Prediction of turning points based on ZZ on fractal calc....

[Kilian19]

Prediction of turning points based on ZZ on fractal calculation

Good evening fellow traders.

I am here to share an indicator concept which was developed by gg53 over at Forex Factory. It's not like a simple indicator but more like a system which consist of pattern
probability analysis and filters to narrow down the probability of a trend reversal. (keywords: neural network and self learning). It's probably the most accurate indicator I have seen in my years of trading (and I am being realistic here and not one of those guys who wants to get rich quickly with any kind of martingale).

Lets get you all started with some pictures, please note that the arrows are not repainting and appear at the close of the candle:

Images of live prediction:

eurusd.m5_post.png

eurusd.m5_post2.png

eurusd.m5_post3.png

eurusd.m5_post4.png

As you see the prediction is not always pinpoint accurate, but this is as good as it gets. There are countless of chart examples like this.

There are a few indications that those images are not just cherry picking:

• about half of the charts he posted were with live predictions which all turned out to be true.
• He always posts real time charts when you ask him
• The author doesn't try to sell the system
• I was part of the early stages of the development, made some suggestions and have seen the forecasting ability of the system improve

Now since everyone is hyped here comes the mayor drawback (of cause there has to be one). The source code of the indicator is not publicly available! The logic somehow is, “gg53” the creator of the system has shared his thoughts and rules throughout the thread. (up to 80 – 90% of it at least).


I am able to do the necessary coding for this indicator and already started but I can't get my head around every instruction gg has given us. I need a 2nd or 3rd person to look over some sentences and tell me if they read something else out of what the author has said than I do.

Bonus + drawback?: Nothing good comes without labor and as much as I can say this indicator is the real deal. It is self adapting as it takes in past patterns with putting more weight to newer occurrences. I won't give any information yet about how this system works as I really recommend everyone interested in spending some hours reading the whole thread at FF. http://www.forexfactory.com/showthread.php?t=440372 It is important to get the basics and to be honest it's quite interesting as this thread is not based on trading or pips but rather about math.
Of course it will take some time but in my opinion the development is definitely worth it.

Is anyone interested to join me?

Regards Kilian19

[phil_trade]

Hi Kilian

The thread begin with :

This thread will prove to everyone - once and for all - that it is possible to know where price will go on any time frame.

I will let you get in at the very start of trends on all time frames and you will also know specific numbers that a given bar on a particular time-frame MUST hit. I will post pictures for a while just so you can verify the results for yourself and try and figure out how I did it.

The algorithm I am using does not run on Empty4, therefore, this is a thought-provoking thread.
In this game : results speak the truth! Look forward to this!

so, where does this ForexGT_Arrow indicator come from ? Could you post it here ?

I will run deeper in the thread deeper this week end.
cheers

Philippe

[Kilian19]

so, where does this ForexGT_Arrow indicator come from ? Could you post it here ?

I will run deeper in the thread deeper this week end.
cheers

Philippe

The calculations are take place outside of Empty4. "GG" is using C# or C++ if I am correct and I am exporting my data into Java. The results are than imported back into Empty4 to draw the arrows.

No I can't post it here since this is the reason I created this thread. We have to develop the indicator for ourself as the source code is not publicly available.

I created this thread so we can work together to decrypt the algorithm which is stated over at FF. There isn't a single post containing all the information we need. Tiny breadcrumbs of the algorithm can be found all alone the way.

[milanese]

Hi Kilian,

first a welcome too from me at the SHF
Thank you for this thread, the project seems very interesting, I will use the weekeend to read to ful thread on FF.

Cheers

Tommaso

[phil_trade]

so, where does this ForexGT_Arrow indicator come from ? Could you post it here ?

I will run deeper in the thread deeper this week end.
cheers

Philippe

The calculations are take place outside of Empty4. "GG" is using C# or C++ if I am correct and I am exporting my data into Java. The results are than imported back into Empty4 to draw the arrows.

No I can't post it here since this is the reason I created this thread. We have to develop the indicator for ourself as the source code is not publicly available.

I created this thread so we can work together to decrypt the algorithm which is stated over at FF. There isn't a single post containing all the information we need. Tiny breadcrumbs of the algorithm can be found all alone the way.

I can also read data from Empty4 and stock them in SQL HF to do complex calculation in Windev and then draw object with Empty4. More with ForexTester, I can backtest with M1 data since 2006. So first thing, is to decrypt the algorithm and write down rules on a PDF.

Mike and Milanese are excellent for this job

[Kilian19]

I can also read data from Empty4 and stock them in SQL HF to do complex calculation in Windev and then draw object with Empty4. More with ForexTester, I can backtest with M1 data since 2006. So first thing, is to decrypt the algorithm and write down rules on a PDF.

Mike and Milanese are excellent for this job

Exactly. The implementation of the actual algorithm comes at the very end of our journey and how we do it actually doesn't matter at all. The top priority is to get a very well defined set of rules. Also note that there is a training period required maybe 1 month - 1 year of data prior to start of predictions. One mistake people have made at the beginning was to test the forecasting ability of the algorithm on in of sample data (the data which was used to train the algo). But back tests are just valid if they are performed on out of sample data. Anyways I am getting ahead of the topic again ...

Attached you will find a pdf of all of "GG's" post. I actually didn't want to upload it so early but now I will do it anyways. Just note that it doesn't help to read just this pdf as the basics are described in the 300 posts prior to this pdf.

first a welcome too from me at the SHF
Thank you for this thread, the project seems very interesting, I will use the weekeend to read to ful thread on FF.

.

Actually I am already part of this forum for more than a year I just haven't actively contributed. The more people join the project the better right

[kwanann]

Looks interesting, reminds me of the itching ea in this forum

[zTB]

Give it a go, gentlemen. Nothing ventured, nothing gained.

But I do not know why this thread has reminded me of the Trillion Dollar Bet.

Z
aka Mrs. Killjoy

[Kilian19]

Just give me a heads up when you are done with the thread and we can start outlining the basic steps needed to create this algorithm.

[dudest]

Wow, a great gathering of minds!, thanks Kilian

A super addition to the minds looking at this would be: slipshod, aka Andrew Sumner [ he of LibSSSR fame (based on fractals), http://www.stevehopwoodforex.com/phpBB3 ... 796#p11796 ]

Gary (garyfritz) would also be a super team-mate to have. I will PM them both and ask them to have a look.

I cut my teeth learning to code MQL this year on a LibSSSR-based EA (very much a work-in-progress), so super interested in this as well. Will read through the thread and come back to here.

Cheers

libsssr in action.png

.