Hidden divergence - page 56
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
Wrote a message to your script. Maybe you can help with https://www.mql5.com/ru/forum/110344/page4? and dragging and dropping the line is very handy.
and this to the previous post
The trend consists of the following Formations:
- Movement Formation:
High[i+1]>High[i+2]&&Low[i+1]>Low[i+2] - Bullish Formation, and
High[i+1]<High[i+2]&&Low[i+1]<Low[i+2] - Bearish Formation ,
- further Pause Formation:
High[i+1]<High[i+2]&&Low[i+1]>Low[i+2] - Compression Formation, and
High[i+1]>High[i+2]&&Low[i+1]<Low[i+2] - Extension Formation,
- then a Correction or U-turn Formation:
High[i+1]<High[i+2]&&High[i+2]>High[i+3] or Low[i+1]>Low[i+2]&&Low[i+2]<Low[i+3] - Extremum Formation.
The coordinatesof Close and Open are irrelevant.
Next Analysis:
Each complete bar is considered together with the previous one and a Formation is determined.
To enter a position you need to open it in the direction of the Movement Formation.
In order to pause or close a position, the bars of an Expansion Formation must
have moved beyond the preceding Formation
Low[i+2]<Low[i+3] for a Bull Formation or High[i+2]>High[i+3] for a Bearish Formation or
reversed Formation through Extremus Formation
High[i+1]<High[i+2]&&High[i+2]<High[i+3]&&High[i+3]>High[i+4] or Low[i+1]>Low[i+2]&&Low[i+2]>Low[i+3]<Low[i+4].
Thus,using a simple visual analysis you can stay in the trend for a long time without using any indicators.
-I agree with Geronimo, similar method (he already gave a link to it) works well on M1-10 and H6-W1.
That they take the trend out of the price is for sure. I tried to take it out with averages and regression, and of course all this with normalization, but I could not get the values they have. And in general I have a suspicion that they calculate not the correlation coefficient but the determination coefficient R^2. It would be cool to be able to calculate the correlation and divergence as they go along and set the optimal period without any adjustment using the tester. But optimal periods vary so much... I tried to optimize periods in the tester from 1 day to a week with one day step. But the results were very unattractive. And of course in the market from February till about 2 weeks ago the divergence had the best results, but unfortunately it is a tester adjustment. I would say that the market fluctuations are always changing and there is no way to catch this bastard.
to ANG3110
Thanks! very valuable experience, - I will save on this direction!
-my attempts to optimize on the fly only worsen the result (it's true that the attempts themselves were not enough))
It seems that heuristic methods, i.e. programmatic approaches - are simply useless, and behind the curtain some theories are used.
You are such monsters! You have no time to think it over before you save your time and replies :)
Maybe another tip? For example, we take the last 1000 bars, find the average bar volume Av, calculate S=P*Av, start reverse calculation of volumes, the bar number at which the sum >= S becomes the indicator period for calculations...... did you understand? :)
There is also an option with number of movements in +- by minutes...... It's just that if you've tried it, you'd also like to save some money :)
to rider
I see. Tried it a long time ago and gave it up - the result is nasty, especially in the candles)))
Well, who are these candlesticks for?
it was a bit more interesting to build an equi volumetric MA, however
you lose the connection to the natural periods of the market !!!!! and have difficulties interpreting MAs.
- well there is a crossover, but it's not time bound(((
It's like the same thing if you plot a topo map by live volume.
Well, you monsters, no sooner have you thought about it than you've already answered and saved time :)
Can you suggest something else? Did anybody tried to create periods on the basis of volumes: the idea is the following - the period of indicator in parameters - P, for example, take the last 1000 bars, find the average bar volume - Av, calculate S=P*Av, start reverse summing of volumes, the number of bars where the sum >= S becomes the indicator period for calculations...... I understand? :)
There is also a variant with number of movements in +- by minutes...... Just if you have tried it, I would also like to save some money :)
I haven't tried adapting indicator periods by volume. I tried to adapt them by standard deviation and linear regression angle. Here are the figures: OsMa 9/21/5 - hours, on the upper one the normal one on EMA, on the lower one with the same periods, but adapted by Std.
-
-
And these are MACD - on the top the standard one from MT4 package, on the bottom one adapted by linear regression.
-
Despite some improvements on the purely technical side - smoother and more unambiguous, the problem of not falling into resonance is not solved by this.
ANG3110 wrote (a) >>.
Thank you.>> MACD, looks kinda nicer....
OsMa 9/21/5 - hours, on the top normal on the EMA, on the bottom with the same periods but adapted by Std.
and the horizontals on the OsMA are also calculated somehow?
Are the horizontals on the OsMA calculated as well?
It simply uses -1/+1 rationing and the indicator readings are in relative units of -1 to +1. But of course there is a provision for withdrawal of normalization. The periods are fixed to the hours, in order not to change readings when switching from one timeframe to another: p = hrma * 60 / Period(); where hrma is the period in hours.
It simply applies a normalisation range of -1/+1 and the indicator readings are in relative units of -1 to +1. But of course there is also provision to remove the normalisation. The periods are bound to hours, so that at switching from one timeframe to another the readings would not change: p = hrma * 60 / Period(); where hrma is period in hours.
I'm not good at arithmetic anymore )).... Can you give me the formula to calculate this relative unit?