The market is a controlled dynamic system. - page 339
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
.
On the subject of stationarity/non-stationarity.
Many people here on the forum have been trying for a long time (some for years) to get to "stationarity", which is supposed to be in quotations, saying that we only need to find it, and then everything will be easy and simple. At the same time they admit that the process (BP) is non-stationary. But they do not understand the essence of the phenomenon. They do not understand the difference between a stationary process and a nonstationary process.
I shall show this distinction on a simple example.
Description of the process in the most general form :
.
1) Stationary process :
.
2) Unsteady process :
.
Only in sections where the transition matrix is constant (or nearly constant)
the non-stationary process degenerates into a stationary process.
No tricks (differences of first, second, ..., hundredths) eliminate the existing non-stationarity.
However, this does not mean that the nonstationary process cannot be controlled. You can. But for this purpose first of all it is necessary to understand the essence of the phenomenon.
I wish success in acquisition of such understanding to all local searchers.
On the question of stationarity/non-stationarity.
Only in sections where the transition matrix is constant (or nearly constant)
the non-stationary process degenerates into a stationary one.
No tricks (differences first, second, ..., hundredths) eliminate the existing non-stationarity.
Oleg, first of all, I would like to apologize for some excessively impudent statements and judgments.
But my posts should not interfere with the case. And we have one business - the Grail, isn't it?
So the question is:
If market processes are fundamentally non-stationary and finding stationary areas is extremely problematic, then it turns out that using neural networks as a forecasting tool is not applicable in principle.
I think I'm not the only one who needs to understand this point. In order not to waste invaluable time on studying neural networks.
Oleg, first of all, I would like to apologise for some of my overbearing statements and judgements.
But, my posts should not interfere with the case. And we have one business - the Grail, don't we?
So the question is:
If market processes are fundamentally non-stationary and finding stationary areas is extremely problematic, then it turns out that using neural networks as a forecasting tool is not applicable in principle.
I think I'm not the only one who needs to understand this point. So as not to waste invaluable time studying neural networks.
hmm... "some..."... some "childish pranks" followed by "I won't do it again"... so streamlined... cutting corners... No, I'm not going down that road with you. And we're not on the same road.
Your posts work for your cause. What your business is, I don't know. But I know I don't and won't have any business with you.
And to answer your question:
You don't want to waste your precious time studying neural networks. But without it you will not have an understanding of what opportunities neural networks have and what they do not have. By the way, it won't take a very long time to study them, but once you have the knowledge you may get an understanding. For the time being you haven't got understanding: why do you think that neural networks are applicable only to stationary parts, and are inapplicable to non-stationary ones? That's not true, and you should also ask yourself what you are going to load a neural network with.
You don't want to "spend invaluable time studying neural networks". But without it, you will not have an understanding of what features neural networks have and what they do not have. By the way, it will not take a lot of time to study them, but once you have gained knowledge, you will probably gain an understanding as well. For the time being you haven't got an understanding: why do you think that neural networks are applicable only to stationary parts, and are inapplicable to non-stationary ones? They are not. And you'd better ask yourself what you are going to load the neural network with.
The next incremental value, of course. There's nothing else to predict.
OK. In the topic "Machine learning..." can you give references to literature on approximation of non-stationary series whose authors are on a par with Kolmogorov and Wiener? I have not found such works, and I think many would be interested.
P.S. I came here for the Grail, and not for building relations with forum participants. Please understand this and participate more actively in the top branches.
1) The next incremental value, of course. There's nothing else to predict there.
2) OK. In the topic "Machine learning..." can you give references to literature on approximation of non-stationary series, authors equal in level to Kolmogorov and Wiener? I have not found such works, and I think many would be interested.
3) P.S. I came here for a grail, and not for building relations with forum participants. Please understand that and participate more actively in the top threads.
1) In fact, you just don't see or understand what, other than increments, can be predicted.
2) Your screening is too rigid, with such screening "by name" you won't catch new fresh ideas for you.
3) You think to findthe "grail" here, without building relationships? but that's what you do. Or don't you understand that either...? And in the so-called "top branches" there has been a lot of filth lately, and therefore I have no desire to participate in them.
Published: 18 Apr. 2013 г
A conversation between S.P. Kapitsa and V.I. Arnold on the 100th anniversary of A.N. Kolmogorov.
.
.
.