the two teams
win_rate_diff	Difference in win rates
goals_diff	Difference in average goals
is_neutral	Whether match is played on neutral ground

4. ML Models Trained

• ✅ Logistic Regression
• ✅ Random Forest Classifier
• ✅ XGBoost Classifier

5. Tournament Simulation

• Full 48-team group stage simulation (12 groups × 4 teams)
• Knockout rounds: Round of 32 → Round of 16 → QF → SF → Final
• Monte Carlo Simulation (1000+ runs) to calculate win probabilities

---

📊 Results & Visualizations

Model Accuracy Comparison

Feature Importance

Predicted Win Probabilities

World Cup Wins by Country (Historical)

---

🚀 How to Run

Prerequisites

pip install pandas numpy matplotlib seaborn scikit-learn xgboost jupyter

Steps

1. Clone the repository:

git clone https://github.com/Sourik-10/Fifa-world-cup-2026.git
cd Fifa-world-cup-2026

2. Launch Jupyter Notebook:

jupyter notebook worldcup_2026_prediction.ipynb

3. Run all cells from top to bottom ✅

---

📦 Data Sources

• International Football Results (Kaggle) — 47,000+ matches
• FIFA World Rankings (Kaggle) — Rankings since 1992

---

🛠️ Tech Stack

---

🔮 Future Improvements

• Add real FIFA Rankings as a live feature
• Include player-level stats (EA FC 25 ratings / Transfermarkt values)
• Use Elo rating system instead of simple win rates
• Increase Monte Carlo simulations to 10,000+
• Update predictions as real tournament matches are played
• Deploy as a web app using Streamlit

---

👤 Author

Sourik — @Sourik-10

---

⭐ Show Your Support

If you found this project useful, please give it a star ⭐ on GitHub!

---

Made with ❤️ and lots of football passion 🏟️