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

VlanFx · Post by **VlanFx** » Sat Nov 22, 2014 5:56 pm

gg53 » Wed Sep 24, 2014 7:28 pm wrote:
crodzilla » Thu May 22, 2014 2:03 pm wrote:
Someone mentioned to me that it is a form of normalizing... like what is done with data for a neural network. Longer bars that are not "normal" lengths actually make the zigzag finish sooner, so that is why we need to add lengths to the whole zigzag.

Carl
That's NOT what I'm doing...
ZZ length remains the same.
If i'm getting a longer bar (double or triple the avg) I consider it as two or three bars in the COUNT of bars between Fractals and between ZZ end point.
Example:
If I have 3 bars, and one of them is double in size, I COUNT it as 4 bars.

G.

Hi everybody,

here is another indicator that I use. It is based on the quoted GG's post. It calculates how many bars a bar should be counted as, based on its size in relation to the ATR of previous 5 (user defined) bars.
Hope it helps.

Vlan

BarCountByAtrEURUSDM5_20141120.png

pedma · Post by **pedma** » Sat Apr 11, 2015 7:29 am

VlanFx » Sun Nov 23, 2014 12:56 am wrote:
gg53 » Wed Sep 24, 2014 7:28 pm wrote:
crodzilla » Thu May 22, 2014 2:03 pm wrote:
Someone mentioned to me that it is a form of normalizing... like what is done with data for a neural network. Longer bars that are not "normal" lengths actually make the zigzag finish sooner, so that is why we need to add lengths to the whole zigzag.

Carl
That's NOT what I'm doing...
ZZ length remains the same.
If i'm getting a longer bar (double or triple the avg) I consider it as two or three bars in the COUNT of bars between Fractals and between ZZ end point.
Example:
If I have 3 bars, and one of them is double in size, I COUNT it as 4 bars.

G.
Hi everybody,

here is another indicator that I use. It is based on the quoted GG's post. It calculates how many bars a bar should be counted as, based on its size in relation to the ATR of previous 5 (user defined) bars.
Hope it helps.

Vlan

Hi Vlan,
Thanks for this nice indicator. May i know, what must we do if we get 0 (zero) result ? We can see that there is no "orange" histogram at some places. We count it as "0" or as "1" candle ?
Thanks
edit: May i know how you calculate it ? Do you use MathFloor (rounding down) or MathCeil (rounding up) ? Example: we have 2.1234 pips as the result of ATR(5), calculate from i+1 to i+5. Current closing bar (i+0) has 5.1 pips (Hi - Lo). So we get 5.1 / 2.1234 = 2.402 candles. Using MathFloor we will get = 2 bars, but using MathCeil we will get = 3 bars. Thanks in advance.

VlanFx · Post by **VlanFx** » Tue Apr 14, 2015 7:50 pm

Hi pedma,

if there is no histogram bar, then you ignore that candle, as if it wasn't there.

I use MathRound().
MathRound(2.49) == 2
MathRound(2.5) == 3

Vlan

pedma · Post by **pedma** » Fri Apr 17, 2015 5:13 am

VlanFx » Wed Apr 15, 2015 2:50 am wrote:Hi pedma,

if there is no histogram bar, then you ignore that candle, as if it wasn't there.

I use MathRound().
MathRound(2.49) == 2
MathRound(2.5) == 3

Vlan

Thanks Vlan.
How do you handle the 0 (zero) count candle that has fractal ? Please see attachment.
Do you eliminate that candle totally (including its fractal) , or take the fractal but do not take the candle (because it has 0 count) or take the candle and the fractal (although the candle count is zero) ?
Thanks in advance

VlanFx · Post by **VlanFx** » Mon Apr 20, 2015 6:01 am

As far as I know, you take count of bars (sliced) between fractals. So in your example, you would have 3 bars.

pedma · Post by **pedma** » Tue Apr 21, 2015 7:25 am

VlanFx » Mon Apr 20, 2015 1:01 pm wrote:As far as I know, you take count of bars (sliced) between fractals. So in your example, you would have 3 bars.

So, do you mean that the fractal at a 0 (zero) candle count (after sliced) is still valid? Thanks Vlan.

VlanFx · Post by **VlanFx** » Tue Apr 21, 2015 4:56 pm

Good thing you brought this up pedma. I checked with GG53 and he said that even though the slice count for a bar is zero, it should still be taken and also it's fractal. In other words, minimum count for any bar, no matter how small, should be 1.

Attached is updated BarCountByAtr that reflects this logic.

pedma · Post by **pedma** » Wed Apr 22, 2015 3:02 am

VlanFx » Tue Apr 21, 2015 11:56 pm wrote:Good thing you brought this up pedma. I checked with GG53 and he said that even though the slice count for a bar is zero, it should still be taken and also it's fractal. In other words, minimum count for any bar, no matter how small, should be 1.

Attached is updated BarCountByAtr that reflects this logic.

Great, this is good news Vlan. Thanks a lot

And thanks again for your attachment

Sharky · Post by **Sharky** » Sun Sep 13, 2015 2:09 pm

Did you saw GG53 latest results on real account @penguintraders "1 pip compounding" thread?
can we join forces and try to replicate this?

gregorious · Post by **gregorious** » Mon Apr 18, 2016 3:01 pm

Good day everyone. I see that this thread has died. Hopefully we can bring some life back to it. I am confused about something. Why is it when gg53 says that his indicator doesn't use something, we continue to try to make that something work? I would think it would be better to follow or uncover his method exactly. If anyone has a working model, please contact me. After talking to gg53, I believe I have two pieces of the puzzle that I am not seeing anyone talk about. I am not a coder so I need help in this area.