Requests & Ideas - page 44

 

constant volatility bars

mladen:
Every missing candle (and only then you are really going to notice it) is stretching the extrapolated values as if it was rubber band. Some new general solution for that issue (so mtf-ing the predicted values) has to be found and right now nothing meaningful comes to mind

Maybe instead of subjectively blacking out certain candles or news events before you feed price to the Goertzel browser, you can use a concept called "instrinsic time" (also "theta time"); It's one of the discoveries of Olsen & associates: This was originally published by Dacorogna et al. (1993b). The concept is simple: instead of sampling the rate at constant points in time (every 1min/1H/...), you try to "de-seasonalize" the time series in a way that time speeds up when the information flow rises. This probably makes you think of renko/crb/equal volume charts, but these techniques destroy spectral content.

Equal volume charts are based on the tick rate, but individual broker volume is not exactly the most objective measure to base a strategy on (imho). This is different; you can call it is a sort of "constant volatility bar". The short-term volatility of a 1-minute (1H/...) chart could be an input for creating a new kind of chart: One that produces more bars if there is more volatility. It's like you "spread" the calendar/seasonal effects evenly over the X-axis to get "constant seasonality bars".

http://www.olsen.ch/fileadmin/Publications/Working_Papers/950920_NonLinTimeScale.pdf

OlsenWorld: Published Papers

 

Simba

I really benefit from your comments and I'm sure I'm not the only one. Thank you for going to the trouble.

Cheers

Rodney

SIMBA:

...

Yes,I know some of you are here for the indicators and not to be taught how to trade,so,skip my messages,but there are many others that appreciate my suggestions when using cyclical tools for trading,and those may benefit from my comments.

...

 
MrM:
Maybe instead of subjectively blacking out certain candles or news events before you feed price to the Goertzel browser, you can use a concept called "instrinsic time" (also "theta time"); It's one of the discoveries of Olsen & associates: This was originally published by Dacorogna et al. (1993b). The concept is simple: instead of sampling the rate at constant points in time (every 1min/1H/...), you try to "de-seasonalize" the time series in a way that time speeds up when the information flow rises. This probably makes you think of renko/crb/equal volume charts, but these techniques destroy spectral content.

Equal volume charts are based on the tick rate, but individual broker volume is not exactly the most objective measure to base a strategy on (imho). This is different; you can call it is a sort of "constant volatility bar". The short-term volatility of a 1-minute (1H/...) chart could be an input for creating a new kind of chart: One that produces more bars if there is more volatility. It's like you "spread" the calendar/seasonal effects evenly over the X-axis to get "constant seasonality bars".

http://www.olsen.ch/fileadmin/Publications/Working_Papers/950920_NonLinTimeScale.pdf

OlsenWorld: Published Papers

MrM,

Very interesting,just one question....What would be the difference between volatility and ATR....?If none or very similar,the end result will be similar to using CRBs,which as you pointed out,are of very limited use with cycles.

We can try using "normalized range"(which I presume will be very similar to normalizing by a higher TF ATR) or "implicit volatility"(Doable,but not precisely easy to code,since we would need to link with Futures options markets),but,besides those, I do not know how to escape the link between atr and volatility...JM2Cents.

Regards

S

 
SIMBA:
What would be the difference between volatility and ATR....?If none or very similar,the end result will be similar to using CRBs,which as you pointed out,are of very limited use with cycles.

This would be DRB: dynamic range bars. CRB's are defined as a constant number of pips, not as an ATR/variance/volatility multiplier. I'm reading the paper right now to know exactly what they meant.

We can try using "normalized range"(which I presume will be very similar to normalizing by a higher TF ATR) or "implicit volatility"(Doable,but not precisely easy to code,since we would need to link with Futures options markets),but,besides those, I do not know how to escape the link between atr and volatility...JM2Cents.

I think this is more about short-term realized volatility than about implied vol. Michal from mqlservice did the first CRB script for me, he told me it would be easily adaptable to change from #pips into a variable. Now I need to do my homework and see what that variable is ...

 

Hello mladen.

When you have time please take a look this indicator

and if it is possible to make multi time frame version

please make it for me.

thanks

http://www.multiupload.com/AO27J5LI45

Regards

t2g

 
t2g:

Hello mladen.

