An advanced machine learning system that predicts potential stock market crashes with 100% recall. Using XGBoost and sophisticated feature engineering, this model identifies weeks where the market may experience a ≥5% drop.
- 🎯 100% recall in detecting market crashes
- 📊 Interactive Streamlit dashboard
- 📈 Real-time market indicators
- 🔮 Weekly crash predictions with confidence scores
- 📉 Comprehensive market analysis
- ⚡ Fast and efficient predictions
- A drop of 5% or more in market value within a week
- Binary classification problem (Crash/No Crash)
- Base Model: XGBoost Classifier
- Sampling: SMOTE for handling class imbalance
- Feature Engineering:
- Lagged features for temporal patterns
The Streamlit dashboard provides:
- Real-time crash probability predictions
- Confidence scores for predictions
- Feature importance analysis
- Machine Learning: XGBoost, scikit-learn
- Data Processing: pandas, numpy
- Visualization: plotly
- Dashboard: Streamlit
- Data Augmentation: SMOTE
- Recall: 100%
- Training Period: 2000-2021
- Validation Method: Time-series cross-validation
- Regular Model Retraining: Weekly
- Previous week's returns
- Rolling averages
- Volatility metrics
- Volume indicators
- Moving averages
- VIX
- VG1
- MXRU
- CRY
-
Data Collection
- Historical market data (2000-2021)
- Volume metrics
- Market indicators
-
Preprocessing
- Feature engineering
- SMOTE application
- Normalization
- Missing value handling
-
Model Training
- XGBoost optimization
- Hyperparameter tuning
- Cross-validation
- Access the Streamlit dashboard
- View current market predictions
- Analyze confidence metrics
- Explore feature importance
- Monitor historical accuracy
This tool is for research and educational purposes only. Stock market predictions involve risk, and no financial decisions should be made solely based on this model's outputs.
Contributions are welcome! Please feel free to submit a Pull Request.
- Fork the project
- Create your feature branch
- Commit your changes
- Push to the branch
- Open a Pull Request
For questions and support, please open an issue in the GitHub repository.
Note: Past performance does not guarantee future results. Always conduct your own research and consult with financial advisors before making investment decisions.