Machine learning in trading: theory, models, practice and algo-trading - page 2631

[mytarmailS]

Maxim Dmitrievsky #:

So in the trailer, if there are any 'dumb' flipping strategies you can try to copy

[Maxim Dmitrievsky]

mytarmailS #:
So in the trailer, if there are any 'dumb' flipping strategies, you can try to copy

I have to look, I don't have any. I only arbitrage and MO

[mytarmailS]

Maxim Dmitrievsky #:
I have to look for it, I don't have it. I only arbitrage and MO

I do not understand how we can restore the trade on the table copiers trades, there is no identifier of the position itself (

i can't understand how you can restore trades in the copier table, no identifier for the position itself( by the way, interesting robot, it opens positions simultaneously in different directions on the same instrument, but closes them at different times with a profit, at the expense of which the profit is unclear, but it is eerily interesting)

[Maxim Dmitrievsky]

mytarmailS #:

I do not understand how we can restore the trade on the table copiers trades, there is no identifier of the position itself (

I do not know what makes profit, but it's terribly interesting)

Not a copier. You download the bot from the market, run it in the MT5 tester, then there is an option to save the report with all trades and other information

example of a simple macd bot report

[mytarmailS]

Maxim Dmitrievsky #:
Not a photocopier. I download the bot from the market, run it in the MT5 tester, then there is an option to save the report with all trades and other info.

Why can't they do it right? They are crooked methaquots, everything is fucked up...

then they don't like R because they cannot make a package for it, then they cannot present deals properly in tables, and they even cannot make a text editor in the goddamn personal account, fucking geniuses... it makes me sick...

I found this on myfxbook .

[Maxim Dmitrievsky]

mytarmailS #:

Why can't they do it right? They are crooked methaquots, everything is fucked up...

then they don't like R because they can't make a package for it, then they can't present transactions properly in tables, and even in a fucking personal account they haven't thought of making a text editor, fucking geniuses... it's sickening...

or from any signal service, yeah.

You can save tables of trades from the signals here too, just that the signal may have a short history, but the bot can be run for the whole thing

[mytarmailS]

Maxim Dmitrievsky #:

Or from any signal service, yes

You can save tables of deals from the signals, but the signal may have a short history, and the bot can be run for the whole time

I will think of something

[Aleksey Nikolayev]

Has anyone studied in more or less detail the issue of dealing with non-fixed dimensionality of predictor vector? Interested in a more or less broad overview of the topic and what (if any) established terminology there is. Something like variable length feature vectors?

[Forester]

Aleksey Nikolayev #:
Has anyone studied in more or less detail the issue of working with unfixed dimensionality of predictor vector? More or less broad overview of the topic and what (if any) established terminology is there. Something like variable length feature vectors?

Trees look for splits by sorting each column.
Apparently you should take as many columns as possible and fill NAN in those rows where they are not used. If the model can handle NAN.

Or something else: -INF, 0, +INF... so that all unused rows are on the same side when sorting.

[mytarmailS]

Aleksey Nikolayev #:
Has anyone studied in more or less detail the issue of dealing with non-fixed dimensionality of predictor vector? Interested in a more or less broad overview of the topic and what (if any) established terminology there is. Something like variable length feature vectors?

What do you mean by that? Describe the problem