9

My basic question is: What is an adaptive copula?

I have slides from a presentation (unfortunately, I cannot ask the author of the slides) about adaptive copulae and I am not getting, what this means resp. what this is good for?

Here are the slides: sl1 sl2 Then the slides continue with a change-point Test. I am wondering what this is about and why I need this in connection to copulae?

The slides end with an adaptively estimated parameter plot: sl3 sl4

This seems to show, that my estimates are lagged behind. Any other interpretations, comments would be great!

Community
  • 1
Copuleros
  • 337
  • 1
  • 4
  • 1
    That appearance might be a consequence of the intervals that were chosen, or of something else. There's really very little to go on here. When you say "my estimates" you suggest the depicted estimates are yours, but your question earlier suggests the slides belong to someone else. Are the estimates yours, and if not, why call them yours? – Glen_b Jul 07 '13 at 04:48
  • 1
    @Glen_b no, I just were talking from my point of view, so these are not my estimates. – Copuleros Jul 13 '13 at 07:54

2 Answers2

1

I think adaptive in this context just means the reestimation on a rolling basis. So the parameter should not change until there is a change point. Then the true parameter increases and stays constant after it decreases again because of the second change point. The estimated parameter is evaluated compared to the true parameter: How fast does it get the change point? How fast does it adapt to the new environment?

user1690846
  • 236
  • 1
  • 3
  • 6
1

It means re-estimation on a rolling basis. However, I do not understand why one would want to do this when there's perfectly good parametric dynamic copulas invented by Patton (2006) and extended by others with various forcing equations, as well as the more recent stochastic autoregressive copula (SCAR). Read here [1][www.wisostat.uni-koeln.de/Institut/.../Manner_Reznikova(2010ER).pdf‎].

user2763361
  • 671
  • 6
  • 20