Adaptive lookback indicators - page 23
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
A friend with some really great ideas requested this, and i got it all except emphasized rsi
, but still like it and its making some 
all ready.Its alb jurik smoothed rsi with 2 ma averages of that rsi. The rsi has 3 methods regular,rsx,and wilders. The alerts and arrows happens when all 3 lines cross.The picture showing the 3 methods H4 period on Hi chart.
This is upgrade of above post the arrows are still the rsi and 2 ma cross, added divergence lines and trend lines, seems it may be a possibility using it along side smoothed ichimoku.
Causation v Correlation
Bump...
If this is a bad idea or not worth the effort - I'm good.
But, if it was lost in the huge amount of work going on, I thought I'd bump it (it has been over a month) to see what moderators / everybody thinks.
The idea seems to have merit, at least to me.
Yes, we've seen autocorrelation holds that near lags are well-correlated. (Thanks Mladen for that excellent work AND illustration of variance and covariance in code - beautiful stuff)
However, correlation is not causation.
I think we'd be remiss for not testing distributions as causation.
What do the esteemed Elite group and moderators think?
Or, have they already spoken and I just didn't get it?
Not sure whether to post this to 'Adaptive' or 'Dynamic' thread
Mladen, how about picking a few of your favorite adaptive and/or dynamic indicators and modify them to use the distribution of values to vary / adapt / 'dynamize' the indicator?
At present, the (adaptive/dynamic) modification of the indicator is driven by a (an adjustable) near term window or smoothing.
Instead, scan the previous n (100s, 1000s ?) bars and create deciles (octiles, n-tiles ?) to hold the distribution counts.
If not too difficult, add the capability to use (A) center or (B) +/- n-standard deviation values
In this way, the indicator is driven by...
(A) the statistical 'nature' of the indicator and not merely the recent past (as in, perhaps the indicator's 'nature' or real tendency is reflected in its distribution not just where it was 14 bars ago)
-OR-
(B) reversals (or even breakouts, I suppose) played off of extreme values based on the +/- n-standard deviation.Bump...
If this is a bad idea or not worth the effort - I'm good.
But, if it was lost in the huge amount of work going on, I thought I'd bump it (it has been over a month) to see what moderators / everybody thinks.
The idea seems to have merit, at least to me.
Yes, we've seen autocorrelation holds that near lags are well-correlated. (Thanks Mladen for that excellent work AND illustration of variance and covariance in code - beautiful stuff)
However, correlation is not causation.
I think we'd be remiss for not testing distributions as causation.
What do the esteemed Elite group and moderators think?
Or, have they already spoken and I just didn't get it?Hi Adeo,
I believe it has merit, just to know how to go about doing it.
Hi Mr MLaden
Thank you for this one, can you tell me what is the MA tracking the price and is is possible to have a parameter to hide it . Sorry i'am talking about Mogdalef bands.
many thx
Flytox
It is a linear regression value (or lsma as some may call it). The simplest way to hide it is to set the color of it (7th color) to none (that way the code does not have to be changed at all)
regards
Mladen
Hi Mr MLaden
Thank you for this one, can you tell me what is the MA tracking the price and is is possible to have a parameter to hide it . Sorry i'am talking about Mogdalef bands.
many thxSimba,
Here is the adaptive look back version of Mogalef bands.
If you look into the code you will notice one difference compared to the indicator posted in the elite section earlier. In this one the slope used for intersection projection is a real linear regression slope (in the previous I used an approximation of the last 2 bars slope) Difference is not big (in most cases, when periods are short (like in default Mogalef indicator), it is neglectable), but thought to have a "to the letter" linear regression and bands intercept in this one. Also, it seems that the alb is indeed adding to the indicator (but that is my first impression only) Here is just a comparison : upper is the regular Mogalef band
PS: it is not centered yet. Will see if I can cook up some extrapolation for a linear regression value (without having it "too linear"
) that would logically belong to it
regards
MladenMladen,
Thanks a lot...Try them on EU and GU M15-H1 with DeviationsPeriod=30-40 and SwingCount=1-2
Regards
S
Flytox
It is a linear regression value (or lsma as some may call it). The simplest way to hide it is to set the color of it (7th color) to none (that way the code does not have to be changed at all)
regards
MladenYou know what, I'am using MT4 for years now, never noticed this "none" color. Thx Mr Mladen.