Request for Help on Pattern recognition AFL's

#16
Kalyan,
I am not a TS user. But have tried it for a while.

TS do not allow to refer in future.
But there is workaround for this "software of workarounds"...
combine 3 functions... refer back, highest high & displace to get what you want, you may also have to create array.

OR just study open code, pattern recognition 'ready systems'.

OR may anybody else will be able to guide.
 

kkseal

Well-Known Member
#17
Kalyan,
I am not a TS user. But have tried it for a while.

TS do not allow to refer in future.
But there is workaround for this "software of workarounds"...
combine 3 functions... refer back, highest high & displace to get what you want, you may also have to create array.

OR just study open code, pattern recognition 'ready systems'.

OR may anybody else will be able to guide.
From what i know the actual zig-zag code can be accessed via the PowerEditor window (possibly in TS6 or higher versions), though personally i do not have any idea of TS or it's features.

The TS zig/zigzag has one important advantage in that it does not allow the right (last) leg to be formed till the percent retracement (as specified by the user) is made. This takes care of the vanishing right leg problem*, which is why i'm interested in this particular implementation. This form of the zigzag coded in Ami/MS would be a better one to use than the native implementation.

I have some idea of the algorithm, but the actual code would obviously provide better guidance and also save time.

Please see if you can access the code (whether in EL or C) in TS. And thanks for your response.

Regards,
Kalyan.

* The position of the zig pivot can still get extended & slope of the zigzag line change, however, with incoming data & as such should be deemed provisional.
 
Last edited:
#18
Although I was looking for zig in TS but never found it.

If you g00gle, you'll find many sites that displays "different codes, for different soft, for single indicator".
even there you'll notice that any indicator that uses zig, have workaround code OR use of dll s instead of zig in TS.

one example is here...it uses zig in MS but dll for TS.
http://www.tsresearch.com/support/online_help/tsexcellink/indicator/

**********************

My idea was just to create Zig from scratch, in TS (the same way used in MS or AB). It'll have dynamic last leg. & you are looking for something else.

BTW. there is some AFL in AB's online library that tries to produce Valid Entry and Exit Points avoiding this problem of last leg. I dont have the neat idea but you can have a look.
http://www.amibroker.com/library/detail.php?id=353
 

kkseal

Well-Known Member
#19
Not impressed with the Ami code. If you notice the valid Buy/Sell signal hinges on the Close of the Pivot Low/High, but the Pivot Low/High is determined by looking 1 period into the future. (But many people do seem to be quite impressed; so maybe there's something to it after all)

My proposition is simple - to create a better zig-zag rather than try various kind of manipulations on the existing one. And as far as i see there are two major problems, the 'vanishing leg' & the 'unanchored pivot'.

Regards,
Kalyan.
 
#20
I myself never read that AFL. just saw the small description.



There is difference in point of view.

I work on Zig as a tool to implement & back test (manually) custom elliott calculations. For me dynamic last leg is not the problem. Just working on making my formula lighter on system, as its slow for tick data.

Seems you are looking the solution for better zigzag indicator.
(IMHO its not there in TS...better way is to clear the idea (construction) in mind & convert it in to AFL.)

Regards
 

Similar threads