Fast Fourier Transform - Cycle Extraction - page 22
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
Tradability,
one point to keep in mind that the indicator cited uses Fourier Transform algorithm to determine the cycles and FT by default re-calculates values historically, so perhaps that carries over to the resultant wave
. I did not look at the code to confirm.
Cheers,
Hi Pip
Perhaps you're right but the thing is the amplitude resultant is not calculated, either correctly or by default. People here just assume everything is correct. What I believe and can see being displayed is a higher order cycle.
Nonetheless does anyone here know how to calculate the Real Amplitude of the resultant wave rather than displaying the amplitude of a higher wave? They are totally different.
The way you can tell if it's the resultant vector is that it will display a less smooth and more jaggered wave. I will help anyone produce this mathematically if anyone want to calculate the true values.
Let me know if anyone can do this.
TB
Please explain what in means in the math terms.
Thanks for clarifying this BUT.. the RED is not the sum of the amplitudes. I have examined this at specific points and can see that the red should be at a different level. So possible confusion is that whomever coded it altered where the resultant vector is. The question is where is the starting base? Thats critical as it will show the overall position relative to all the cycles. I think the RED is the sum of slopes perhaps at specific points but not sure, however it's definately not the sum of amplitudes at all. Look at the 7th Feb time 11 the RED should be above all the cycles and it's not so if in this point its not then its not at many other points.
Well, I've programmed this indicator and I am sure that red is the sum of three other waves. Perhaps you can use more math what are you want to explain as I'm lost in what you are looking for. For example, what do you mean by the 'starting base'? There is nothing with this definition in FFT.
7th Feb looks just fine. Red is at correct place, 2 waves are above DC component and one below. The sum is exactly at correct point. Perhaps showing what calculations and formulas you are using will explain why they differ from the FFT?
Mind that what you see on this picture is in fact IFFT as this is time series, not frequency. Vectors are added in the frequency domain, or better said the lower energy tons are subtracted in the frequency domain from the FFT of the time series. When in the indicator you set the number of frequencies higher the IFFT would produce a wave closer to the time series, which is of course obvious knowing how FFT/IFFT works.
I don't understand you at all. Please use math for explanation.
Hi Pip
Perhaps you're right but the thing is the amplitude resultant is not calculated, either correctly or by default. People here just assume everything is correct. What I believe and can see being displayed is a higher order cycle.
Nonetheless does anyone here know how to calculate the Real Amplitude of the resultant wave rather than displaying the amplitude of a higher wave? They are totally different.
The way you can tell if it's the resultant vector is that it will display a less smooth and more jaggered wave. I will help anyone produce this mathematically if anyone want to calculate the true values.
Let me know if anyone can do this.
TBPlease define what do you mean by the 'higher order cycle' and which line is according to you this 'higher order cycle'? In FFT usually terms like frequency are used. I never heard about the 'higher order cycle' therefore I'm asking you for the clarification.
Michal
Hi Michal,
Please don't take me wrong but I'm not here to degrade you or your programming skills, however your mathematical knowledge can use a bit of touch up.
Let me enlighten you about adding vectors. Take the distance from for our arguments sake to be the fixed line at a point in the middle of all 3 cycles. Summing the vertical components means your resultant vectors' vertical component will be the sum of all the vertical components of each cycle individually at any given point. So at the point I've mentioned on the 7th at time 11 the resultant point would be in a position much higher than what it is on the graph but secondly what clearly and positively assures me that it hasn't been done correctly is the fact that the resultant vector is smooth lol. It will very seldome be smooth.
I'm wondering what your resultant vector would be for y = sinx + 2.cosx
The method of determining the final graph would be exactly the same.
Perhaps you can put all the cycles in a separate window to see clearly your output and you frame of reference
I hope I'm clear.
TB
Hi Michal,
Please don't take me wrong but I'm not here to degrade you or your programming skills, however your mathematical knowledge can use a bit of touch up.
Let me enlighten you about adding vectors. Take the distance from for our arguments sake to be the fixed line at a point in the middle of all 3 cycles. Summing the vertical components means your resultant vectors' vertical component will be the sum of all the vertical components of each cycle individually at any given point. So at the point I've mentioned on the 7th at time 11 the resultant point would be in a position much higher than what it is on the graph but secondly what clearly and positively assures me that it hasn't been done correctly is the fact that the resultant vector is smooth lol. It will very seldome be smooth.
