A TEST OF THE PREDICTIVE POWER OF VOLATILITY FORECASTING MODELS

Open Access
Author:
Binder, Christian N
Area of Honors:
Finance
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
  • Lou Gattis, Thesis Supervisor
  • Brian Davis, Honors Advisor
Keywords:
  • Finance
  • Volatility Forecasting
  • Econometrics
  • Statistics
  • Volatility
  • Time Series Analysis
  • GARCH
  • Forecasting
  • Implied Volatility
Abstract:
In this paper, a quantitative evaluation and ranking of commonly used volatility forecasting models and metrics is presented. The models and metrics included are Generalized Autoregressive Heteroscedasticity (GARCH), Exponentially-Weighted Moving Average (EWMA), Implied Volatility, long-term historical mean volatility, and a moving average volatility. The metrics were tested on a randomly-selected sample of US equities, international equity indices, currencies, and commodities on both historical data and in real-time, making this study unique. Models are ranked and compared via various statistical loss functions.