Machine learning in trading: theory, models, practice and algo-trading - page 1723

New comment
 
Colleagues, do you think there is such a function that I give it an array of shit profits and losses, and it gives me in response to the result of the Z-score, but when I try to calculate it myself, the result still seems to be different from the official data :-(
 
Mihail Marchukajtes:
I tried to calculate it myself but the result still differs from the official data.
I have sort of got it sorted out. Thank you all :-)
 

Regarding cycles... if we take the deviations of the minute closing prices (their increments with some lag) from the opening price of some hour:

the picture for the year for all the minutes of the 7th hour, as an example

the distribution of all minutes

We can distinguish a certain regularity by adding all the increments for the seven hours:

you can see in what minutes of a particular hour it makes sense to sell, and in what minutes to buy (purely statistically)

For example, to sell at the 10th minute, to buy at the 31st minute, etc. You can check it in the tester.

This is kind of a statistical pattern that can be applied to the MO, and so on for every minute of every hour.

 
Maxim Dmitrievsky:

Regarding cycles... if you take the deviations of the minute close prices (their increments with some lag) from the opening price of some hour:

a year's picture of all the minutes of the 7th hour, as an example..........

you're looking a bit crooked.

it is better to differentiate 5 minutes (for example) for each hour and then display their cumulative sum separately



as we see the hours "0" , "1" и "23" have a kind of deterministic trend

"0" We should buy at "1" and sell at "23

 
mytarmailS:

You're looking at it all wrong.

better to differentiate 5 -minutes (for example) for each hour and then separately derive their cumulative amount

Why differentiate then? :D

you made some bullshit.

 
Maxim Dmitrievsky:

Why differentiate then? :D

So they all fit together without any gaps due to absolute prices

 
mytarmailS:

so that they all join together without gaps due to absolute prices

no price gaps, equalize by the opening of each hour, first increment = 0

 
Maxim Dmitrievsky:

No price gaps, equalize by the opening of each hour, first increment = 0

I'm fine there, I've had this script for half a year already written and tested, and I've built it, so...

You better check it yourself )

I have five-minute glue for a single hour, then glue for another hour, etc.
 
mytarmailS:

Yes, everything is fine there, I have this script for half a year as written and tested and TS built, so...

you better check it yourself )

I'm gluing five minutes purely for a particular hour, then gluing for another hour, etc.

i don't need gluing.

 
Maxim Dmitrievsky:

Did you read what I meant? Why do I need to glue

I got it, I got it wrong.

1...171617171718171917201721172217231724172517261727172817291730...3699
New comment