Price Predictions and Probabilities

[Jayzon]

Was messing around and the drawing looks kind of interesting as shown in the image, but after some more discussion with Lowphat, I realized that I'm missing the Mid TZ zones based on my current logic. I'm going to have to rework this project and draw the potentials based on price instead of high and lows. I think that I see what is going on better now hopefully though.

[simplex]

hi all, my first post here!
i coded the RecurrenceStatistic indy that lowphat posted here (he was kind to ask for my permission before posting).
if there would be further ideas or questions regarding it, let me know.
cheers
k

Hi kprsa,
First of all I would like to thank you for sharing your indicator here!

As posted before, I continued working with your RecurrenceStatistic V4.2 indicator posted by lowphat. Now trying to understand your algorithm I applied a minor modification and got some weird results. I simply added two new buffers to display the upper and lower border of a transient or fractal zone your algorithm identified on a certain center bar. I did not touch your core logic at all, just try to display what according to my interpretation are the upper and lower border of the zones in focus.

My modifications in short:
- removed unused buffer Mean[]
- added buffers upperBorder[], lowerBorder[]
- added variable tzDisplay

Upper and lower fractal zones seem to be allright.
When display mode is switched to one of the transient zones' borders things become a bit more complicated. Some seem allright, some do not. See attached pic to show two examples:

RecurrenceStatisticV4_2_sx.png

This behaviour may have one of two reasons, IMO: I misinterpreted your use of the variables, or there's something wrong in the part of your algorithm that identifies the TZ bars. To be honest: I did not understand this part of your algorithm fully.

I attach my modified code to this post, so you might have a look at it if you like to. When looking for my modifications inside the code just search for comments containing simplex.

Cheers, Jürgen

EDIT:
My point is not to ask you for help in coding a new indicator that displays TZ - I already have that.

But I'm afraid - if I did not misinterpret your algorithm - that the statistics displayed just depend on some transient zones that are not being identified correctly. IMO, the fractal zones are all correct. The only reason I added my two buffers to your code was to visualize this without having to deal with those graphical objects.

[Thx]

Thanks for commenting on it Thx, I might release the code for this later when I get some basic stats showing, but I haven't fully decided on that yet. It is still fun messing with it though. I love some old Pink Floyd sometimes too

...

kinda reminds me of pink floyd album (dark side of the moon)

Pink Floyd's We Don't Need No Education somehow makes me think I should have attended more statistics classes during college time ...

Cheers, simplex

Totally Agree ...

:hi:

[simplex]

In the similarity thread over at FF (http://www.forexfactory.com/showthread. ... ost7925594) kprsa agreed to my view that there's a bug in his Recurrence Statistic V4.2 indicator.

The above mentioned thread has been moved to the recycle bin by FF admins and will certainly disappear soon, so I will cite kprsa's answer here:

Quoting duplex
{quote}But I'm afraid - if I did not misinterpret your algorithm - that the statistics displayed just depend on some transient zones that are not being identified correctly.

Hi, you are right, thank you. There was indeed a bug in the stats of the mid-bar zones.

What was (quite stupidly) incorrect in my code was the condition for the cleared mid-bar TZ being resolved involving High/Low instead of Close. There are still minor issues left that I am aware of and will perhaps get back to them once in an appropriate mood.

Anyway, since this changes the stats for the mid-bar TZs to an extent, I will call this update Version 4_3. The pdf from the 4.2 version is still valid.

Voila,
k

By PM at FF kprsa gave me permission to post his now corrected V4.3 here - and so I do. I will also add my modified version that does display the upper and lower boundaries of the fractal and transient zones identified by kprsa's modified algorithm. To me looks perfect now - thank you, kprsa!

The pdf kprsa mentioned can be found here: http://www.stevehopwoodforex.com/phpBB3 ... 80#p104293

Cheers, simplex

[fasttrade]

thanx simplex

[sirius778]

In the similarity thread over at FF (http://www.forexfactory.com/showthread. ... ost7925594) kprsa agreed to my view that there's a bug in his Recurrence Statistic V4.2 indicator.

Simplex, please can you put an alert for red/green arrows on RecurrenceStatisticV4_3.mq4?

[Lowphat]

seen this coming

[Jayzon]

Spent more time with actual trading today instead of coding. I used the PTZ indicator's basic stats (no probabilities yet) for my logic. After 63 manual trades within 14.5 hours, I was able to pull off 137.2 pips across many different pairs on a live account. I hope everyone else did well today.

[simplex]

Simplex, please can you put an alert for red/green arrows on RecurrenceStatisticV4_3.mq4?

Hi sirius,

Thanks for your interest! I do not believe that alerts really make sense for this 'indicator', because it's no classical indicator intended to provide specific trading signals.

Its only intention is to provide statistical data in hindsight about resolving / non-resolving of possible fractal and transient zones. Please bear in mind that the arrows and potential signals would be subject to change (i.e. repainting) until the end of a calculation period (h). This is very much comparable to a classic ZigZag / Fractal Indicator.

The statistical data provided may be used to calculate reasonable entries / TPs / SLs that might have an enhanced success rate based on this statistics. I'm still working on the details of a possible strategy that incorporates this method.

There was a huge thread (The Similarity System, ~14000 posts in 18 months) over at FF covering this methodology. That particular thread seemed to be ending in scams and has been moved to the recycle bin by FF admins. And personally, I never believed in those 97 - 99 % probabilities, with 'approaching 99.9999 %' like some of the FF contributors stated.

Anyway: I am sure that there's a real potential in this method, because it relies on relatively simple, straightforward statistics to enhance success probability. This is why I put some effort in building statistics for different pairs and timeframes about this at the moment. My personal goal is to distinguish between entries that provide a say 65 % win probability and those providing maybe 80 % - based on nothing but pure statistics.

Best thing is: this does not introduce any lag. The statistical probabilities provided are valid at any bar, based on all the bars available on chart and a respective number of potential trend and counter trend entries.

The only thing you must not do is take the arrows painted by such a statistics indicator as entry signals. It would only lead to desaster.

The whole methodology is still new within SHF. The theoretical background has not been discussed here, yet this knowledge is absolutely necessary to use this method / its data successfully. If there's some interest in this topic here at SHF it might be interesting to set up a new thread that deals with it and builds up knowledge.

Personally, I'm still trying to get a grip to this topic (for less than 2 weeks by now). So I feel I'm not the right guy to start this thread by now. If there's anybody around who has followed the topic at FF longer that I did we might team up in setting up such a thread.

Cheers, simplex

[sirius778]

Simplex, please can you put an alert for red/green arrows on RecurrenceStatisticV4_3.mq4?

Hi sirius,

Thanks for your interest! I do not believe that alerts really make sense for this 'indicator', because it's no classical indicator intended to provide specific trading signals.

Thanks for the explanation and advice.