From theory to practice - page 167
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
Yuri, create a real signal please, if possible
I don't need it. I don't discuss real anywhere, and I'm not interested in anyone else's either.
ZS For a signal on futures, there is not enough liquidity and the system will stop working. That's right, by the way.
Nevertheless, friends, this week's profit =+10% and would have been more if not for this disgraceful trend. I will try to apply Hearst to define it.
Firstly (in order not to extinguish the ardor) all that is written on the forum should be read as "in my humble opinion" IMHO.
Secondly, there was already such a precedent, even in competitions that hosted MQ, investigated whether the result is random at the championships, and then MQ introduced a strict track to one participant does not make a few accounts.
The essence of the problem is the dummy theorem: in any limited market segment it is possible to find such a set of intersection of dummies that their signals will give profit.
As usual it is selected retroactively, that's why the tester runs of EAs (EAs) with high degree of freedom are treated incorrectly.
I personally when I started to study MQL and to check it I made a scoop which opens trades on random and then I fixed initialization parameter in the tester and found srand(X) when scoop traded in profit. It is clear that only at optimization.
In the hope that some of the parameter sets will work, we tried owls with different parameters.
You may encounter the same problem, the randomness of the result. The system parameters just happened to be randomly matched to a given section (again, this is a cautionary tale, not a criticism).
Especially this problem often arises due to insufficient data, e.g. as with ticks, so I methodically advised to go up to minutes, they are, after all, for 6 years, some kind of statistics.
About the trend/float and their separation. There's a good cartoon. In the market(s) it's about the same nonsense, you need to see something, something whole. .
Nevertheless, friends, this week's profit =+10% and would have been more if not for this disgraceful trend. I'll try to apply Hearst to identify it.
Hearst is lagging
You need stable probabilities of transitions (or probabilities of stable transitions?) from one state to another, and where to get them... :)Hearst is lagging.
I need stable probabilities of transitions (or probabilities of stable transitions?) from one state to another, but where to get them... :)Yeah, Hearst doesn't fit. At least with me, it doesn't see anything at all. Maybe I'm miscalculating. Different sources have different methods.
Yeah, Hearst doesn't fit. At least with me it doesn't see anything at all. Maybe I'm calculating it wrong. Different sources have different methods.
https://www.mql5.com/ru/articles/2930
There was a discussion here in the comments, SanSanych didn't like it as usual, he wrote about "other" methods :)
Yeah, Hearst doesn't fit. At least with me it doesn't see anything at all. Maybe I'm calculating it wrong. Different sources have different methods.
I wrote above (Have you read?) - A classic method is not suitable, because it requires a large (2-4 thousand points) amount of data. We get "the average hospital temperature" - when the market has gone through all the possible phases (and trend and fdat and SB), and we mixed them together.
There are algorithms for "fast" calculation - you need about 16 points or so. In this case, the accuracy is acceptable.
I do not claim that Hurst is a panacea, but IMHO, it is worth a try. If you want to.
I wrote above (Have you read?)) - the classical method is not suitable because it requires a large (2-4 thousand points) amount of data. We get an "average hospital temperature" - when the market has gone through all the possible phases (both trend and fdat and SB), and we have mixed them up.
There are 'quick' calculation algorithms - 16 points or so are required. At the same time, the accuracy is acceptable.
I do not claim that Hurst is a panacea, but IMHO, it is worth a try. If you want to.
Read, of course.
That is what I suggest to you, Dmitry, and all the readers of this forum and this branch in particular.
It is simply obvious that together we are as close as possible to solving this problem.
There are parts, which may have value, and such that even their owners do not fully understand their meaning (to put it simply).
I will not ask for anything out of principle - it means to admit myself a moron and a beggar. If you want some of my models or studies in exchange for your own know-how, in private or in the public domain, it would be my pleasure.
Yes! ALL decisions must involve the dependence of price on the square root of time and be accompanied by a brief physico-mathematical justification.
I have read it, of course.
Here is what I suggest to you, Dimitri, and all the readers of this forum and this branch in particular.
It is simply obvious that together we have come as close as possible to solving this problem.
There are parts, which may have value, so that even their owners do not fully understand their meaning (they get cold, to put it simply).
I will not ask for anything out of principle - it means to admit myself a moron and a beggar. If you want some of my models or studies in exchange for your own know-how, in private or in the public domain, it would be my pleasure.
Yes! ALL solutions must be related to the dependence of price on the square root of time and be accompanied by a brief physico-mathematical justification.
Dear Alexander, please briefly formulate the main points of the theory in question, so that they can be tested on the whole history of the chosen instrument to overcome it with a positive mathematical expectation. If it cannot be done, then there is no need to waste time on the development of this theory.