![MQL5 - Language of trade strategies built-in the MetaTrader 5 client terminal](https://c.mql5.com/i/registerlandings/logo-2.png)
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
Is this person from Donetsk? Donetsk University? Aleksandr Vladimirovich? I have been in contact with him. I am aware of his works, I mean synthetic moving averages, ZPMAs of all kinds. And I have not seen anything in them. It's fantastic, but it's true. He took offence at me and stopped answering my e-mails.
He didn't really communicate much himself. He only referred to his various works. Here's an example:
You can find more detailed information in my article "optimization of algorithms of sliding averaging without delay" in proceedings of Donetsk national technical university, series "informatics, cybernetics and computer engineering", release 11 (164), 2010.
I also communicated with his master's student, who has a master's dissertation with the same supervisor, of course. He even agreed that the most important thing is not described in the articles and the most important thing is that they would not tell anybody, but they are looking for an investor. :-)
The authors rivet their articles without fully understanding what they are doing.
The algorithm there is linear by input. But there are no miracles; we deal with an ordinary FIR filter but its implementation is tricky. But in the linear case you can always open brackets. They simply describe a new method of synthesis of usual linear FIR filters. That's all, there is not and cannot be any smell of non-delayedness there. I wrote, there is such a theorem. At this point he shut up and did not comment on the open brackets.
Backwards and forwards averaging by itself does not compensate for any delay, not even "partially". Yurick's result to which they refer is not due to the use of backwards and forwards averaging, but to the fact that his algorithm is non-linear in its input. He uses a "pre-prediction-final averaging" scheme, i.e. some assumptions about the type of signal. That's obvious. There must be "secret knowledge" about the signal, a priori. This is the essence of the full analogy of non-classical spectral analysis and its ability to crack the "uncertainty ratio". The design of a non-linear filter should not start with averaging back and forth, but with what properties of the input signal we can reasonably assume.
By the way Smirnov himself also agreed in one of his letters that the articles supposedly lack the most important thing :-)))
"Hello, *****! ***** is indeed my master's student. However, more than he told you, he could not say. His specialty is economic cybernetics. He hasn't studied filtering theory, is not familiar with spectral analysis, etc. In fact, he is not a bad programmer and is also quite ambitious. I have been developing these methods for 7 (!!!) years. Many my graduate and post-graduate students did this work "blindly" not understanding essence of tasks in view. Why was this so? Simply because it was impossible for them to understand it. Once about 12 years ago my chief manager of a private company "got" me with his crying about the quality of Mark Juric's indicators. And after that I started working on the subject. Why don't I reveal everything? I just feel sorry for my efforts. We don't value intellectual production. One has worked for years, and the other wants to learn everything in 20 minutes. And he will consider an inventor a "sucker" and himself a man who knows how to live."
OK, you are not related, but you are aware of each other's existence.
So what "secret knowledge" about the signal, a priori, are you exploiting? What properties of the input signal do you reasonably assume?
Many things can be assumed. For example, one of the trivial assumptions everyone knows:
Price increments are white Gaussian noise. Sometimes GVH is assumed to be the logarithm of price increments. For forex it is the same thing due to small dynamic range of price on the averaging interval. A property of the spectrum follows from this model: it falls as 1/f. This property can be used to design a filter. For design of an ordinary linear filter even: it is not necessary to aim at big suppression of high frequencies, they are small as it is. In practice, however, this is of little use (as the price increments are not BHP :-)) ), but you can assume a lot of things.
Again. The filter is a black box. I don't set out to discuss the filter. I'm demonstrating the trading system and the reality of outweighing probability.
Dr.Drain has clearly moved to MathCade to make his filter into a coordinate system for "price". What he doesn't realise is that price knows nothing about his filter and therefore doesn't care whether it returns to it or not.
If the quote stream was a "purely random process", everyone would have given up trying to build efficient trading algorithms a long time ago. So I don't really understand what "working with a purely random process" means. Watching it? What else is there to do with it? Non-randomness is obvious from the most trivial ideas. For example, if EURUSD were "random", it could have long ago become not 1.25, but, say. 0.1 or 10. However, it does not. There are mechanisms (and very strong ones, which determine the nature of events) that effectively balance trade relations between countries (and the exchange rate is a trade advantage) in such a way that strong (many times a year, for example) changes of ratios are practically impossible. Unless the system is tiny and closed to the outside world (Belarus, for example). Then natives can do anything, for example, say that tomorrow the tugrik will be 100 times cheaper. Since there are mechanisms, which make impossible such trends, which are easily realizable for random processes, it is obvious that the process is not random. That is enough. Although it is possible to justify it strictly, mathematically, for example, by looking at the properties of ACF. Again, though, calculating ACF from a short sample, for example, is a tricky task. And so on.
To C4: "Until he realizes that the price knows nothing about his filter and therefore does not give a damn if it returns to him or not ".
Heh. It's you who doesn't yet understand that the price is obliged to bounce around the filter by definition. If you're confused by the word "price must" and want to shout "doesn't owe anything to anyone" then you don't understand the cause and effect relationship. Hint: the filter is a derivative of the price.
Heh. You don't yet understand that the price is obliged to wiggle around the filter by definition. If you're confused by the word "price must" and want to shout "owe nothing to anyone", then you don't understand the cause and effect relationship. Hint: the filter is a derivative of the price.
Not convincing. Everything you say is 100% applicable to the same MACD. It does not work. So without demagogy explain to people what is special about your theory. So far it is nothing but illogisms and "postulates" based on which it is not clear.
Not convincing. Everything you say is 100% applicable to the same MACD. It does not work.
Not convincing. Everything you say is 100% applicable to the same MACD. It does not work. So without demagogy explain to people what is special about your theory. So far it is nothing but illogisms and "postulates" based on which it is not clear.