[Gretl-devel] Re: nonlinearity tests, etc. (2)
On Mon, 1 Mar 2021, Riccardo (Jack) Lucchetti wrote:
On Sun, 28 Feb 2021, Allin Cottrell wrote:
> I'm currently exploring bootstrapping of the BDS test, which looks quite
> promising.
Hm, interesting. I'm too lazy to look at the code ;-) What kind of bootstrap
are you trying? Maybe our resident bootstrap guru (Stefano Fachin) can weigh
in with his opinion.
It's a very simple bootstrap to get a p-value: scramble the input
vector and recompute the statistics a bunch of times, then give the
p-value as the fraction of these statistics more extreme than the
original.
However, I'm now taking a look at Kanzler's code, which may be
better.
Allin