QUANTILE REGRESSION MODEL FOR VALUE-AT-RISK PREDICTION

• Main Author: Dian Fahira, Audrey
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/50074
• Institution: Institut Teknologi Bandung

The quantile regression model is a modern regression model that appears as an alternative to fulfilling the assumptions that failed to be obtained in the classical regression model. This model is able to model the entire distribution of the data. This feature is obtained because the quantile regression model defines the dependent variable as a quantile function. The quantile that defines Value-at-Risk can be a tool to quantify the possible risks that occur. The quantile is the bridge of the relationship between the quantile regression model and the Value-at-Risk concept so that Valueat- Risk predictions can be made using the quantile regression model. The quantile regression model was able to model a lot of data including the type of data with a thick tail distribution. This form of data distribution is one of the characteristics of financial data. Therefore, the prediction of Value-at-Risk using quantile regression model will be applied to financial data. Various quantile regression models can be applied to model financial data, but the ones that will be used are the QAR (Quantile Autoregression), CAViaR (Conditional Autoregressive Value-at-Risk), and HARQREG ( Heterogenous Autoregressive - Quantile Regression Model).