⚽ FIFA 2026 Winner Prediction using XGBoost and ELO Ratings

Overview

This project predicts FIFA World Cup 2026 match outcomes and potential tournament winners using Machine Learning, ELO ratings, team form statistics, head-to-head records, and historical international football match data.

The system collects historical match results, engineers predictive features, trains an XGBoost classifier, and generates predictions for FIFA World Cup 2026 fixtures.

---

🚀 Features

• Historical international football match analysis
• Dynamic ELO rating calculation
• Team form analysis
• Goal scoring and conceding statistics
• Head-to-head performance tracking
• Tournament importance weighting
• XGBoost-based match outcome prediction
• FIFA World Cup 2026 fixture simulation
• Probability-based match forecasts

---

📂 Project Structure

FIFA-2026-Winner-Prediction/
│
├── data/
│   ├── match_results.csv
│   ├── elo_ratings.csv
│   ├── current_elo.csv
│   ├── 2026_teams.csv
│   ├── features_train.csv
│   ├── features_test.csv
│   ├── features_2026.csv
│   └── 2026_predictions.csv
│
├── notebooks/
│   ├── collect_data.ipynb
│   ├── feature_engineering.ipynb
│   └── model_training.ipynb
│
├── README.md
├── requirements.txt
└── .gitignore

---

📊 Dataset

Historical Match Results

The dataset contains:

• International football matches
• Match dates
• Home and away teams
• Scores
• Tournament information
• Neutral venue information

ELO Ratings

ELO ratings are used to estimate team strength over time and are updated after each match based on actual and expected results.

---

🔧 Feature Engineering

The model uses the following features:

ELO Features

• Home ELO
• Away ELO
• ELO Difference
• ELO Win Probability

Team Form Features

• Home Team Recent Form
• Away Team Recent Form
• Form Difference

Goal Statistics

• Average Goals Scored
• Average Goals Conceded
• Expected Goal Difference (XGD)

Head-to-Head Features

• Home Team Win Rate
• Historical Draw Rate

Match Context Features

• Tournament Weight
• Neutral Venue Indicator
• World Cup Indicator

---

🤖 Machine Learning Model

Algorithm

XGBoost Classifier

Prediction Classes

Class	Meaning
0	Away Win
1	Draw
2	Home Win

Model Configuration

• 200 Estimators
• Maximum Tree Depth: 6
• Learning Rate: 0.1
• Subsample: 0.8
• Feature Subsample: 0.8

---

🔄 Workflow

Historical Match Data
          ↓
ELO Calculation
          ↓
Feature Engineering
          ↓
Train/Test Split
          ↓
XGBoost Training
          ↓
Model Evaluation
          ↓
2026 Fixture Prediction
          ↓
World Cup Simulation

---

⚙️ Installation

Clone the repository:

git clone https://github.com/your-username/FIFA-2026-Winner-Prediction.git
cd FIFA-2026-Winner-Prediction

Install dependencies:

pip install -r requirements.txt

---

▶️ Running the Project

Step 1: Data Collection

Run:

collect_data.ipynb

Generates:

• match_results.csv
• elo_ratings.csv
• current_elo.csv
• 2026_teams.csv

Step 2: Feature Engineering

Run:

feature_engineering.ipynb

Generates:

• features_train.csv
• features_test.csv
• features_2026.csv

Step 3: Model Training

Run:

model_training.ipynb

Outputs:

• Model Accuracy
• Classification Report
• Feature Importance
• 2026 Match Predictions

---

📈 Example Prediction

Argentina vs Mexico

Home Win Probability : 68.4%
Draw Probability     : 18.2%
Away Win Probability : 13.4%

Predicted Result: Argentina Win

---

🔮 Future Improvements

• Streamlit Web Application
• Interactive Tournament Simulator
• Knockout Stage Simulation
• Monte Carlo Tournament Forecasting
• Real-Time FIFA Ranking Updates
• REST API Integration
• Mobile Application

---

🛠 Technologies Used

• Python
• Pandas
• NumPy
• Scikit-Learn
• XGBoost
• Matplotlib
• Jupyter Notebook

---

👨‍💻 Author

Protyay Saha

---

📜 License

This project is licensed under the MIT License.