Author's dialogue. Alexander Smirnov. - page 36

New comment
 
Integer:
Mathemat:
Is the curve something like a weight function of the k-types for the mach?
yes, it is.

It is roughly clear. All this fiddling with polynomials is, so to speak, an attempt to compress information about the filter weight function: if the k-values are 34, they can be plotted using 5 numbers.

Of course to polynomial regression your indicator has the same relation as Fibo level has to RSI :) But the idea is very interesting.

I put a couple of your filters with parameters 8 and 21 on the chart and found a nice full stop in July 2007. So I have a question: what trend indicator should be to work on that area (i.e. ignore it)?

 
Mathemat:
Integer:
Mathemat:
Is the curve something like a weight function of the k-ts for the mach?
yes, it is.

. So the question arises: what should the trend indicator be so that it works in this area (i.e. ignores it)?



counter-trend, enter at maximum difference
 
lna01:
Mathemat:

In principle RMS^2 = M[X^2] - (M[X])^2. You can try to build something recurrenceally from here.

Think about it.

I did some thinking. The result is a bit unexpected: LRMA is a bit faster - 1000 msec vs 1219 via iMA. It doesn't count the RMS yet, but it's a matter of technique and a couple of pieces of paper.
Files:
movinglr_2.mq4  4 kb
 
lna01:
It is true that it does not yet calculate the RMS, but it is a matter of technique and a couple of pieces of paper.
The RMS is calculated by the built-in MT4 indicator:

double iStdDev( string symbol, int timeframe, int ma_period, int ma_shift, int ma_method, int applied_price, int shift)
 

Yura, I want to count the RMS faster than the standard function. What if it works? For a single call it should be faster than any code written in the language, but for a massive call (counting the whole chart) we can save on costs.

 
VBAG:
Prival
I tweaked it, everything should be fine now.


Still error 2008.02.15 17:07:22 2007.01.11 12:15 OTF_1 EURUSD,M1: negative argument for MathSqrt function

Although Ihmo, instead of bars we should use three numbers instead of one. Open, (H+L)/2 (middle of time interval) and Close. But it will be another indicator

 
Mathemat:

Yura, you want to count the RMS faster than the standard function. What if it works? For a single call it should be faster than any code written in the language, but for a massive call (counting the whole chart) we can save on costs.

Hopefully it will be faster, although it's hard to compete with native code :). By the way, the refusal of zero bar calculation should lead to even more substantial actual gain in testing - in theory iMashcats should work even at opening prices twice. Of course, this is a bonus for those who think that there is no need to calculate it.
 
Mathemat:

Yura, you want to count the RMS faster than the standard function. What if it works? With a single call it should be faster than any code written in the language, but with mass (counting the whole chart) it's possible to save on costs.

That's exactly what was done in Kaufman's optimized AMA : Perry Kaufman AMA optimized

Prival:
VBAG:
Prival
I tweaked it and everything should be fine.


Still got error 2008.02.15 17:07:22 2007.01.11 12:15 OTF_1 EURUSD,M1: negative argument for MathSqrt function

Although Ihmo, instead of bars you should use three numbers instead of one. Open, (H+L)/2 (middle point of time interval) and Close. But it will be another indicator

There was such a thing.

lna01:
Mathemat:

Yura, I want to calculate the RMS faster than the standard function. What if it works? When you call it once, it should be faster than any code written in a language, but when you call it en masse (the whole chart), you can save on costs.

I hope it gets exactly faster, although it's hard to compete with native code :). By the way, refusal to calculate zero bar should lead to even more significant actual gain in testing - in idea iMashes should work even at opening prices twice. Of course, this is a bonus for those who think that there is no need to calculate it.
It will work, as analytical optimisation in complex cases is almost always possible.
 
Rosh:

Although Ihmo, instead of a bar you should put three numbers instead of one. Open, (H+L)/2 (middle of time interval) and Close. But this will be another indicator

I had such a case.


If you don't mind, where and in which indicator is this implemented ?
 
Prival:

If you don't mind, where and in which indicator is this implemented ?
There was such an error due to accumulation of errors.
1...29303132333435363738394041424344
New comment
Reason: