Rolling Calculation or Regression
Rolling windows calculation or regression is a useful tool to examine the stability of the model over the sample data, and detect whether a structural break(s) exists in the duration of the sample.
For rolling calculation and analysis, we'd need to decide on the window size (optimal, adaptive, etc.), window type (i.e. disjoint or overlapping), and finally, analyzing the calculations (e.g. parameters' values) of each windows.
This will be a great function to have in NumXL.
Comments
This is also know as Piece-wise or Segmented Regression.
Please sign in to leave a comment.