Optimized Trend Trading | Page 39 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Optimized Risk vs Reward Equation 335 replies

Loss Less Optimized Strategy 63 replies

Optimized Parameters in Metatrader Testing? 3 replies

"Optimized Parameter" feature in EA? 0 replies

Strategy with Optimized parameter 1 reply

Attachments: Optimized Trend Trading
Exit Attachments

Optimized Trend Trading

  • #761
  • May 10, 2010 10:55pm May 10, 2010 10:55pm
I don't suppose you know of a simple way to implement polystab?
matlab implements it as a matrix, which are difficult to deal with in normal programming languages.
  • #762
  • May 11, 2010 3:40am May 11, 2010 3:40am
There's a simpler way because minimal phase filters are generally implemented as IIR.
So if you take a 2d order IIR butterworth filter you will have the same result but with simpler and faster computation.
You can use the Biquad class I give you :
Attached File(s)
File Type: zip Biquad.zip   < 1 KB | 1,074 downloads

Here's a simple example in pseudo java of how to use it :

Inserted Code
double[] input = {...};
double[] output;
 
double period = 8.75;
// Q must be set to sqrt(2)/2 to have butterworth caracteristics
double Q = Math.sqrt(2.0)*0.5;
Biquad b = new Biquad();
b.setLowpass(period, Q);
 
for(int n = 0; n < input.length; n++)
{
    output[n] = directFormI(input[n], false);
    // the boolean value is usefull when filtering close price in a realtime feed
}
  • #763
  • May 11, 2010 6:20pm May 11, 2010 6:20pm
Quoting asm8086
There's a simpler way because minimal phase filters are generally implemented as IIR.
So if you take a 2d order IIR butterworth filter you will have the same result but with simpler and faster computation.
You can use the Biquad class I give you : Attachment
Here's a simple example in pseudo java of how to use it :

[code]double[] input = {...};
double[] output;

double period = 8.75;
// Q must be set to sqrt(2)/2 to have butterworth caracteristics
double Q = Math.sqrt(2.0)*0.5;
Biquad b = new Biquad();
b.setLowpass(period, Q);

for(int...
asm, is that the same 8088 sinc based filter you showed way back on the thread? It looked much tighter in your posted zipped graph compared to flyboy's response. Was that because you used something beyond the basic 39 tap coefficient set you utilized?

Also, just curious what prompted you to choose 39 coefficients? Trial and error based optimization, or was there a more optimal rule that drove that choice?
thanks,
jjj
  • #764
  • May 11, 2010 11:32pm May 11, 2010 11:32pm
Quoting jamjamjam
asm, is that the same 8088 sinc based filter you showed way back on the thread? It looked much tighter in your posted zipped graph compared to flyboy's response. Was that because you used something beyond the basic 39 tap coefficient set you utilized?

Also, just curious what prompted you to choose 39 coefficients? Trial and error based optimization, or was there a more optimal rule that drove that choice?
thanks,
jjj
No, as he mentioned this is a butterworth filter. Quite different. But it should be useful.
  • #765
  • Edited May 12, 2010 12:14am May 11, 2010 11:58pm | Edited May 12, 2010 12:14am
Quoting flyboy
No, as he mentioned this is a butterworth filter. Quite different. But it should be useful.

Right, I got that the recent post is a butterworth IIR. I was referring to the so named 8086 FIR nested in the ullma zip file, with the asm8086 title (post #526 http://www.forexfactory.com/showthread.php?p=3149590).
I looked at the code of coefficients, and it appears to be a single sided 39 tap FIR sinc rolloff with some of the lobes cutoff; the time domain graph looks very low delay compared to your sinc example (maybe because yours is double sided or weighted differently?).

Anyways, just checking how he did that original design and why he chose 39 coefficients.

edit: ok, not exactly a sinc, looks like a single sided raised cosine rolloff?
see impulse response here:
http://www.mathworks.com/access/help...seshaping.html
http://i39.tinypic.com/2s61u90.jpg
I entered the 39 coefficient convolution function and plotted in excel. Looks like a very simple FIR to implement, with a very tight fit in time domain.
Just wondering what drove the design parameters (39 tap, filter type) here; besides low phase, was there a specific reason?
  • #766
  • May 12, 2010 1:30am May 12, 2010 1:30am
Okay here's the butterworth filter. It tracks very nicely but doesn't smooth terribly well. Playing with the period doesn't affect things at all as far as I can see.
Attached Image (click to enlarge)
Click to Enlarge Name: Screen shot 2010-05-11 at 11.00.jpg Size: 203 KB
  • #767
  • May 12, 2010 3:37am May 12, 2010 3:37am
jamjamjam, I choose 39 because CB choose 39. When I made these filters I was just cloning and improving his ullma. All the codes you need to compute my filters coefficients are in the thread.
  • #768
  • Edited 3:56am May 12, 2010 3:41am | Edited 3:56am
flyboy, can you try to plot on the same chart several butterworth with different cutoff periods, like for example : 8, 16, 32, 64... ?
What was the cutoff period and the Q used in your last chart ?
  • #769
  • May 13, 2010 3:41am May 13, 2010 3:41am
Fixed it. It was a coding error on my part. So now I have a decent hull moving average, a sinc filter (with lag) and a butterworth filter (without lag). The butterworth still seems to have more lag than the hull moving average though.

