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
And I am gradually leaning towards that view. Something remains to be clarified, namely. Suppose, reasoning by analogy, Close[] is the water level in a reservoir. We have no way to measure the exact inflow (rainfall, numerous rivers, etc.), but will estimate the inflow, by the level.
When the water level increases, everything is clear - the inflow will be a positive difference. But if we started the observation in a dry summer and as a consequence - negative values of the change in the level. After all, a decrease in the level does not mean that there was no inflow, just that more has flowed out than inflow. The average inflow should be equal to the volume released from the reservoir each year. Again some cyclicality comes up.
I want to clarify what is the correct methodologically to take as inflow for my case:
- Only the difference
- Only the difference modulo
- Only the positive difference
Intuitively, I understand that the main thing is to correctly determine the inflow.
I also doubt that there is a correlation between Close[] and Close[i]-Close[i+1]. I once tried to find it using DSP methods (as a test, I was debugging functions to determine correlation and autocorrelation of signals at the time), the calculation result - not related in any way. This, in general, can be seen with the naked eye, if you look at the shape of the signal. But these are DSP methods, it may not be logical to use them in this case.
Also, one cannot reconstruct the original series by Close[i]-Close[i+1] alone - the necessary information is just lost, and the Hearst calculation is performed, in fact, by a different series from the original one.
-Difference only
- modulo difference
- only positive difference
Intuitively, I understand that the main thing is to correctly determine the inflow.
I also doubt that there is a correlation between Close[] and Close[i]-Close[i+1].
...
Also, one cannot reconstruct the original series by Close[i]-Close[i+1] alone - the necessary information is just lost, and the Hearst calculation is performed, in fact, by a different series from the original one.
I would only take the difference. Also, if you've got a good reason, don't say you can't use it :)
http://forex.kbpauk.ru/showflat.php/Cat/0/Number/120077/an/0/page/0#Post120077
And I am not a professional "digitizer". Three months ago I've come across an article praising the use of DSP in trading. I found it interesting, and decided to make a basic set of digital filters. Well, it was done. Went to a bookstore, chose two thickest books and bought them. After reading the first 200 pages of the first book I encountered, I closed it and, muttering swear words, went to pick up the dusty two-volume book of mathematics.
I was interested not to buy a ready-made product (I have enough money for that), but to make my own, assuming that the study of this area - new ideas will appear. And so it happened. Within 3 months, without any distractions I got what I wanted - DSP-based indicators and a lot of ideas.
But perfection knows no boundaries....) :о))) I want to ask professionals (it's faster or I'll get it myself, but later) to define some specifications of existing filters and write really adaptive filter.
From what is offered - no one has them, I mean no adaptive filters.
PS: And I'll get to spectra soon :o))))
I would only take the difference.
...
I'm thinking the same thing. It's a lot of right data, well ... almost right.
- DSP based indicators and a bunch of ideas.
......
From what's on offer - no one has them, I mean no adaptive filters.
PS: And I'll get to spectra soon :o))))
Here I am, I believe in DSP and "not a professional "digitizer""
I use this:
"Package for technical analysis by digital filtering methods
Developed thanks to Keny (Oleg, Krasnoyarsk, unclekenny@yandex.ru) and
Goodman (Zyabrev Ilya, Moscow, info@goodman.ru)
Help in testing - Tsygankov Andrey gypsy@mail.ru, Elizarov Oleg TOPpoint@yandex.ru
Distributed for free according to "AS-IS" principles.
Please inform the author about all remarks and errors."
and there is one unrealized idea how to optimize filters on the fly:
There is nothing new. This has been done manually hundreds of times, but how to automate it? -
I'm talking about FATL SATL (FATL-SATL, FATL SATL CD ...) and adjusting their parameters
manually using spectrum analyzer (included in the package).
A library is needed that performs spectrum analysis, giving output periods of
maximum signal density (time series). Analysing this data
for different fragments of time series you may try to guess periods, phases, weighted coefficients (amplitudes) of harmonic components (how? I don't know yet)
The more harmonic components may be determined the more probable it is to forecast development of series. That's all.
Yeshe. If the forecast differs a lot from reality, we can suppose the reaction
of the market to strong news. Now that is all.
Not a programmer and not a mathematician. :)
This branch is clearly beyond my skills. I've been chewing on it for a week now. :(
Two screenshots made with only one difference in calculation version setting (OldVersion: true and false)
You can see that the old version shows habitual levels, while the new one shows a failure with the same parameters - "sticky" Murray levels. Can you comment on this situation? Are "adhered" Murray levels in the new version of the indicator just a technical failure or do they have a deeper meaning? For example, no Murray levels can exist today because of the strongest movement and we should just wait until they are correctly calculated for example on Monday and then we can make decisions about market entry? And today, for example, if you did not manage to enter yesterday, it would be better to stay out of the market? I would like to hear your opinion on this issue.
PS: But now with the new bar the new version has started to display the same Murray levels as the old one. Probably it is some technical error of the new version of calculation of the Murray levels.
PPS: A couple more bars came and levels in the new version are stuck together again.
where can i see it?
google is silent
...............................................
PS: But now with the arrival of the new bar and the new version began to show the same Murray levels as the old one. Probably it is some technical mistake of the new version of Murray's levels calculation.
PPS: A couple more bars came and levels in the new version are stuck together again.
Strange - I'll keep an eye on it - of course it shouldn't be like that. Need to look at the initialisation of the arrays I'll check.
I actually don't have it, but I don't use it as a separate indicator - I just built the code into the Expert Advisor.
HH I found one more place where there was a typo. You were right - the error is technical. Here is the corrected code :
Good luck and happy trends.
Really fixing one more error eliminated the problem, as only the original version of adding code at the beginning of start() did not solve the problem and I was about to send you quotes and screenshots to reproduce the problem on your computer. But now it is not needed because you have completely solved the problem judging by my observation of the latest corrected code. I will move the new corrected code to my Expert Advisor.
Thank you very much for the prompt solution of the problem!!!
There is also a paid one from Finware. haven't tried it