Sultonov differential indicator - page 16

New comment
 
Yousufkhodja Sultonov:
Yes, Artem, I also, having just returned from work, discovered this unpleasant fact. It redraws its readings when you reset it. This is a result of the fact that, it includes historical data outside the calculation period, which is not correct. Please make it work only with the latest data sanctioned by the period specified in the settings and do not go beyond that towards history. Or, it remains to restart the indicator programmatically, which is not good. Let it read in that last range, which it has been prescribed. The cat, now the indicator shows the market conditions with period of 1000, but it was a mess before reinstallation, because it takes unnecessary data from the history, since the initial installation:https://www.mql5.com/ru/charts/7574577/eurusd-m1-e-global-trade

Yusuf, go to the stock exchange - there this data is given without any calculations and without any delay/overrides.

Above is the price, below is "your" indicator (delta between the lines)


The only thing is, you can't call it by its name))
 
Maxim Kuznetsov:

You will get a sliding sum and, in the limit, an ATR over a long period


You just have to try it out and see the results. It's not difficult. And new ideas will emerge from the new results.

 
Artyom Trishkin:

Your problem is that it is accumulating data from history. Now imagine if it only takes a certain number of bars to accumulate data, then with each new bar the amount of data will change, and the line will be completely redrawn with a different look.

Suppose we have data to accumulate in the amount of 10. Make a table with different values and then shift the beginning of calculation (on the left) one bar to the right, thus emulating the appearance of new bars - the beginning of calculation will also constantly shift to the right and, accordingly - with each new bar the accumulated data amount will change that will lead to changes in the appearance of the indicator line - its complete redrawing.

But after all calculations the indicator should have the same line at the output, regardless of the bar from which the calculation has been started.

Artem, this is how any algorithm, including this one, should work with the authorized number of bars, and not accumulate an unknown number of bars from the history, even from "his" recent history. It should from bar to bar, as new information comes in, redraws, but will show the true state of the market within the given N bars. If you need more bars, we will set their greater number in the indicator settings. Definitely, we must correct this mistake - it is the cost of remote communication.
 
Олег avtomat:

As far as I understood from the previous one, the indicator simply gives cumulative sums of increments of different directions. The results can be improved by introducing a "forgetting" operator, which slowly reduces to zero the old irrelevant values.

Oleg, the problem is worthless. Is it difficult to designate in the code to take only the last N bars of history for calculations, than to bother with "forgetting"?
 
Dmitriy Skub:

Yusuf, go to the stock exchange - there this data is given without any calculations and without any delay/overrides.

Price at the top, "your" indicator at the bottom (delta between the lines)


The only thing I can't give my name to))
I don't pretend to be someone else's development, but, I try not to miss mine as much as possible. My elaborations on "Cauchy difference", "Geometric moving average", "Moving average with fading period calculation in proportion 1/N" (I do not remember the name exactly) are placed in kodobase with other people's priorities. Haven't I had enough? Why does everyone seem to be jealous of this fact?
 
Dmitry Fedoseev:

That wasn't the question, it was about their algorithms.

I got acquainted with Widler's algorithm. It is based on the differences of the Hig and Low of the last two bars as (+D) and (-U) . They are accumulated, moving averages of each D and U lines are calculated and their difference is ADX. I have a completely different approach to the construction of the two lines B and M, the difference of which gives the indicator DDA, analogue of ADX. but constructed on different principles. With the application of the difference between the two lines ends the similarity of ADX and DDA. And I have not found a close analogue to DA in Widler.
 
Олег avtomat:

As far as I understood from the previous one, the indicator simply gives cumulative sums of increments of different directions. You can improve the results by introducing a "forgetting" operator for the old history, which slowly reduces the old irrelevant values to zero.

Now that's a sensible idea! If the "forgetting" algorithm is correct, the results are much more interesting.
 
Yousufkhodja Sultonov:
I do not pretend to develop someone else's work, but I try not to miss my own, if possible. My developments on "Cauchy difference", "Geometric moving average", "Moving average with fading period calculation in proportion 1/N" (I do not remember the name exactly) placed in kodobase with others' priorities. Haven't I had enough? Why does it make everyone cringe?

You have missed the main point: instead of looking for a black cat (in semi-synthetic forex quotes) it is suggested that your analytical mind be used for its intended purpose - on a real exchange with a much larger and more diverse flow of information, apart from the price.

Personally, I am not offended by any of your ideas. Why do you think so? It's just a pity when such a mind goes to waste. This is all an IMHO opinion, of course.

And this is a continuation of the "banquet":


 
Yousufkhodja Sultonov:
Oleg, the problem is not worth a damn. Is it difficult to assign in the code to take for calculations only the last N bars of history, than to bother with "forgetting"?

Yusuf, you don't see the problem, so it's not really there for you. But it is there. "Take only the last N bars of history" is not difficult, but that is not the problem.

If at each bar (which you have every minute) "take for calculations only the last N bars of history", each time the reference point will change, and therefore the indicator lines on each bar will show values relative to this new reference point. That is, the whole picture will change on every bar.

Try it and you will see.

 
Dmitriy Skub:

You have not understood the main point: instead of looking for a black cat (in semi-synthetic forex quotes) it is suggested to apply your analytical mind to the real exchange with a much larger and more diverse information flow, apart from the price.

Personally I am not confused by your ideas. Why do you think so? It's just a pity when such an intelligence goes to waste. All this is IMHO, of course.

And this is a continuation of the "banquet":


Thanks Dimitri for the suggestion to extend the scope of research by a real exchange. But, I will not be able to finance hypothesis testing projects on the real with a salary of an assistant professor of a university of 250$. And again, all development would be theoretical in nature. Exit the situation could be working remotely as part of a wealthy investment company as a developer of various market projects. But I have not yet received any such offers.
1...91011121314151617181920212223...49
New comment