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

[Kilian19]

Now we only need to get more traffic to that subforum

Attached a small little document which won't give you much information about the algorithm if you have no clue but hopefully keeps you interested and lets people read Steves and gg's former post.


As promised I will come back once every part of the algorithm is clear. When I created this thread I didn't believe it would be so much work to get the algorithm together. Maybe in a month we are good to go. Nothing great and innovative comes cheap huh

Remember to donate

[milanese]

Now we only need to get more traffic to that subforum

watch here http://www.stevehopwoodforex.com/phpBB3 ... 4&start=24
I think it will come

Cheers
Tommaso

[SteveHopwood]

When I created this thread I didn't believe it would be so much work to get the algorithm together.

Hehe. Whenever we read something along these lines we chorus, "Yep. Know exactly how you feel."

[Kilian19]

When I created this thread I didn't believe it would be so much work to get the algorithm together.

Hehe. Whenever we read something along these lines we chorus, "Yep. Know exactly how you feel."

For me this happens on a regular scale . If I want to program a web crawler, or want to try out a little idea I always end up putting much more time into the project than intended. There are just so many ideas coming to my mind and either I overestimate my programming skills or I underestimate the problem

The countless of hours thinking about this little monster already payed itself. I got in touch with so many great minds and the hunt really developed into a new hobby of mine. I enjoy it, I have fun a challenge to work on and if one day I also get monetary compensated for it that would be the little icing on all of it.

After all social contests are the most valuable asset you can have after all

Do you remember the post when you read that higher timeframes look exactly like smaller timeframes?

this might open your eyes:

http://www.youtube.com/watch?v=lmxJ1KDR_s0

Highly recommended

[gg53]

Please try to donate to some charity/organization helping the Philippines.

It really is a small thing to do and it can really make a difference.
In the current situation there - that small thing can mean the difference between life and death.

poocher send me this:
Deal News through this link will match any donation made:
http://www.redcross.org/cm/dealnews-pub

I'm working on receiving info on more credible organizations that are directly working in the Philippines right now.

I'll post more links as soon as I receive them.

Thanks you in advance.

G.

I agree. milanese brought your post to my attention. I have created a new sub-forum at the top of the Board Index, created a new thread in your name and moved your post to it.

Nice one, my friend.

Thanks, my friend.

G.

[kesk]

Basically I am a learner in forex so lurking here, trying to assimilate knowledge. Have opened an a/c with globalprime 2 months ago, but still couldn't trade full due to fear.

I was interested in Neural networks for trading and have read some papers and articles.

The posts here seems to have some relation to NN, isn't. Instead of us trying to decode the DNA sequence or the learning info, why couldn't NN try and learn the same.

And based on the learning, one can make prediction.

Am I making sense here?

[Kilian19]

It's not directly a neural network, but more of a self adapting trading system.

A program is not intelligent.

Instead of us trying to decode the DNA sequence or the learning info, why couldn't NN try and learn the same.

If you say it like that it sounds so simple. But we can't simply say

Code: Select all

NeuralNetwork n = new NeuralNewtwork();
n.input(Data);
n.doMagic();
signal = n.output();

We have to tell the program what to do with the data and to do that we have to understand the logic before we can tell out electronic circuits what to do.

If you want to read about NNs I recommend this page:

http://natureofcode.com/book/chapter-10 ... -networks/

It's really interesting and even shows how to implement a NN.

It's really easy to understand and pretty clear. It contains information about fractal geometry as well.

[scheinwerfer]

I've been playing with this turning point prediction for some time, but the pieces don't seem to click together.
Signal generation as by GG53 is

• 1. fractal pattern match

• 2. probable length match of ZZ leg

• 3. opposite direction bar prediction

With pattern match it is easy, it is either current or the longer version of current pattern(current+1 or current+0). If the database is of 2 years of data, then it is almost given that there is a pattern match during each bar of the ZZ sequence. Length, therefore, is of most importance!

GG says to look for most probable total length. This means we are actually looking for that specific length, where the respective pattern has ended statistically more than not. Therefore, we can get a length match only in that region. If current ZZ leg overshoots or stops short of that statistically significant region, the signal will surely be a miss. It can be tradeable, but it might not. Furthermore, longer patterns amplify the possible gap between signals and the real ZZ leg end. That is the reason why signals can occur in the middle of nowhere. Does anyone agree with me on this?

[gg53]

The ZZ leg is our "unit of measurement" to analyse its components (fractals & "total length"=bar count). It's NOT our target. Our target is the trading SWING point. Sometimes it's the same and other times it's not.
The ZZ leg end point is the "Highest high" which not always is the best entry point on bar close.
So, many times you visually see a "miss" - but if you look at bar close as your TRADING entry point - it's not.

G.

[intcust]

very interesting work you might want to read CJ naked trading on how patterns repeat on every time frame