The market is a controlled dynamic system. - page 15

 
Nafany:

Looking at the picture it seems that the series is non-stationary - usual market quotes. But by its origin it is stationary - the sum of sinusoids. How can we show on the interval of a given sample that intuition is wrong? For example, in order to distinguish strongly/weakly non-stationary areas in real quotes?


Again, the answer to such a question is ambiguous. But you can try using, for example, the Dickey-Fuller test, a description of which can be found on the internet.

.

and also

Cooper. Probabilistic methods for the analysis of signals and systems.

 
-Aleksey-:

p.s. I hope the author does not mind a short remark, faa1947 will never understand that his favourite package is not designed to analyse market series with unstable probability characteristics.

Exactly to the contrary, it is designed and has a huge set of tools for analysis, in particular it has a set of tests to analyse the stability of the model.
 
Nafany:

Here's a picture - it looks like an ordinary quote, in fact it's a sum of sine waves. I wonder if it can be identified as a stationary series?


No question. Here in my article, this is done for this section as well. It is not the sum of the sinusoids.
 
Vizard:

faa1947


If you don't mind showing the practical application of the econ. package !

Actually the article is based on real quotes and a functionally complete analysis has been done. Prepared a sequel with an EA profit exit, is in coordination with meta-quotes.

The article is frankly unimpressive...

Unfortunately you are not the only one. This is not the first time I've tried to turn forum discussions towards the use of econometric methods (mathematical statistics). Instead the forum thrives on other theories drawn by the ears instead of special methods and mathematical apparatuses tuned to economics. They promote "bicycles", e.g. "slow and fast", although this is called a timeframe and is available in any terminal and these properties were widely used by the Japanese as long ago as 300 years ago.

Ok, you whined and that's enough. Keep pedalling.

 
avtomat:
By the way, it would be interesting to see what this economic package would say on the data given.
This is available in the article and is not the only answer to the stationarity question.
 
faa1947:
Exactly the opposite is intended and has a huge set of tools for analysis, in particular it has a set of tests to analyse the stability of the model.
Exactly the opposite: "in particular, it has a set of tests to analyse the stability of the model". And in the market, a series is not described by a stable model, because the properties of the series change. Hence what is the point of evaluating the stability of an unchanging model, what are you doing? For a series whose properties change, you have to evaluate the stability of dynamic models that try to describe these changes in the properties of the series. Can your package do that? What dynamic models does it have? It is so clear - it cannot be explained any easier, can it?
 
faa1947:

If you don't mind showing the practical application of the econ package !

Actually the article is based on real quotes and a functionally complete analysis has been performed. Prepared a sequel with the exit to the profit of the EA, is in coordination with metaquotes.

Frankly speaking the article is not impressed...

Unfortunately, you are not the only one. This is not the first time I've tried to turn discussions on the forum towards the use of econometric methods (mathematical statistics). Instead the forum thrives on other theories dragged to the ears instead of specific econometric methods and mathematical tools. They promote "bicycles", such as "slow and fast", although it is called a timeframe and is available in any terminal and these properties were widely used by the Japanese 300 years ago.

OK, whine and that's enough. Keep pedalling.

In general, econometrics involves the use of econometric indicators, but there are problems with them for obvious reasons (mostly for complete analysis on history) ... but this is not about that ...

i'm not against any method in general ... as long as it works ... so i'd like to see the real market application ... do you have a model? - show me...

just for fun...

 
faa1947:
No question about it. Here in my article, this is done for this section as well. It is not the sum of the sinusoids.
I love confident people - somehow you believe them immediately! I would believe you, too, if I had not generated this data myself. Believe me this is the sum of sinusoids, and the number of these sinusoids is small - less than a dozen. And if you persist, I'll post a continuation of the picture):
 
-Aleksey-:

And in a market, a series is not described by a stable model because the properties of the series change.

Stable models are made in an unsteady market.

 
faa1947:

... "slow and fast", although this is called a timeframe and is available in any terminal ...

Wrong again..... These are fundamentally different, incomparable things.... Well yes I see it's useless to explain it to you....