When you have time please take a look this indicator

and if it is possible to make multi time frame version

please make it for me.

thanks

http://www.multiupload.com/AO27J5LI45

Regards

t2g

t2g

That is a hull moving average and it is a part (hull moving average as an option) of averages indicator from the elite indicators thread and also there is one more non-repainting version posted here, but, for the sake of making it the same as the one you have (in the averages I do not allow averages method change since by definition hull moving average is calculated using linear weighted moving average) here is one that calculates only hma and allows you to choose moving averages method too (as in the one you have).

So, it is a multi time frame and, as far as colors are concerned, it is a non-repainting one (so, do not be surprised on occasional differences in color - this one does it as it should)
regards

Mladen

Files:
hma.gif  24 kb
hma.mq4  8 kb
 
MrM:
This would be DRB: dynamic range bars. CRB's are defined as a constant number of pips, not as an ATR/variance/volatility multiplier. I'm reading the paper right now to know exactly what they meant. I think this is more about short-term realized volatility than about implied vol. Michal from mqlservice did the first CRB script for me, he told me it would be easily adaptable to change from #pips into a variable. Now I need to do my homework and see what that variable is ...

MrM,

Is the Kase bar same or similar with what you are trying to achieve?

http://www.traderinterviews.com/content/KaseBarsHandout.pdf

Video Interview: Kase Bars

Kase Bars Explained - A New Type of Bar Chart

 

Goertzel Cycle Distribution

I tried, but got lost in the process...

Would someone be interested in modifying Goertzel to loop through the bars on the chart (or extern thereof) and count the number of times a period count presents?

The array dimension could be resized with each new, larger period number.

At the end, you would have an array of size n (where n was the largest number of periods found in a cycle) representing the distribution of periods that present in the cyclebuffer[] array of the current indicator.

After all the counting was done, the array contents would be output to a .csv file.

Obviously, the purpose of this exercise would be to analyze the dominant cycles.

This could be used to preferentially select cycles from the current list of the current indicator.

The current indicator display changes as new, dominant (higher amplitude) cycles are determined. Preferentially specifying the cycle list, especially during selected, critical times - say, the time near a trade entrance or exit - would be more stable for automated (EA) use.

Such output would also be useful in comparing trade instruments (e.g., EurUsd v GbpUsd).

Such 'forcing' of the Goertzel does defeat its overall purpose. However, if used sparingly and for short duration (that is, tactically) it would avoid having an aspect change at a critical time. For example, while waiting for a precise entry point, the presentation may change from 19-period dominant to 83-period dominant. One cannot simply use the cycle list to achieve this aim as the interesting period may move up or down in the list.

To change the capability, one must loop through the cyclebuffer[] array, match the interesting period, get the cycle index and populate the cycle list (manually or programmatically).

Of course, an alternative would be an option to specify a list of interesting periods in much the same way the cycle indices are specified now. If the interested period is found in the rank-ordered list, it is included in the Goertzel output. Said another way, this would be a method to filter noise with preference given to periods that had previously been determined to have a higher incidence in the distribution.

A next question might be, "Can we do that for the more active portions of the more active sessions." since that is often the most desired time to trade.

Of course, all this may be moot; it is relatively easy and straightforward to learn the common periods by being "Men Who Stare At Lines" - but, it would be nice to have objective output instead of passed out goats.

 

Hello mladen.

thanks a lot for your quick reply.

I appreciate. you are the best.

I will check your indicator.

Thank you.

t2g

 

Hello mladen.

I tried to make an EA by using HMA indicator which you gave me.

I want to get signals when HMA changes color.

I coded like this.

double HMA_up1 = iCustom(NULL,0,"HMA",1,1);

double HMA_up 2= iCustom(NULL,0,"HMA",1,2);

double HMA_down1 = iCustom(NULL,0,"HMA",3,1);

double HMA_down2 = iCustom(NULL,0,"HMA",3,2);

if (HMA_down2 != EMPTY_VALUE && HMA_up1 != EMPTY_VALUE) ====> BUY;

if (HMA_up2 != EMPTY_VALUE && HMA_down1 != EMPTY_VALUE) ====> SELL;

but EA doesn't trade as I wanted.

I can't understand which buffers I have to use and how to code them......

I just want to get buy sell signals when HMA changes color.

I'm beginner of coding. please teach me.

sorry for my stupid question.

thanks

t2g