VOLATILITY MODELS AND SDPP PREDICTION BASED ON CONDITIONAL MEAN AND VARIANCE
Conditional mean and variance random variables are the main components of SDPP (Standard Deviation Premium-Principle) risk measure that can be used to quantify the risk (loss). In this Final Project, the calculation of conditional mean and variance with its predictors are analyzed through two kinds...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/27005 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Conditional mean and variance random variables are the main components of SDPP (Standard Deviation Premium-Principle) risk measure that can be used to quantify the risk (loss). In this Final Project, the calculation of conditional mean and variance with its predictors are analyzed through two kinds of illustrations, illustration on dependence of two random variables and stochastic processes. Moreover, comparison of empirical properties, such as the kurtosis and autocorrelation function, in the GARCH(1,1) and SVAR(1) volatility models are investigated to accommodate the characteristics of data loss, which have heavy-tailed distribution and volatility clustering. Based on simulation results, it is shown that the SVAR(1) model is better in accommodate the heavy-tailed characteristic, while the GARCH(1,1) model is better in accommodate the volatility clustering characteristic. Furthermore, in this Final Project, the formulation of SDPP representation as a function of kurtosis in the GARCH(1,1) and SVAR(1) models are derived. Through the simulation on real and generated data, it is shown that the SDPP prediction value of the SVAR(1) model is greater than the GARCH(1,1) model. |
|---|