Requests & Ideas - page 42

273

More Wavelets - MODWT and volatility.

Alex

Files:
9493

Feg

Fourier extrapolation of Goertzel browser 2 and Fourier extrapolation of Goertzel browser 2.1

ps) you need fourier extrapolation dll in libraries folder and of course Goertzel browser 2 and 2.1 in indicators folder.

Files:
163384

MrM:
are there any other ideas we can try here?
mrtools:
Fourier extrapolation of Goertzel browser 2 and Fourier extrapolation of Goertzel browser 2.1

An idea

mrtools reminded me with his post what was the whole idea of calculating composite waves all about. Hurst and Millard intended that not for reconstructing the past, but to construct the future. Once when components are knows (lengths, amplitudes and phases in current point) apart from being able to "reconstruct" a composite wave in the past and to see how well it fits the past, there is no limitation whatsoever to do that same thing for the future too, So this one is not using an extrapolation for future bars but is using the same technique as it is for past bars too

_____________________________

For those that are not so familiar with what and how does this indicator create a composite wave here are some explanations. Goetzel cycles calculation (more info about it here : Goertzel algorithm - Wikipedia, the free encyclopedia ) is used in this indicators to identify currently existing cycles in signal (in case of this indicator, signal is the price that is used by the indicator) and that is the list of cycles that the indicator is writing out.

Those cycles alone are not much : depending on what you choose (the UseCosineForWavesoption was introduced for that purpose : to remind - crodzillas browser (regardless of the error it had in calculating wave from current phase - we are far, far from that version now and as far as I am aware there are no more errors in this one) used sine for basic waves while it turns out that cosine is a better choice) If you set the UseTopCyclesto 1 you can get to see i single sine or cosine wave which may, or may not coincide significantly with current "waves" in price (here is an example of a 2 single cosines - the most significant in the upper (period 84) and 5th (period 38) in this case, just as an example)
So the most dominant cycle Goertzel cycles calculation have found in signal is the upper, the 5th dominant is the lower and so on and so on. If you put all the cycles together you would get something like this (it would get even more complicated since on this one I used only 7 out of 16 cycles) :
(Amplitudes in this case are all made equal in order to make clear how exactly do they look) And from that data (since those are all elements of current signal) one would need to decide what is the current cycle. Obviously, the way it is in the upper picture is of no use to anybody (except to lovers of complicated pictures that are posted from time to time at mql forum ) Instead of showing it that way the indicator itself is combining those waves into a single wave (that is the "composite wave" that has been mentioned a couple of times) which is showing much more "readable" information. It is doing it to the left (past) in combining waves to reconstruct past, bat it can be done as easily to the right (future) by simply extending those waves to the future and doing the same combining there. And that is what this version is doing : simply extending the whole thing to the future. Compared to the regular one it adds one line more - the future composite wave value and uses one additional parameter for that : WindowSizeFuture- which is simply the size in bars you wish to extend it to the future. Here is a comparison of the two :
After seeing the end-point results makes me curios as of how accurate this method of predicting will be (since deviations of end-point version from current wave values are smaller than I was expecting : frankly I was expecting much bigger differences after seeing all the other end-point versions, so it is a bit of surprise that those differences are not much bigger) and how will Fourirer extrapolation compare to this method, so ... having what to look for - from the way it reconstructs past waves it seems that it is a sound method, but now we can see for ourselves if it is really so and does something else has to be done to make it even "sounder" .

_____________________________

PS: as in any "future prediction tool" be careful with the portion of future you expect to be accurate. The firther it goes, the less is the probability that it will be accurate, but let us see if some "moderate" expectations can be fulfilled (so, I do not expect all the default 100 bars the indicator is drawing in the future to be correct - 99 would be enough too )

updated version posted here: https://www.mql5.com/en/forum/179807

Files:
9493

G3

Thanks Mladen,Simba and Mr M, just checking out the EurUsd using Browser 3 on H1 timeframe will step out on a limb and say bearish

Files:
bearish.gif 60 kb
2078

519

"Williams Percent range" on Accumulation / Distribution

Can anyone make "Williams Percent range" on Accumulation / Distribution, or "Phase Accumulation ",... not on the price.

...and moving average on the same indicator.

If Mladen know the indicator "Power Momentum" of Martin Pring, and if you make this indicator .

Thanks

Files:
519

"Williams Accumulation / Distribution" or "WilliamsPhase accumulation"

I mean, the indicator "Williams Percent Range "on Accumulation / Distribution "is made on the Accumulation / distribution, not on price.

or

the indicator "Williams Percent Range "on "Phase accumulation"

If Mladen know the indicator "Power Momentum" of Martin Pring, and if you make this indicator.

Thanks

118

Extraordinary stuff - really

My thanks to all involved - on any and every level.

I know many will think me obsessed with PAL (my hammer) and thus see all problems as PAL-nails. Perhaps, but nobody else is watching that store so, someone should. Besides, plenty of other, sharper minds are pushing the edge elsewhere.

This sort of decomposition is just what I have been (mostly unsucessfully) pursuing for months after deciding that there was definite signal content in PAL.

After sifting through Goertzel and related posts, I concur with Simba's analysis and requests. I eagerly anticipate Mladen's general Goertzel output and mtf offering.

Speaking of value, best 70 bucks I've ever spent. If you cannot or do not see the value, you just haven't been around long enough.

Speaking of crazy, have a look at default, out-of-the-box settings for this guy.

Thanks for making it general Mladen. Extraordinary.

Files:
110

118

Adeo thanks for posting that version, would love to try it but it doesnt plot anything for me - do I need a DLL or anything else? PAL 1.54 loads fine and goertzel browser 3 loads fine, not sure what my problem is (I just get an empty window).

Thanks

Odysseus

My thanks to all involved - on any and every level.

I know many will think me obsessed with PAL (my hammer) and thus see all problems as PAL-nails. Perhaps, but nobody else is watching that store so, someone should. Besides, plenty of other, sharper minds are pushing the edge elsewhere.

This sort of decomposition is just what I have been (mostly unsucessfully) pursuing for months after deciding that there was definite signal content in PAL.

After sifting through Goertzel and related posts, I concur with Simba's analysis and requests. I eagerly anticipate Mladen's general Goertzel output and mtf offering.

Speaking of value, best 70 bucks I've ever spent. If you cannot or do not see the value, you just haven't been around long enough.

Speaking of crazy, have a look at default, out-of-the-box settings for this guy.

Thanks for making it general Mladen. Extraordinary.