From theory to practice - page 569

 

Anyway did a modulo increment and plotted a histogram window of about 30000 values, very similar with this

https://www.mql5.com/ru/forum/221552/page567#comment_8665711

 
Evgeniy Chumakov:
Here you go

Well, also converging to normal, h.t.d.

Now look at the subject - I've already mentioned it in "Machine learning...".

It is possible, as I do now - to wait for the sum of increments to go beyond a certain interval, etc., etc. But, deals are extremely rare... Terribly boring and tortuous...

If this junk (amount of increments) is shoved into a neuronet, it will stupidly hack it and the trades will be at every time step - 5 min. Fun. Profitable. A grail. No?

 
Alexander_K:

If this junk (amount of increments) is shoved into a neural network, it will bluntly hack it and there will be trades at every time step - 5 min. It's fun. Profitable. A grail. No?

NS is complicated for me.

 
Evgeniy Chumakov:

NS is difficult for me.

He's the one who's fooling around.

 

And erlang how to count, in the sum of the increments count not all in a row, but at a certain interval. I.e. = increment 1 +

increment 3 + increment 5, etc.

 
Evgeniy Chumakov:

And erlang how to count, in the sum of the increments count not all in a row, but at a certain interval. I.e. = increment 1 +

increment 3 + increment 5 etc.

No. It's much more complicated than that... I'll post the data later.

 
Олег avtomat:

He's the one who's playing dumb.

I didn't turn on anything. I'm just at a standstill right now - looking for different options, communicating with young people in their language. What else is there to do?! It's so depressing...

 
Alexander_K:

No. It's much more complicated than that... I'll post the data later.


Write the algorithm, maybe I'll understand what's what.


By the way, should the M5 increments be calculated on a five-minute chart? Or it would be better to use Close(current) - Close(5) on a minute chart?

 
Evgeniy Chumakov:


Write down the algorithm, maybe I'll understand what's what.


By the way, do the M5 increments count on a five-minute chart? Or it would be more reasonable to use Close(current) - Close(5) on a minute chart?

On the five-minute chart.

Erlang's Flows Algorithm? We should accept data through the sum of exponential time intervals. There are no such archives and cannot be - that's why I cannot use strategy tester.

 
Novaja:

Anyway did a modulo increment and plotted a histogram window of about 30,000 values, very similar to this

https://www.mql5.com/ru/forum/221552/page567#comment_8665711

Well, it's xi-squared, obviously.

It's the variance. I'm wrong of course - the sum of the moduli of the increments is directly proportional to the variance.

We should just count the sum of the increments - we have a normal distribution.

But again, what should we do with it? Should I work with the levels in the weekly window? I'd rather stop working on the market altogether - I'm bored and I don't need any money.

Nothing else but neural networks comes to mind...