[Tams]

Originally Posted by mtorri »

Hi all.

I'm also searching for an easylanguage implementation of the center of gravity bands, also known as Hurst bands.
I have the following code which works well in proprealtime.
I tried to translate it in tradestation, it looks easy, but the results are wrong.
My problem is the DPO function used in the prorealtime code.
I substituted it using the detrend function in tradestation but it's not the same.
So I'm not able to go on.
Anyone can help?
...

do you have the code for DPO ?

[mtorri]

unfortunately not, i searched for it without sucess.
The prorealtime platform, describing how DPO works, says it does exactly what the tradestation detrend function do.
But if you plot the DPO you can see two big differences compared to the detrend
1) it is shifted on the left by LEN/2 bars (where LEN is the input length)
2) it is normalized so that it moves form -100 to 100

here is an example of the DPO and hurst bands on prorealtime.

[mtorri]

Originally Posted by moneymarkets »

Thanks for your input, would you mind posting what you have for easylanguage so far, to see if we can help?
thanks

This is the tradestatione code

A function:
---------
Inputs: p3(numericsimple), XX(numericsimple), HurstM(numericRef),HurstH (numericRef),HurstL(numer icref);

Variables: k(0), de48(0), n(0), p(0), d100(0), moy100(0), co(0), hh(0), x(0), w(0), aa(0),b(0),e(0), z(0),
alpha(0), bbeta(0), flag(0), p33(0), ii(0),cc(0), dd(0);

k=p3;
de48=Detrend( CLOSE, k*2);

if de48=de48[1] and de48[1]=de48[2] and de48[2]<>de48[3] then flag=1;
n=(k*2)-4;
p=(n/2)-1;
d100=Detrend(close,n);
moy100=close-d100;

co=(moy100-moy100[1]+(close[p])/n)*n;
//co=(moy100-moy100[1]+div(close[p],n))*n;

if flag[1]=1 and flag[2]=0 then hh=co[1];
if flag[1]=1 then co=hh;
n=mod(p3,2);
p=(p3-n)/2;
p33=(2*p)+1;

//once x=0
w=absvalue((p-x)/p);
//w=absvalue( div(p-x,p) );

w=w*w*w;
w=(1-w);
w=w*w*w;
x=x+1;
if barnumber=p33 then
begin
aa=0;
b=0;
e=0;
for ii=1 to p33 begin
z=barnumber-ii+1;
aa=aa+w[z];
b=b+w[z]*(ii);
e=e+(ii)*(ii)*w[z];
end;
end;
if barnumber>p33 then
begin
cc=0;
dd=0;
for ii=1 to p33 begin
z=barnumber-ii+1;
cc=cc+co[p33+p-ii]*w[z];
dd=dd+co[p33+p-ii]*w[z]*(ii);
end;
end;
//alpha=(aa*dd-b*cc)/(aa*e-b*b);
alpha=div((aa*dd-b*cc),(aa*e-b*b));

//bbeta=(cc*e-b*dd)/(aa*e-b*b);
bbeta=div((cc*e-b*dd),(aa*e-b*b));

HurstM=alpha*(p+1)+bbeta;
if barnumber<p33*2 then
begin
HurstM=0;
HurstH=0;
HurstL=0;
end;

HurstH=HurstM+xx;
HurstL=HurstM-xx;

AA_BandeHurst=0;
--------------------

and its indicator:
-------------
inputs: p3(83), XX(0.75);

Variables: HurstM(0), HurstH(0), HurstL(0);

Value1= AA_BandeHurst(p3, XX, HurstM, HurstH, HurstL);

Plot1( HurstM, "HurstM" ) ;

Plot2( HurstH, "HurstH" ) ;

Plot3( HurstL, "HurstL" ) ;
-----------

I also attach 2 images on the same bund future contract daily where I show the bad results on tradestation and the nice hurst bands on prorealtime.

And here is the prorealtme code creating this bands
I shifted the upper and the lower band only by 0.75 so the bands are not "perfect".
But the most important thing is the central line.
-----

k=p3 //(input value)
de48=DPO[k*2](close)
if de48=de48[1] and de48[1]=de48[2] and de48[2]<>de48[3] then
flag=1
endif
n=(k*2)-4
p=(n/2)-1
d100=DPO[n](close)
moy100=close-d100
co=(moy100-moy100[1]+(close[p])/n)*n
if flag[1]=1 and flag[2]=0 then
hh=co[1]
endif
if flag[1]=1 then
co=hh
endif
n=p3 mod 2
p=(p3-n)/2
p3=(2*p)+1
once x=0
w=abs((p-x)/p)
w=w*w*w
w=(1-w)
w=w*w*w
x=x+1
if barindex=p3 then
a=0
b=0
e=0
for i=1 to p3
z=barindex-i+1
a=a+w[z]
b=b+w[z]*(i)
e=e+(i)*(i)*w[z]
next
endif
if barindex>p3 then
c=0
d=0
for i=1 to p3
z=barindex-i+1
c=c+co[p3+p-i]*w[z]
d=d+co[p3+p-i]*w[z]*(i)
next
endif
alpha=(a*d-b*c)/(a*e-b*b)
beta=(c*e-b*d)/(a*e-b*b)
lowess=alpha*(p+1)+beta
if barindex<p3*2 then
lowess=undefined
endif
lowess1=lowess+xx
lowess2=lowess-xx
Segno=SGN(lowess-lowess[1])
return lowess as "H media", lowess1 coloured by (Segno) as "UpperBand", lowess2 coloured by (Segno) as "LowerBand"
-----------

[moneymarkets]

Thanks for your information and post.
I get an error when trying to verify the below statement--needs more inputs???
alpha=div((aa*dd-b*cc),(aa*e-b*b));

[Trendup_]

Did you find the COG easy language version?
Can anyone share it?