I'm wondering what your resultant vector would be for y = sinx + 2.cosx
The method of determining the final graph would be exactly the same.
Perhaps you can put all the cycles in a separate window to see clearly your output and you frame of reference
I hope I'm clear.
TBI'm very sorry, but without proper explanation and answering questions which I've asked you I'm unable to help you. You are just repeating yourself with the ridiculous argument that sum of vectors is seldom smooth. I wonder where you got it (care to send a reference)? Anyway, I would advise you to study what DFT and IDFT is. There are plenty resources on the net. Moreover, you don't need to worry about my math skills as I have an university degree in it.
Regards,
Michal
Gentlemen (Tradability & Michal),
No need for anyone to take things personal or as grounds for one party to prove that they are better than the other. Also, there is no need for anyone to belittle the other regardless of their skills. We are all here to learn and share ideas and make it better.
So, Tradability and Michal, if you could be so kind as to expand on your points with analytical examples accompanied with some figures or graphs that would make it easier for those involved to follow your point and perhaps clarify any misconceptions or miscommunications that you may perceive so that others can chip in.
Cheers,
Pip
Michal
I've already explained how it should be calculated for you but you ignored that fact and yes you do have a degree but what do I have? lol People like you come under my care year in year out for decades now in tertiary institutions. But that's irrelevant.
Review what I've said and I have even asked you to show me how and what the resultant of the sinasoidal wave I've given would be to make sure it's been calculated correctly.
I think what you have done is perhaps shifted the resultant but even if you did your resultant curve is too smooth to any degree of accuracy.
The best I can do is print your graph and hand sketch the resultant curve on top of the cycles. Post just a pic of JUST the 3 cycles and ill print produce by hand the resultant and post it here. You want that?
Also I found a pic on a Russina site I saw that looks like what the resultant curve should look like. This is most likely to me a resultant curve to cycles with differing frequencies,
take a look at the yellow curve below....
I can't read Russian but what ever the indicator is in the pic I can't load it and it seems like a very good indicator
Point taken Mr Pip
I think were miss-communicating here, it's the difference in the accents.lol
Probably last post on this topic from me, I have more important things to do.
The best I can do is print your graph and hand sketch the resultant curve on top of the cycles. Post just a pic of JUST the 3 cycles and ill print produce by hand the resultant and post it here. You want that?
Also I found a pic on a Russina site I saw that looks like what the resultant curve should look like. This is most likely to me a resultant curve to cycles with differing frequencies,
take a look at the yellow curve below....
I can't read Russian but what ever the indicator is in the pic I can't load it and it seems like a very good indicator
Point taken Mr Pip
I think were miss-communicating here, it's the difference in the accents.lolI don't know what you see, anyway I've updated the graph to a little newer with the 3 highest power frequencies and horizontal line on the DC level (X axis). All green sinuses are extracted from time diagram and the red line is the sum. You can draw your version on top of it, so perhaps we find what is the miscommunications here.
Your yellow graph does not seem as IFT to me at all because it should be on the time graph, unless it is not properly scaled, then it is obviously wrong. The time series is always > 0 while your yellow curve has values below 0. This is not correct, again I advise you to search some info about Fourier Transform and educate yourself. I know that complete understanding of the Fourier Transform requires an university level of math but general grasp of the concept should be understandable for everyone.
Regards,
Michal
This is not what you had before and I can see that this is the resultant curve now. Perhaps the update you made changed something. You must have one Degree only right, at which Uni? I don't recall seeing you there? Did I pass you?
Can I have a copy of this Red curve you have now please provided where I can alter the frequencies of the other curves.
LOL @ time >0 ...do you know anything that has time <0 ?The resultant need not be on the same graph. The horizontal scale is the time scale as is in all charting graphs with regards to trading. You can draw any resultant anywhere provided the scales and all important info is shown.
You impress me with your maths. Do you know anything about my maths? lol
Lets take this a little further by using calculus and SMH shall we Michal to see what we can come up with together?