Fast Fourier Transform - Cycle Extraction - page 87
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
"One thing"?!...do not be so stingy...
One thing I will give you credit for is your "glass half full" attitude.
"One thing"?!...do not be so stingy...
Don't push your luck...
Don't push your luck...
On that response to Pava a stern mouth expression with that goatee could have it come across threateningly instead of pleasant like it is now. TeeHee.
Back to biz: Thx Pip for cpo, could you point me to post that has your method you do with said indicators, or was that ever availed?
Larger biz for this FFT forum. Would any experienced indy programmer be interested in back modifying these extrapolator indies to their pure fft outputs? I.E., volume of sin/cos components, say 8? That be a start. Thereafter a calculating weighting "analyzer" could be further programmed on these outputs from there. It be a new indy game as what has occured is presumption these components needed to be added back ("re-compiled") for "predictive" purposes which may or not be only way to go unless one is of Pip Trader character to use in correct way. ;-)
Thanks.
Jerry
"goatee"?...he looks like Vladimir Lenin reincarnation!
On that response to Pava a stern mouth expression with that goatee could have it come across threateningly instead of pleasant like it is now. TeeHee.
Back to biz: Thx Pip for cpo, could you point me to post that has your method you do with said indicators, or was that ever availed?
Larger biz for this FFT forum. Would any experienced indy programmer be interested in back modifying these extrapolator indies to their pure fft outputs? I.E., volume of sin/cos components, say 8? That be a start. Thereafter a calculating weighting "analyzer" could be further programmed on these outputs from there. It be a new indy game as what has occured is presumption these components needed to be added back ("re-compiled") for "predictive" purposes which may or not be only way to go unless one is of Pip Trader character to use in correct way. ;-)
Thanks.
JerryJust a question : why not using the FFT itself (that is available at this thread too) for cycles/frequencies/amplitudes extraction/calculation?
Just a question : why not using the FFT itself (that is available at this thread too) for cycles/frequencies/amplitudes extraction/calculation?
Which one, post #9 or #10? What is difference of these two? Do these give one or multitude of sin/cos volume outputs? Do you think this is what I'm looking for for bar-open-only strategyquant output usage?
Which one, post #9 or #10? What is difference of these two? Do these give one or multitude of sin/cos volume outputs? Do you think this is what I'm looking for for bar-open-only strategyquant output usage?
I though how to answer simply. Instead, posting this video that might help when it comes to FFT (actually Fourier transform - FFT is a special case of Fourier transform) with some simple explanation. Each Fourier transform gives you signals/amplitudes/cycles in the "first pass" (when you get the frequency domain) of computation. If you take those same results and input them to the same Fourier transform you will get the original signal (the time domain) if you do not eliminate some of the cycles (frequencies) that are part of a composite signal. One of the limitations of the FFT (compared to "simple" Fourier transform) is that FFT can process only certain lengths of data (data length must be power of 2). If you are not planing to use very long data samples even a "simple" Fourier transform will do. If you plan to take a shot on long samples (2048,4096 or longer) then it is better to use (for the sake of efficiency and speed) FFT, otherwise, for the sake of the flexibility of choosing any sample size, use a "simple" Fourier transform or some mixed radix Fourier transform (but those are yet another story)
To avoid the risk of over complicating, will stop now. Take a look at the video (those are guys from Tektronix) - it might clarify some things
On that response to Pava a stern mouth expression with that goatee could have it come across threateningly instead of pleasant like it is now. TeeHee. ...
You are a bit late to the party Jerry but don't worry, Pava is a big boy and he can handle it.
If you haven't done so already i highly suggest going through the entire thread. There is a wealth of information that you should not moss on. As per your request, here are the relevant
https://www.mql5.com/en/forum/178842/page22
https://www.mql5.com/en/forum/178842/page23
Regards,
Pip
Awesome video! Thank you so much for sharing Mladen
Hi Batchboy,
If your goal is to export to StrategyQuant, you are going to run into lots of problems with recalculating indicators. You are much better off looking into adaptive indicators that have their history intact. Since you are exporting the current bar as a csv file your resulting time series is going to look very different from what you might expect. Take a look at end pointed FFT or SSA to see what I am talking about.
regards