This is a group project of SDSC6013 Topics in Financial Engineering and Technology at City University of Hong Kong (CityU). We built a value-at-risk model directly modeling the quantile return directly by referring the paper CAViaR: Conditional Autoregressive Value at Risk by Regression Quantiles by Engle and Manganelli (2004).
As I found the original optimization approach is computational costly, I have modified a bit the box constraints as well as the starting approach (the initial guess/start of the estimated parameters). For details, you may want to take a look on the documentation (you may also easily change the setting back accordingly in the source code) or the presentation.pdf about the experiments and results. Although the differences are insignificant, please use the package caviar with caution.
You are welcome to report bug in https://github.com/yatshunlee/CAViaR-Project/issues. :)
We constructed two libraries: caviar and var_tests to model the value at risk and backtest the VaR estimate. For presentation, we constructed a dashboard to showcase how it can be possibly applied. If you are looking for some inspiration, we strongly suggest to take a look of the documentation below and the code in notebook-example.
Libraries:
- CAViaR Model
caviar - VaR Test
var_tests
Demo Application:
Documentation:
Main Author(s) / Maintainer(s) of caviar and var_tests:
- Jasper Lee