From theory to practice - page 563

[Yuriy Asaulenko]

Alexander_K:

Thank you. I'll be watching. Get your pockets ready, Yuri.

PS. corrected the pre-post a bit.

If VisSim has the right LPFs, it's probably best to use them. There's no need to go further than 2nd order. The best for delays etc is the Butterworth filter.

Coat nada.

[Alexander_K]

Igor Makanu:

Hmmm, there you are... You're basically in all the forum threads at once and nowhere in particular

from the thread about machine learning, here are the incremental modules on Close draws the indicator that has a sliding window - where and to where? is it a regular MA?

I am Schrodinger's cat.

Can I have a picture? Because computer A_K2 is busy and won't let me get close.

[Igor Makanu]

Alexander_K:

I'm Schrodinger's cat.

Can I have a picture? A_K2's computer has taken over and won't let me near it.

It's a Close incremental module.

what to do with them? well I'll screw it in tomorrow.... need a star in the sky... well we will have to run... let's just refute the grail idea and calm down...

[Yuriy Asaulenko]

Alexander_K:

I'm Schrodinger's cat.

Can I have a picture? A_K2's computer's taken over and won't let me near it.

Schrodinger's cat died from a potassium cyanide vial a long time ago.

Schrodinger's cat ghost.)

[Alexander_K]

Igor Makanu:

Cool... Can I see a histogram of these values?

[Igor Makanu]

Alexander_K:

Cool... Can I see a histogram of these values?

I kind of gave you the source in the message above, I don't know what else... Can I get you my computer? Or your laptop? - what?

)))

ZS: formulas or your thoughts write, tomorrow I have time, maybe we will find the grail

HH: Your task should be solved by several stages, i.e. Close increment modules - they should be unloaded to a text file and then processed in Matlab in all possible perverse formulas, then, having found the Grail, we should return everything back to MT... back again, so to speak.

[Alexander_K]

Igor Makanu:

More precisely, the sum of the CLOSE M5 increments in the weekly window. Right?

[Alexander_K]

If it's a Gaussian distribution with a suitable ACF, a neural network will hack it in seconds.

https://ru.wikipedia.org/wiki/Гауссовский_процесс

[Igor Makanu]

Alexander_K:

More precisely, the sum of the CLOSE M5 increments in the weekly window. Right?

there is no sum yet, these are incremental modules in pips

what do we need to sum up? make a MA for these modules?

what is the weekly window? what is the starting point at the beginning of the week? and then ...?

all this research needs to be done in the matlab anyway, there is a pond of ready-made material for the matlab, google it in 1-2 queries, here is the ACF for the matlab:

function [AKF]=akf(signal,TAU)
Tn=length(signal);
for tau=0:TAU-1
    Tt=Tn-tau;
    AKF1=0;
    for t=1:Tt
        temp=signal(t+tau)*signal(t);
        AKF1=AKF1+temp;
    end
    if tau==0
        AKFc=AKF1/Tn;
    else
        AKFpr(tau)=AKF1/Tt;
    end
end
for i=1:(TAU-1)
    AKFlev(i)=AKFpr(TAU-i);
end
for i=1:TAU-1
    AKF(i)=AKFlev(i);
end
AKF(TAU)=AKFc;
for i=1:TAU-1
    AKF(TAU+i)=AKFpr(i);
end

//параметр signal — вектор входных значений. TAU — порядок АКФ.

[Yuriy Asaulenko]

Alexander_K:

If it's a Gaussian distribution with a suitable ACF, a neural network will hack it in seconds.

https://ru.wikipedia.org/wiki/Гауссовский_процесс

I laughed for a long time.)

The NS will absolutely not hack you if you don't hack it yourself beforehand.

Funny about the seconds part too). It takes me 24 hours to learn. The neural network is sparse - ~60 neurons.