Machine learning in trading: theory, models, practice and algo-trading - page 1735

 
mytarmailS:

sad )

lag increased?, every minute is too thin can be

increments are probably to blame. Doesn't work even if you buy every hour

 
Rorschach:

I would know how to do it already. And I would have predicted at most 1 inflection and retrained on each bar.

I would divide this task into 3 parts: data preparation, breakdown into components, and prediction. I know how to do the first 2, but forecasting is a problem. I would like to use NS, but it is a "hot" field, besides I haven't been engaged in it since NS.

If I want to play with it quickly, I can use these indicators to make predictions. Parameters should be selected in the optimizer.

I don't understand, what do these Lysaju figures show, the connection between the two components in the expansion? That is, there are two components in the expansion that are shifted one relative to the other? Conditional sin and cos. That's redundant, you can replace it with something with a different initial phase.

It will be obvious that the indicator is an adequate person. Frequency analysis is only used up to the first dash of the second refraction at most, there are just such parameters that allow you to work much longer up to 10 signals. But everything beyond the second prism is already at the level of lucky not lucky.

I don't understand what kind of forecasting CSSA is talking about, it selects a frequency that is ahead of the quote and uses it in general. There are two frequencies in the green window and if you look closely, one is ahead of the other.

Right, there are two components, one of which is the klose itself and the other is a found cycle that consists of two components marked in the green window where their intersection points are the inflection points of the cycle and those inflection points are earlier than the klose itself.

There was also some talk about how you can watch that the cycle we use is ahead of the cloze cycle, and not vice versa. I don't remember exactly. It's a long time ago.

 
Rorschach:

Listen, are you familiar with R in general? If so, let's do it together, based on my experience I will try to clarify all the things needed to build the script, of course without the basic components, such as CSSA and egeinvectors itself nothing will work, but I think I can make them correctly in TS for hypothetical check. The main thing is to come to automatic optimization, namely to select period of build and find these two components, which will be the perfect circle of found the cycle with respect to current cycle klose. If there is such a possibility, I would conduct a branching by fight. The period of construction of the cycle KSSA 0-250, and the other two parameters from 0 to 20 approximately change. And I'm sure if you always optimize to the last bar you can't always get the ideal circle, i.e. you have to optimize bar by bar and wait until such a circle is found, i.e. the beginning of the origin of cyclic quotes, will tell. Thus, the trading will pass those sectors of the market where there are no cycles or where the cyclic activity has decreased and finished for the moment. That speaks in favor of this method again, that it is not a grail because it does not always work and we have to wait for the occurence of the cyclic activity. But if the quote's cyclic behavior starts to grow you will be the first to know about it, because the cycle will be found. This is the theory, which I want to check empirically.

 
Naturally I am against the use of NSh, because then it will be a hassle and do not do it automatically. The problem with NSh is that it has a lot of stuff inside, so that only the developer knows what it is like. In essence it is a black box. It is better to take the principle of the idea there and implement it on the R, KSSA is not only for NS, and the logic of its construction is also in the internet.
 

In short, everything is clear about it all, now we just need to rip off the clustering and see what the alpha is all about

The Trickster's got some nice rattles, but he's decided not to give any more gifts.

 
Maxim Dmitrievsky:

In short, everything is clear about it all, now we just need to rip off the clustering and see what the alpha is all about

the wizard has some nice rattles, but he decided not to give gifts anymore

What did he give presents other than tips???? Where is mine? :-))
 
Mihail Marchukajtes:
What did he give out gifts other than tips???? Where is mine? :-))

He promised me for the New Year, but you did not record the clip and the gift went away...

 
Maxim Dmitrievsky:

He promised me for the New Year, but you did not record the clip and the gift went away...

And what he promised? Did he at least said?
 
Mihail Marchukajtes:
What did he promise? Did he at least tell you?

He promised panties, but gave me a bow

 
Maxim Dmitrievsky:

He promised panties, but gave me a bow

Be glad it wasn't a sucking candy and thank him for that :)
Reason: