Author's dialogue. Alexander Smirnov. - page 33

[Sceptic Philozoff]

Well yes, Candid, the native iMA () is hardly calculated recursively. All without taking into account the preceding values, by a straightforward formula.

[Candid]

Mathemat:
Well yes, Candid, the native iMA () is hardly calculated recursively. All from scratch, by a direct formula.

No, iMA() is even faster than direct implementation in MQL. Nothing prevents counting recurrently, just the same code can be written in C and compiled together with terminal.

P.S. I just needed to write the numbers in a different order.

[Sceptic Philozoff]

Ah, there it is. Sorry, misunderstood you.

[Aleksandr Pak]

to Mathemat
Cleaned indices. M_qRMA requires a compiled M_qWMA
P.S. I have some doubts about the consistency of the six. Maybe it's easier to cycle the calculation as it happens? (see f-la in the comments)

[Aleksandr Pak]

I'm surprised at the window, Someone's not catching two files at a time,

[Candid]

Mathemat:
Ah, there it is. Sorry, misunderstood you.

You could have disavowed the paranoid thing, too :)

[pisara]

Mathemat:

What is HMA, pisara?

P.S. Found it: 'HMA'. What's the idea behind it?

HullMA formula, standard LWMA method:

halvedLength:= = if((ceiling(length/2) - (length/2) <= 0.5), ceiling(length/2), floor(length/2));
sqrRootLength:= if((ceiling(sqrt(length) - sqrt(length) <= 0.5), ceiling(sqrt(length)), floor(sqrt(length));
Value1:= 2 * mov(price,length,method);
Value2:= mov(price,length,method);
HMA:= mov((Value1-Value2),sqrRootLength,method);

here is a variant without colours

[Sceptic Philozoff]

lna01 писал (а): You could have disavowed the paranoid thing, too :)

Okay, I'm disavowing it. You're not paranoid. It's a normal measure to ensure the purity of the experiment.

2 Korey: the six is absolutely correct, if everything is counted accurately. It results from the summation of the squares of natural 1 to N. The sum is N(N+1)(2N+1)/6. Direct software summation will give the same result, just a bit longer.

You calculate the normalizing k-value incorrectly, you don't need to subtract one from the sum there. And you have a formula which is commented out wrong: not

k=1./( N*(N+1)*(2*(N+1)) );

а

k=6./( N*(N+1)*(2*N+1) );

[Candid]

Mathemat, how do you calculate the RMS in your approach?

[Sceptic Philozoff]

Through the muwings? :)