From theory to practice - page 415

[Alexander_K2]

igrok333:

go to dukascopy. there's seconds data there.

??!! Thank you.

[Alexander_K2]

A post for nothing. For those in the know.

1. The "root of T" law for calculating variance is valid for any time sampling interval in Erlang flows - whether the average is =1 sec, =30 sec or whatever.

2. The variance calculation should not be related in any way to the deviation of price from any moving average. Only on the basis of the sum of incremental moduli in the selected observation time window.

[Vitali Kadel]

Alexander_K2:

A post for nothing. For those in the know.

1. The "root of T" law for calculating variance is valid for any time sampling interval in Erlang flows - whether the average is =1 sec, =30 sec, or whatever.

2. The variance calculation should not be related in any way to the deviation of price from any moving average. Only on the basis of the sum of incremental moduli in the selected observation time window.

Brainwashing

[Alexander_K2]

Vitali Kadel:
Brainwashing

:))) Got frustrated with the theme? It happens! But, keep an eye on the signal once in a while - come back here in good spirits when there's a +.

[secret]

Alexander_K2:

2. The calculation of the variance must not be related in any way to the deviation of the price from any moving average. Only on the basis of the sum of incremental moduli in the selected observation time window.

Dispersion based on increments assumes a starting point of t=0. You don't have it.

More precisely, you would get the variance in this way relative to the price at the starting point of the window, not relative to the centre or the average, whatever it may be. No credit.

[Yuriy Asaulenko]

Alexander_K2:

Mm-hm....

The situation in this thread is no better than in "Machine Learning..."

Farmers! Ugh, i.e. gentlemen!

I beg you for the 100th time, do a good deed - take some tick archive, convert it to a second even series, using a second CLOSE. If there has been no new tick for a second, then write down the value of the previous second.

Post this second BP to the public.

In return, I will teach you how to work with Erlang flows with respect to such a second BP and obtain Laplace distribution on the increments.

Go fuck yourself. Why do you keep begging - do me this, do me that? Either do it yourself or see a psychiatrist.

[Alexander_K2]

And another question - in which program is it possible to make multiplication of distribution of increments?

That is, redraw the histogram at each step of the FIFO buffer and save it in avi format, for example? I was given such a task by Warlock and I didn't do it... I've been haunted by failure ever since. Magic...

Example:

https://ru.wikipedia.org/wiki/Уравнение_Фоккера_-_Планка

[Renat Akhtyamov]

Alexander_K2:

:))) Pardon. I'm a collective farmer myself, I agree. But, not surrendered, but staggering under the blows of the market.

Exactly.

fuck that strategy.

You're suggesting we all just dump it?

that's just crazy...

[Aleksey Nikolayev]

A big problem for TC is the lack of mastery of the fundamentals of theorising. Price increments are likely to be non-stationary and hence their sampling distribution does not make much sense. This is the domain of statistics of random variables, where methods like maximum likelihood are usually used.

[Renat Akhtyamov]

Alexander_K2:

And another question - in which program is it possible to make multiplication of distribution of increments?

That is, redraw the histogram at each step of the FIFO buffer and save it in avi format, for example? I was given such a task by Warlock and I didn't do it... I've been haunted by failure ever since. Magic...

Example:

https://ru.wikipedia.org/wiki/Уравнение_Фоккера_-_Планка

A normal MQL indicator...

will re-calculate on every tick

don't believe it, it will be the same

don't worry, MQ made this feature on 5p a long time ago and it's in the codebase