jingmouren/Chamkang6093-Levy-ARMA-GARCH_Risk_Model

Mainly modeled asset returns with ARMA-GARCH model and convolution-closed distributions, e.g., Generalized Hyperbolic or Normal Inverse Gaussian, to get 1-step forecasted p.d.f.s of returns.

Lévy - ARMA - GARCH Risk Model Implementation and its Evaluation

Descriptions

• Modeled asset returns with ARMA-GARCH model and convolution-closed distributions, e.g., Generalized Hyperbolic or Normal Inverse Gaussian, to get 1-step forecasted p.d.f.s of returns.
• Evaluated the effectiveness of this model by some risk measures like CVaR and coverage tests.

Notes

• _snp_tickers_sectors.csv  is found at https://en.wikipedia.org/wiki/List_of_S%26P_500_companies (In this project, last change was on April 21, 2021).
• In the part of tests, the code about coverage tests is mainly from the library named "vartests" (Link: https://github.com/rafa-rod/vartests), when some minor modifications were made.
• In the part of fitting Lévy - ARMA - GARCH model, some lines of the code are omitted due to potential copyright issue, where some of them are from private code of our professor and his assistant. You can try to incorporate NIG and GH distributions into some other open-source ARMA-GARCH code as an alternative if you are interested in this part. Sorry for the incovenience.

Main Reference

• https://econpapers.repec.org/article/taflstaxx/v_3a47_3ay_3a2018_3ai_3a15_3ap_3a3616-3639.htm
• https://www.semanticscholar.org/paper/Backtesting-Value-at-Risk-Models-Nieppola/5fbbf20d7f3e03c4da3a9d33aa618fdfe5d0a9d4
• https://link.springer.com/book/10.1007/978-0-387-77827-3