Machine learning in trading: theory, models, practice and algo-trading - page 2244

New comment
 
Valeriy Yastremskiy:

If through these decompositions to isolate all varieties of stable sites and make a bible of states (it is clear that they are infinite), why not)))))

so why don't the respected digitizers share their fine filters. And you can check their robustness in 5 minutes.

 
In general, looking at the ticks, I noticed that they are very similar to the random minority game in Alexei Nikolaev's realization. But the bar picture does not look like a tick picture. And if you zoom out, you will still end up with ticks. Which, due to the inability to trace the factors influencing them, are random for us.
 
Maxim Dmitrievsky:

so why don't the respected digitizers share their fine filters. And you can check their robustness in 5 minutes

And there the problem of the beginning and the end of the plot. In a good way there should be a sufficient number of filters that with a small enough error to spread the row, so that it can be put back together. And maybe with this picture it will be possible to do something.

 
Valeriy Yastremskiy:

And there is a problem of the beginning and the end of the plot. In a good way, there should be a sufficient number of filters to decompose the row with a small enough error that it can be put back together. And perhaps with such a picture it will be possible to do something.

So far it looks like they do not understand what they are doing.

 
Maxim Dmitrievsky:

I understand that approximating a sine wave (which Asaulenko also did very successfully) looks very inspiring. But could you describe at least one meaningful approach of using digital filters for trading?

Automaton couldn't do it.

I don't know who Asaulenko is (

It's not approximation, but rather noise reduction.


I'll show you what you do ...

Filtering is when you filter out from data what signals are useful to you, so you filter the information...


When you do boosting, your input is data and your output is bye/seat (filtered information)... It is a normal digital filter, but as there are many trees it also has noise reduction...

You think about it? Congratulations, you do DSP.)

 
mytarmailS:

I do not know who Asaulenko is (

It's not approximation, but rather noise reduction ...


I'll show you what you're doing ...

Filtering is when you filter out from data what signals are useful to you, so you filter the information...


When you do boosting, your input is data and your output is bye/seat (filtered information)... It is a normal digital filter, but as there are many trees it also has noise reduction...

You know what? Congratulations on your DSP job.)

I asked a specific question.

 
Maxim Dmitrievsky:

I asked a specific question.

Do you need a ready-made solution? I don't have one...

I'm working on levels, price filtering hasn't been applied yet

 
mytarmailS:

Do you need a ready-made solution? I don't have one...

I'm working on levels, price filtration has not yet been applied there

at least something meaningful, because they write all kinds of rubbish

I know how to use meaningful filtering on quotes, but I'm waiting for Automat to write
 
Maxim Dmitrievsky:

At least something fucking meaningful, otherwise they write all kinds of bullshit.

It all depends on the task, I do not understand what you want to hear

 
mytarmailS:

It all depends on the task, I do not understand what you want to hear.

He wants everything chewed up and put in his mouth and all he has to do is eat it.

1...223722382239224022412242224322442245224622472248224922502251...3495
New comment
Reason: