Volatility Forecasting Using GARCH Model
Objective
Forecasting volatility in asset returns using the GARCH (Generalized Autoregressive Conditional Heteroskedasticity) model.
Data
- Historical stock prices of S&P 500 from 2010-01-01 to 2023-01-01.
Methodology
- Calculated daily returns from stock prices.
- Visualized returns and checked for volatility clustering.
- Implemented GARCH(1, 1) model for volatility forecasting.
- Backtested model using rolling window approach.
Results
- RMSE of GARCH(1, 1): 0.0058
- Value at Risk (95% confidence): 0.0205
Dependencies
- Python 3.x
- pandas, numpy, matplotlib, seaborn, yfinance, arch, sklearn, scipy, statsmodels
Usage
- Clone the repository.
- Install dependencies using the following command:
pip install -r requirements.txt - Run the Jupyter notebook.