8000 GitHub - VB4R0N/Thesis: Code Used in My Bachelor Thesis
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

VB4R0N/Thesis

Repository files navigation

📊 Overview

🎯 Objective

Evaluate and compare the predictive performance of different GARCH-type models in forecasting weekly volatility, using out-of-sample evaluation techniques.

🔍 Key Methodologies

  • 📈 ARCH/GARCH Modeling – Testing for ARCH effects and fitting GARCH, GARCHX (with exogenous variables), and EGARCH models.
  • ⚖ Model Selection – Using the Akaike Information Criterion (AIC) to identify the best in-sample fit.
  • 🔄 Forecasting Approach – Implementing a rolling window framework to generate dynamic volatility predictions.
  • 📊 Forecast Evaluation:
    • 📌 Realized Variance Approach – Weekly volatility forecasts were evaluated using realized variance, computed from daily returns.
    • 📉 Mean Absolute Error (MAE) for both variance and volatility forecasts.
    • 📊 Mincer-Zarnowitz regression to assess forecast optimality.
    • 🔍 Diebold-Mariano tests to compare predictive accuracy between models.

📂 Data

📡 Financial time series data (DAX, Bitcoin, EUR/RUB) sourced from Yahoo Finance, with additional market indicators such as the Volatility Index (VIX) used as an exogenous regressor.

About

Code Used in My Bachelor Thesis

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

0