[Tams]

Originally Posted by ch33ch »

Alright. I've attached three charts, each showing different calculated values for the Ergodic oscillator. I believe the last chart should be the correct one. I'm using a custom weighted moving average (as opposed to EL's WAverage) function for the smoothing. ....

why are you using a custom code?

if the anomaly is observed after the customization, then the error is in the customization.

[Tams]

You cannot use a FC function inside a function.

I have highlighted the error in your code.

Inputs:
	Price( NumericSeries ), 
	Period( NumericSimple );

Variables:
	var0( 0 ),
	var1( 0 );

if CurrentBar = 1 then 
	begin
	for Value1 = 0 to Period - 1
		begin
		var0 = var0 + ( Period - Value1 ) * Price[Value1] ;
		end ;
	end
else
	var0 = var0[1] + Period * Price[0] - SummationFC(Price, Period) ;  {<---- error here}

var1 = ( Period + 1 ) * Period * .5 ;
WMA  = var0 / var1 ;

[ch33ch]

Thanks for pointing this out, Tams. I wasn't aware that you can not use FC within a function. Just out of curiosity, why not?

I'm using this version of the WMA because it is more efficient than WAverage. WAverage has to calculate a loop for each bar, and depending on the length of the smoothing period, these loops can slow down calculations considerably. From Logic's Ergodic oscillator parameters, it appears that triple smoothing is used, so this means that even more loops are calculated per bar.

[Tams]

Originally Posted by ch33ch »

Thanks for pointing this out, Tams. I wasn't aware that you can not use FC within a function. Just out of curiosity, why not?

I'm using this version of the WMA because it is more efficient than WAverage. WAverage has to calculate a loop for each bar, and depending on the length of the smoothing period, these loops can slow down calculations considerably. From Logic's Ergodic oscillator parameters, it appears that triple smoothing is used, so this means that even more loops are calculated per bar.

re: these loops can slow down calculations considerably

this code was created 20 yrs ago for the computers of the day.
it won't make a difference with today's computer.


FC remembers the result from previous calculation, and makes a short cut for the current calculation.
This technique works fine inside an indicator, but the memory gets lost when it is called from within a function.


Try SUMMATION instead of SUMATIONFC, you will get the result you expect.

[BlowFish]

As an aside loops are generally a pretty poor way of averaging though they are 'unintuitively' what people seem to use. Sum the new point and drop the point n periods ago for the win

[tbone]

For ch33ch - I tried to pm you, but it was rejected because I haven't posted on this board 5 times. I'll be happy to assist, but you will need to pm me w/ your email address. Standing by, if you're interested ...

[El Guapo]

Any interest in reviving this thread?

[Logic]

I respond to all inquires. No one seems to care..
Nice to see you here