|
PureBytes Links
Trading Reference Links
|
Ah. I didn't know the Omega code uses 2 for-loops to do the job of one.
But I'm not surprised. And I know from experience that some other Omega
code uses 2 or 3 variables where one will do. You would think that if Omega
wanted to provide an efficient backtesting system, they would clean up their
own EL for starters. Just another notch against them. I cringe when I
think about what the application C/C++ code must look like. Pierre, maybe
if you rewrote all the Omega functions, indicators, and systems and offered
them to Omega at a big enough price, they would buy and use them.
Kent
-----Original Message-----
From: Pierre Orphelin <pierre.orphelin@xxxxxxxxxx>
To: CodeList <code-list@xxxxxxxxxxxxx>; L_Omega <omega-list@xxxxxxxxxx>
Date: Thursday, July 27, 2000 9:37 PM
Subject: Re: CL_Fast Stochastic Or RSI
Obviously, there is a shortcut!
You can compute the highest and lowest values in the same for loop.
Inputs: Price(NumericSeries), Length(NumericSimple);
Variables: Counter(0), MyHigh(-999999);
MyHigh = -999999;
For counter = 0 To Length - 1 Begin
If Price[Counter] > MyHigh Then
MyHigh = Price[Counter];
End;
Same calculation for Mylow occurs here
=============================
If Length <= 0 Then
MyHigh = Price;
Then use MyHigh and Mylow to calculate the %A_K (Amphetamin_ fast K).
Cannot say for the real speed you will get, but it avoids to read
Price[Counter] two times.
Rgds
Pierre Orphelin
www.sirtrade.com
----- Message d'origine -----
De : Kent Rollins <kentr@xxxxxxxxxxxxxx>
À : CodeList <code-list@xxxxxxxxxxxxx>
Envoyé : Friday, July 28, 2000 03:19
Objet : Re: CL_Fast Stochastic Or RSI
> For RSI calculations, you are supposed to use an "average off" method
where
> SumUp and SumDown are computed using a method similar to the way
exponential
> moving averages are calculated. Something like this:
>
> SumUp = (SumUp[1] * (Len - 1)) + (Price - Price[1])) / Len
>
> You are not supposed to use a For-loop for RSI calculations. I don't
think
> there is a shortcut to computing Stochastic values.
>
> Kent
>
>
> -----Original Message-----
> From: Barry Kaufman <102577.325@xxxxxxxxxxxxxx>
> To: code-list@xxxxxxxxxxxxx <code-list@xxxxxxxxxxxxx>
> Date: Thursday, July 27, 2000 3:39 PM
> Subject: CL_Fast Stochastic Or RSI
>
>
>
> I SENT THIS TO OMEGA LIST EARLIER TODAY:
>
> In order to speed up TS optimization, I want to reduce the look-back
> FOR - END looping overhead of the Stochastic and RSI on each bar. We
> know that this can be done with a simple MA by replacing FOR - END
> lookback, with adding the new bar and subtracting the oldest bar and
> dividing by N. Is there a similar trick for Stochastic and RSI? If
> so, please show a snippet of code to point me in the right direction.
>
> Thanks, Barry.
> The Code List -
> Subscribe/Unsubscribe
> http://www.markbrown.com
>
> The Code List -
> Subscribe/Unsubscribe
> http://www.markbrown.com
|