Fast Fourier Transform - Cycle Extraction - page 28
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 echo UK_Kara's sentiment!
MLaden,
I have a question, is it possible to ad a code to adjust the start of the forecast to any bar in history? for example, set the start bar for the forecast to be 5 or 20 bars ago?
Thanks,
Pip
If you want to see the composite cycle only, set the CompositeCycleparameter to true. In that case it shows just one cycle that is the combination of the cycles you chose (cycles 1 to 7, whichever is turned on in parameters by using Cycle1 to Cycle7 parameters). Anyway, found that there was a problem with version 2.02 if some of the sub-cycles was turned off and you wanted to see the composite cycle only (so, if not all 7 cycles were set to true). This version corrects that too.
...
Mr. Pip...I knew it will draw...but I did not know how well it can draw...
Ah Pava, great to see you around mate. I hear you but am only testing mate. No worries
Mr. Pip...I knew it will draw...but I did not know how well it can draw...
...
Ah Pava, great to see you around mate. I hear you but am only testing mate. No worries
my test results came in negative...I guess because our eyes can not see the full spectrum of colors...:)
Thanks for the input P.
my test results came in negative...I guess because our eyes can not see the full spectrum of colors...:)
...
Pip,
Spectrometr is calculating cycle values in a rather strange way (it does not use the amplitude of a calculated wave, but is using a fixed multipliers : 1 for first wave, 2 for second, and so on) In that case the composite cycle is not what it should be but a composite of some "abstract" value : phases of sub-cycles are initially OK but since amplitudes are not, I would not count too much on the composite cycle
See the comparison of spectrometr (upper picture - with it's way of calculating composite cycle) and when real amplitudes are used in calculation for composite cycle (that is the Goertzel browser that you took place in developing too). Both are using first closed bar as a start for calculation in these examples
Thanks for the input P.
Dear Mladen,
Excellent observation and I concur with your analysis; however, i have further observed that the amplitude calculation in the G browser fails to correlate the wave amplitude to price action. In other word, each wave that G browser detect is done so in a vacuum or its starting point does not correlate with the actual price movement.
Spectrometer, on the other hand, seems to do a better job in adjusting the waves and correlates them to market movement. In the two images you have shown you will see that Spectrometer shows the general rise along with the tops and bottoms, G browser does not do that, it simply shows the summation of each cycle independent on how it relates to price.
I don't think i did a good job in explaining what I mean but am hoping you can understand it. If not, let me know and we can discuss further.
Note: if G browser is able to connect the wave detected with actual trend amplitude that would make it that much more powerful. This is a suggestion i made in the AE section before but am not sure if it had been implemented yet.
Cheers,
Pip
Pip,
Spectrometr is calculating cycle values in a rather strange way (it does not use the amplitude of a calculated wave, but is using a fixed multipliers : 1 for first wave, 2 for second, and so on) In that case the composite cycle is not what it should be but a composite of some "abstract" value : phases of sub-cycles are initially OK but since amplitudes are not, I would not count too much on the composite cycle
See the comparison of spectrometr (upper picture - with it's way of calculating composite cycle) and when real amplitudes are used in calculation for composite cycle (that is the Goertzel browser that you took place in developing too). Both are using first closed bar as a start for calculation in these examples
...
Pip
Now this is my personal opinion, so treat it like that
Spektrometr artificially adds strength to some cycles that those cycle do not have. In my opinion we can not use cycles analysis for "general trend" estimation, but we can use the for peaks estimation. Cycles as such (being simple sinusoidal waves), can not reflect long term trends simply because it is not their nature : they oscillate between 1 and -1. If the calculation disregards the amplitude (which is the most important part of a cycle significance) and uses some artificial amplitude (as spektrometer does) then the result, in my opinion, is simply a case of luck. It is not enough to have just the phase for compositing
I know that some people will tell that what Goertzel browser does is similar thing, but, Goertzel algorithm already drops out insignificant cycles and keeps original amplitudes, so, for the bar they are calculating and respecting all he rules Spektrometr is not based on some model while Gortzel browser is. As long as we use it for peaks estimation, we are going to use it in a correct way (but this is my just opinion regarding this matter)
________________________________________
And, a remark not directed to you, but to people not familiar with these kind of things : Fourier transform recalculates. Apart from that, there is no single extrapolation method in the world that is "unchangeable" - so "forecasts" should be used with extreme caution and for estimation only - never, ever should they be used as signals
And for the end : the example from the above posts were similar. But see the example that are valid at this moment (they can even be compared to what actually have happened). and I think that you will see what did I mean when I told about "adding artificial strength"
Dear MLaden,
Excellent observation and I concur with your analysis; however, i have further observed that the amplitude calculation in the G browser fails to correlate the wave amplitude to price action. In other word, each wave that G browser detect is done so in a vacuum or its starting point does not correlate with the actual price movement.
Spectrometer, on the other hand, seems to do a better job in adjusting the waves and correlates them to market movement. In the two images you have shown you will see that Spectrometer shows the general rise along with the tops and bottoms, G browser does not do that, it simply shows the summation of each cycle independent on how it relates to price.
I don't think i did a good job in explaining what I mean but am hoping you can understand it. If not, let me know and we can discuss further.
Note: if G browser is able to connect the wave detected with actual trend amplitude that would make it that much more powerful. This is a suggestion i made in the AE section before but am not sure if it had been implemented yet.
Cheers,
PipWell said Mladen and I am in agreement with your opinion to a degree. As you had kindly pointed out, the key word here is "estimation" and that is exactly what i am after.
I look forward to version 2.04 whenever you get a chance.
Cheers,
Pip
Pip
Now this is my personal opinion, so treat it like that
Spektrometr artificially adds strength to some cycles that those cycle do not have. In my opinion we can not use cycles analysis for "general trend" estimation, but we can use the for peaks estimation. Cycles as such (being simple sinusoidal waves), can not reflect long term trends simply because it is not their nature : they oscillate between 1 and -1. If the calculation disregards the amplitude (which is the most important part of a cycle significance) and uses some artificial amplitude (as spektrometer does) then the result, in my opinion, is simply a case of luck. It is not enough to have just the phase for compositing
I know that some people will tell that what Goertzel browser does is similar thing, but, Goertzel algorithm already drops out insignificant cycles and keeps original amplitudes, so, for the bar they are calculating and respecting all he rules Spektrometr is not based on some model while Gortzel browser is. As long as we use it for peaks estimation, we are going to use it in a correct way (but this is my just opinion regarding this matter)
________________________________________
And, a remark not directed to you, but to people not familiar with these kind of things : Fourier transform recalculates. Apart from that, there is no single extrapolation method in the world that is "unchangeable" - so "forecasts" should be used with extreme caution and for estimation only - never, ever should they be used as signals
And for the end : the example from the above posts were similar. But see the example that are valid at this moment (they can even be compared to what actually have happened). and I think that you will see what did I mean when I told about "adding artificial strength"