Kevin Sheppard

If e has seasonality in it, then CBB and ` X* = X, e*=sample(e), y*=F(X*) + e*` won't build a time series with the same properties as the original data....

If you don't have seasonality, then I think you could do something like 1. Create your bootstrap for e 2. Sample t + m * n where t is the...

This is correct - vector sampling the data will capture the seasonality. The only other change is that you need to adjust the bandwidth so that it is in seasonal...

Yes, the BW is 1 in your example. a BW of 1 is an IID bootstrap. In your problem, BW should reflect any data dependence across seasonal periods. If days...

If you have the time, it would be great to have a writeup in a notebook of the key steps. You can use fake data (simulate as part of the...

Thanks.

@vchernat I don't generally see much of a point in ARFIMA in the types of data I am interested in. It is also relatively difficult to write reliable estimators that...

Sorry, this type of "irregular" GARCH model can't be specified in the package. FWIW, your variance process needs a shock, e.g., `epsilon_[t-?]**2` where ? is 1 or 2.

The only way I could think to do this within the package would be to define a new volatility model that had this structure. I don't think it would be...

There is no explicitt `append` feature. This would be good. For now, the "solution" is to create a new model and then to use `forecast` from the new model passing...