Machine learning in trading: theory, models, practice and algo-trading - page 3602

New comment
 
Maxim Dmitrievsky #:
Isn't it naive without it?)

Well, for example, I took quantum segments, and checked them on 3 other currency pairs for 2010-2024 - everywhere where there was a probability shift - I selected - let's say, it is 30% of the original number, but taking already the section 2007-2010 on the original instrument I found out that it is necessary to throw out another 50% of the selected in this way. At the same time from the original mass works about 60% (if not filtered by other tools). So it seems like a fluke, or there must be another rule clearly switching in the model the use or not of the quantum segment by an external condition - presumably a simple rule that will already work everywhere. Then yes. And so - well, it depends on the variant of cross validation - to pull out of the totality of examples on which everything generally works and evaluate the future - it does not work like that.

 
Aleksey Vyazmikin #:

Well, for example, I took quantum segments, and checked them on 3 other currency pairs for 2010-2024 - everywhere where there was a shift in probability - I selected - let's say, it is 30% of the original number, but taking already a section of 2007-2010 on the original instrument I found that it is necessary to throw out another 50% of the selected in this way. At the same time from the original mass works about 60% (if not filtered by other tools). So it seems like a fluke, or there must be another rule clearly switching in the model the use or not of the quantum segment by an external condition - presumably a simple rule that will already work everywhere. Then yes. And so - well, it depends on the variant of cross validation - to pull out of the totality of examples on which everything generally works and evaluate the future - it does not work like that.

If there is enthusiasm, I can make a script in kolab that fixes labels. Without any chrenrezk and other things. Test it.

By the way, Google has a similar library that fixes labels. There was a link somewhere in my channel.

You're serving up blatant rubbish as tags without basing it on anything, aren't you?
 
Maxim Dmitrievsky #:
If there is enthusiasm, I can make a script in kolab that fixes labels. Without any dickens and so on. Test it.

By the way, Google has a similar library that fixes labels. There was a link somewhere in my channel.

I am open to new ideas. You can just upload the script and don't bother with it.

Maxim Dmitrievsky #:
You submit outright rubbish as labels without basing it on anything, don't you?

In the last example - in general, yes. In other variants - the tags are built by the strategy. It is based on the logic of price behaviour and my observations.

 
Aleksey Vyazmikin #:

I'm open to new ideas. You can just upload a script and not bother.

In the last example, yes, in general. In other variants, the strategy builds the marks. It is based on the logic of price behaviour and my observations.

Purely on new ideas.

All probable clusters of opening orders are already known to certain circles. It's already a take-it-or-leave-it).

 
Uladzimir Izerski #:

Purely on new ideas.

All probable clusters of opening orders are already known to certain circles. It's already a take-it-or-leave-it kind of thing).

We do not belong to these circles.

 
Aleksey Vyazmikin #:

We don't belong in those circles.

I can see that.))

That's why I put the post in.

 
Maxim Dmitrievsky write an article to be famous and don't want to reveal different features that lead to success. I want to hit the moderators, but I don't want to be banned. Constant problem of choice :)
Imho, one should also want to make a normal forum on MO in trading)))
 
fxsaber #:

The further away from zero the average spread is, the weaker its average spikes are visible.

ZY I ran a spread indicator from KB showing the average (time-weighted average) spread at each minute for the last 250 days. It did not show anything similar on EURUSD spikes.

If I understand correctly, within a minute you are looking for an average spread with weights proportional to time. In this case, it is unlikely that such spikes as mine are possible.

I had a task to find not the average, but the average-maximum spread - I took the maximum spread within a minute and then averaged them for each minute of the day.

 
Aleksey Vyazmikin #:
The main thing is how to choose strategies to work at a certain point in time.

Imho, it would be useful to compare economic conditions (macroeconomic indicators) at different sites.

 
Aleksey Nikolayev #:
Imho, you should also want to make a normal forum on MO in trading)))

It is impossible to create a normal forum about MO without normal MO experts...

And normal MO experts even laugh at the word "crypto" let alone "forex".

1...359535963597359835993600360136023603360436053606360736083609...3699
New comment