Is it possible to create your own tool with a generated random price wander chart consisting of minute bars on the MT4 chart? - page 7
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
You can mql5, what difference does it make...
For example, take one year or five years of the existing history of a real instrument, and the output will be 20-100 years synthesized,
but with the characteristics of the input range.
There may be at least 100500 of such scripts. They will be of no use at all.
You will never get synthetics which are identical to a real band with all its characteristics. The point is not even that there are supercomplex characteristics, but in something else: we don't even know what exactly needs to be modeled. So, we will have to give up synthetics that exhaustively simulate the real series, because we don't have the guts.
At one time, when I was infected with the synthetic series virus, I read comics and thought a lot. I came to the conclusion that there's no way around it without a specific TC on which these synthetics will be tested. In other words: the synthetic itself must somehow consider the patterns that the trading system uses. I.e., we need to model something that takes into account the properties of the TS itself in a very specific way.
Example 1: Suppose we want to use TS "two muwings, entry/exit by crossovers". What kind of synthetics should be generated? Normal random walk (SR) with independent increments? Nope. Firstly, the market is not SB. Secondly, it's not going to work on SB anyway - no matter what some people who think they can make a long-term profit on pure SB say. What to generate then? Fuck knows... Probably nothing useful to generate for this particular TS.
Example #2: a system based on Fibo levels. There are people who successfully use them. In other words, Fibo levels are confirmed much more often than if the levels were random. OK, now how do we generate synthetics in which these Fibos occur nearly as often as in the real world? I don't think anyone knows how to do that. But they do exist, the Phoebes! And testing such a TS on synthetics that don't take into account Fibs makes no sense - because that's exactly what this system is using. It will be unfairly rejected, even if it turns out to be profitable.
Summary: in order to generate good synthetics, you have to know 100% about the actual series. And when someone finally knows it, you won't need to generate this synthetic :)
About the subject:
You are unlikely to find a ready-made script. This problem is better solved with more specialized packages like MatLab. You cannot completely generate a sequence identical to a natural one (after all, the market is not a random number generator), but you can get a sequence with identical basic statistics like variance and non-normal distributions. In econometrics, special models such as ARCH, GARCH, etc. are used for this purpose. In Matlab, there is a special toolbox Garch modeling or something like that, it receives basic statistics as inputs and random returns as outputs, which are then converted into a price series of increments. If a better approximation to reality is needed, it is better to discard the primitive volatility models they use and use the volatility of real instruments instead. But what is the point of all this? It doesn't make money.
What is there to test on?
So it seemed that modern possibilities allow to generate a synthetic series similar in characteristics.
I would like to try TC for lice ....
Do you have a ready synthetic? Preferably EURUSD.
Please post it!
Example #2: a system based on Fibo levels. There are people who use them successfully. In other words, Fibo levels are confirmed much more often than if the levels were random. OK, now how do we generate synthetics in which these Fibos occur nearly as often as in the real world? I don't think anyone knows how to do that. But they do exist, Thebes! And testing such a TS on synthetics that don't take into account fibs makes no sense - because that's exactly what this system uses.
Summary: in order to generate a good synthetic, you need a 100% knowledge of the real series. And when someone finally knows it, there will be no need to generate this synthetics :)
As always, forcing you to look at the question from a different angle...
Is there something? Even if "not so good"?
here's what was done at my request.
judge.
P.s. on the title of the work, copyright me
... I don't see any practical sense in this.
Here, dear namesake, I do not agree with you :)
The sense, quite practical, contains (imho) already your "redemptive" statement about senselessness of development of a synthetic after comprehension of essence of what is happening against the background of your statement about impossibility of such development without such comprehension.
I translate: Any attempt of statistical research of previously not revealed sufficiently complex regularity by any method which does not use external information on character of this regularity is absolutely senseless. For example: the actively used here MNA is applicable to the determination of the law of universal gravitation only if the researcher dares to choose a power series as an approximating polynomial, but not sines with any other cotangents... The same, but exactly the opposite - to study periodic processes.
Everybody seems to understand that interpolation and extrapolation are two "big differences", - but: Within the current day only, there are more than a dozen posts and even a rather respectable paper about forecasting of opening price of the next bar based solely on the regression analysis, while completely ignoring preliminary estimate of at least general nature of the investigated process "here and now".
Alexey, if two or three people, after reading your post, realize that the perception of VARIABILITY as an unknown regularity makes ANY subsequent application of probability theory senseless, then the meaning of the mentioned topic will become quite practical. People will gradually stop trying to use this theory to discover the law of universal gravitation, and will apply it to estimate the effect of random factors on the speed and acceleration of free fall.
That's a smile, thank you.
You're welcome!
OK, now how do we generate synthetics, in which these Fibs occur about as often as in real life? I don't think anyone knows how to do that.
An exception - to generate synthetics, which do not contain Phoebes in explicit form in the formula, but in which they are obtained during generation in a natural "unknown" way (note immediately - this is not about me, I do not know how, but optimistically I believe that with time I will learn))
Hi all, by the way, it's been a while since I've been on here!
Привет всем, кстати, давненько сюда не заходил!
Oh, what a bunch of people... Where have you been, namesake?!