Fast Fourier Transform - Cycle Extraction - page 8
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I drew those trendlines manually.
Fft cycle extraction
clahn04 i'm new to this board and my question is, you posted a chart with the different cycles in progress is that software available in metaquotes software ? i use 30 min time frame for trading monsanto, is this software compatible with stocks ?
thanks,
bgp03
Well, its been a while since I started this thread. But I have finally got this Goertzel thing figured out. I am developing a windows based program to do all the calculations for me. I'm using excel for display until I get my graphics engine running.
I combined 9 individual cycle periods, along with thier amplitudes and phases to a daily EURUSD chart. Interesting to say the least.
Here is what it looks like. The next shot are the cycles projected into the future.
I still have a lot of things to look at as far as normalizing data... I'm looking at "hurst-izing" the data to remove some noise to see if that makes a difference. I still have some work to do on my windows program, but I'm still working it. This looks promising though. Enough for me to continue development.
I'll also check the 4-hour, 1-hour etc... will look to combine timeframes to see if there is a good enough correlation.
Carl
PS. Sure am having a hard time uploading this afternoon. Pink = EURUSD_Daily, Blue = Goertzel cycle reconstruction
Here is the 4-Hour chart.
Looks like the cycles flipped starting in November...
(Probably because I used 9000 4h bars in the analysis-so this is a concensus that goes way back... interesting)
Again, second chart includes future projection of the 10 individual cycles.
Carl
Extrapolator
Extrapolator - MQL4 Code Base
Method 1:use Quinn-Fernandes Algorithm
Method 2: Autocorrelation Method
Method 3: Weighted Burg Method
Method 4: Burg Method with Helme-Nikias weighting function
Method 5: Itakura-Saito (geometric) method
Method 6: Modified covariance method
Repainting...
More repainting indicators? We don't need any more of those...
Recommend to use on small TF
Hmmm. I'm reading Simba's comments about how to read the Goertzel frequencies and amplitudes, and how to do the conversions... I think I may have added a step I didn't need when writing the frequency conversions for the graphs.
Its all so confusing. I re-did the graphs. They look alright. But it looks like something is missing. I am going to try to combine ALL the frequencies that Goertzel spits out in-stead of just the highest amplitude freqs. Maybe that would more closely resemble whats going on...
Carl
Extrapolator with
extern int LastBar =0; //Last bar in the past data
Now we are getting somewhere...
Here is the screenshot of the first 40 low freq cycles (out of 128). But it is shaping up nicely. The first 20 didn't look promising, but as the frequencies get higher, the amplitudes are getting higher, which should carve this thing out.
Simba was on the right track. Goertzel, looking at the indicator code uses freq^-1. So that tipped me off that I needed to take the output of Goertzel at face value because the calculation to reconstitute a sine wave is:
Amplitude * (Sin (Freq * (Time * (Percentage of Phase * Cycle))))
Cycle = 1/freq
Time = Bar #
Percentage of Phase = Goertzel outputs the phase in Radians, so you have to convert that to degrees first. Then percent of a complete circle.
Carl