Tracking the size of the estimation window in time-series data

What is it about?

We propose a new method named Variance Rule-based Window size Tracking (VR-WT), which derives a sequence of estimation window sizes. This method simultaneously identifies structural change points and selects the optimal size of the estimation window. Monte Carlo simulations show that the VR-WT visualizes points of structural change and suggests a reasonable and accurate estimation window size for parameter estimation. We apply the VR-WT method in three empirical analyses: first, to the Capital Asset Pricing Model for Microsoft and Walmart; second, by analyzing the interdependence between the KOSPI and S&P 500 markets; and third, in modeling the time dependence of crude oil prices. These applications involve a simple linear regression, an ARDL, and an AR model, respectively. We successfully identify data-driven structural change points and determine the proper window size for model estimation.

Featured Image