PS: does anyone else find the file upload to be frustratingly slow?
Attached Image (click to enlarge)
Click to Enlarge Name: Screen shot 2010-05-13 at 1.23.15 AM.jpg Size: 207 KB
  • #770
  • May 18, 2010 12:42am May 18, 2010 12:42am
This thread has been quiet for a while so I figure I'll try to awaken it.
We're all looking for a decent clutter filter, so I feel a good first approach is to break down what clutter is. For me, it's three things:
1. Bouncing - large high/low spreads in a single tick.
2. ranging - beyond the nyquist frequency of our filters.
3. jaggedness - How closely those ticks follow each other, or not.

The plot I've been using to visualize filter work demonstrates all three of these aspects at some point or another. I don't think anyone will come up with a single clutter filter that is anywhere near robust, nor does it seem reasonable to try. But a composite approach would work.

What does everyone else thing?
  • #771
  • Jul 27, 2010 3:48am Jul 27, 2010 3:48am
Quoting fajst_k
I love this !!!

http://143.53.36.235:8080/1pc.htm

Fuck Forex Factory
  • #772
  • Aug 9, 2010 9:56am Aug 9, 2010 9:56am
ElephantFlea - If your still reading this thread or if anybody else knows about IGIndex spoofing an API - could do with some help.

I had tried some of what ElephantFlea did, i.e running fiddler to capture the HTTP Posts to try to figure out what it was doing. Most of it was straightforward stuff, instrument, price, stop, limit etc.

But there was a couple of things that confused me, especially the "token". Im sure this is some authentication token but not sure how to spoof that.

Im a very compentent c# developer so please understand im not looking to have it done for me, or for any basic help, Im just stuck on the more complicated parts of this endevour

Does anyone have an API/robot hooked up working for IGindex at the moment (or any other spread betting company) at the moment?
Any help greatly appreciated.
  • #773
  • Aug 10, 2010 10:01pm Aug 10, 2010 10:01pm
hi FxTimer

sorry to trouble you,but i have a question.VFFT indicator is repainting,how to make it not repaint? if we can not make it not repaint,which indicator can i use to help to comfirm the direction of VFFT?
hope you can help. if someone have some advice,please help me
thank you all

yourspace
  • #774
  • Aug 11, 2010 1:34am Aug 11, 2010 1:34am
Serch for end point FFT.
  • #775
  • Aug 11, 2010 2:37am Aug 11, 2010 2:37am
Quoting purley
Serch for end point FFT.
hi purley

thank you, i search end point FFT for many times,but i only get the pdf files or some formulas. i can not find end point FFT indicator,if you can give a link. thank you very much.

yourspace
  • #776
  • Aug 11, 2010 7:24am Aug 11, 2010 7:24am
Quoting dliddane
Does anyone have an API/robot hooked up working for IGindex at the moment (or any other spread betting company) at the moment?
Any help greatly appreciated.
I did have via web macros, worked ok but the companies platform itself was too unreliable and i couldnt work around the various errors it would throw up.

Get yourself an account with smart live markets then you wont have to bother with all that faffing around.
Long live dynamic range
  • #777
  • Aug 12, 2010 6:20am Aug 12, 2010 6:20am
Thanks a million Doris - I hadnt realised that any spread bettors were offering MT4 - looks like they are the first - thats just brilliant news. I had a look at them there and read some reviews - they seem positive. Time to go learn how to write MT4 code now
  • #778
  • Aug 13, 2010 7:15am Aug 13, 2010 7:15am
Quoting dliddane
I hadnt realised that any spread bettors were offering MT4
Yeah there doesnt seem to be much buzz about them, certainly on this site, odd considering how long people have wanted it. My experiences of them have been superb.
Long live dynamic range
  • #779
  • Aug 18, 2010 1:44am Aug 18, 2010 1:44am
like this
Attached Image (click to enlarge)
Click to Enlarge Name: FFT END Point2.GIF Size: 14 KB
  • #780
  • Aug 18, 2010 2:42am Aug 18, 2010 2:42am
is this you?


FFT END POINT indicator
Quote:
Originally Posted by Pip http://www.forexfactory.com/images/n...s/viewpost.gif
Members,

I have just confirmed the scam is being perpetrated by a person named Murat Ibrayev in Kazakhstan. This has been confirmed by colleagues of mine who have fell victim to this person.

According to my sources the perpetrator will lour potential buyers by providing screen-shots of the indicator in action. These screen shots will show recent price action, thus bolstering credibility.

The perpetrator will then indicate that he is offering the indicator at a discounted price of $300 (discounted from $500 for a limited time). The potential buyer is then able to negotiate lower purchase price (as low as $125).

Once the price as agreed upon, the perpetrator then will request a money transfer via Western Union promising to deliver the indicator once the funds are collected.

Unfortunately, Mr. Ibrayev collects the funds and never sends anything to the victim.

Please be ware and spread the word.

Regards,

Attention Attention

One of my friends had been hooked on scam .user scam here is dremur.
He is a scam and cheater.

Warning! In order to avoid scam Never pay with Western Union, Moneygram or other anonymous payment services.
0 traders viewing now