Discussing the article: "Two-sample Kolmogorov-Smirnov test as an indicator of time series non-stationarity" - page 6

New comment
 
Aleksey Nikolayev #:

Ok, I'll double-check it sometime. There was just some discussion that GARCH is stationary, although the realisations look non-stationary (in terms of variance?). I think there was non-stationarity when checking one implementation by some test.

PS It is very good that matstat specialists appear on the forum. Be sure to write more articles.

garch is a stationary process with inhomogeneous variance. That is, the variance varies stochastically, but all the time according to the same law, hence stationarity.
But on the market it is not so, the variance also changes stochastically, but at the same time the law itself changes, hence non-stationarity.
If Smirnov is presented with two garch processes with different parameters, I am sure he will recognise them as non-homogeneous at once. An extremely sensitive criterion in my opinion.

P.S. thank you (although I am still far from being a specialist).

 
Евгений Черныш #:

Perhaps, in order to get a tool that will tell you where and when they are non-stationary. It is not possible to determine it all by eye, you need some criterion, that's what we are talking about.

Profit is the best criterion for everything.
 
secret #:
Profit is the best criterion for everything.
Be sure to write an article about this unexpected discovery.
 
Евгений Черныш #:

Perhaps, in order to get a tool that will tell you where and when they are non-stationary. It is not possible to determine it all by eye, you need some criterion, that's what we are talking about.

You can always select a stationary piece from a non-stationary series, but only on history - no practical use for trading
 
Dmytryi Nazarchuk #:
It is always possible to extract a stationary piece from a non-stationary series, but only on history - no practical use for trading

This can be said about almost any method of technical analysis, such as searching for trends. As it happens, we are not able to analyse prices from the future, only history.

Imho, the methods of the article are interesting and fresh. I plan to use them to analyse the behaviour of a zigzag (on history, of course).

123456
New comment