Random Flow Theory and FOREX - page 35

 

to Prival

So this phrase of yours IHMO is wrong

I know there are limitations, but I meant it:

To apply wavelet transform the issue with sampling rate is even tougher, it should ideally be equal to infinity.

Exactly, I will have to work with what I have. I'm OK with (H+L)/2 for prognoses/runs on a clock, and for more accurate representation of series (for my purposes), I use the clock again, but I get a count on one hour bar by stat-processing all minutes for that hour, namely:

[Open{60}, Low{60}, High{60}, Close {60}} - a total of 240 digits to calculate one value per hour

Unfortunately it is not sufficient in FOREX, that's why the gaps are there.

I have no other data and unlikely there will be. Although I came across an American firm that assures you that it provides data from one of the major trading floors. But that's not the point anyway. And the gaps will remain anyway. That's just the way it is.

The only way to decrease the gap (it will become not 40 points, but 38) is to gather a large amount of quotes providers (I wrote about it). But anyway the gaps will remain, maybe their value will be only slightly smaller.

MQ seems to collect its quotes in the same way, Rosh told about methods in some thread and even attached the exol table with primary sources.

P.S. Do not leave a forum, as a result of disputes sometimes truth is born. Only do not concern military as Budyonny who sent cavalry on tanks in 1941. You can meet intelligent and competent people among them sometimes.

I am a former sergeant myself and sometimes I have a sin - I pull my sword and start waving :o)

to Yurixx

Sergey, why does this link http://grasn.narod.ru/002.djvu not work?

I searched my site for extra stuff and accidentally moved file to another directory. It works now.

 

- You think badly of me!
- I don't think of you at all.

Think about it:) And everything will fall into place.

 
+1, specifically for NorthernWind.
 
Mathemat:
+1, specifically for NorthernWind.

Yeah, thank you, you're a real eye opener. Just wanted to ask for some links to interesting discussions.
 
Yurixx:

Archive of ticks since 2000 by year and month: http://ratedata.gaincapital.com/


Help to understand that there in a file is written, not all it is clear. i give an example has taken 1 week of december, 2007

363533816,GBP/USD,2007-12-02 17:00:24.000,2.054500,2.055000,D
363533888,GBP/USD,2007-12-02 17:01:30.000,2.054600,2.055100,D

I wonder what the numbers 363533816 and 363533888 mean

time interval between ticks 17:00:24.000-17:01:30.000, in this example 1 min. 6 sec.

The most surprising thing is if this is a tick archive then why is there two prices for a tick 2.054500 and 2.055000 ?

and what is D I could not figure out

 
Who knows, Prival. The first number looks like the standard Windows system time (the number of seconds since 01.01.1970), but it's a bit small (it should be over a billion). The two prices are bid and ask, I think.
 
Mathemat:
The fic knows, Prival. The first number is similar to the standard representation of system time in Windows (the number of seconds since 01.01.1970), but it's a bit small (it should be more than a billion). The two prices are bid and ask, I think.


I don't remember where I know it from, it must have come from space. :-)

The first big number is the unique tick number in the total tick stream of all currencies. The two quotes are also bid and ask, I think.

 
Yurixx:
Mathemat:
Fuck knows, Prival. The first digit looks like a standard representation of system time in Windows (the number of seconds since 01.01.1970), but it's a bit small (it should be over a billion). The two prices are bid and ask, I think.


I don't remember where I know it from, it must have come from space. :-)

The first big number is the unique tick number in the total tick stream of all currencies. The two quotes are also bid and ask in my opinion.


yes, the first is a unique number in the quotation stream.

the last letter, usually D, is unknowable.

 
Prival:
I need a procedure which calculates coefficients of a and b in equation y(x)=a*x+b using MOC. Then maybe I'll be able to cobble together some curve ACF algorithm in MQL again

#define MAXHIST 200 // maximum length of the analyzed history
#define MAXPOW 10 // maximum polynomial degree
#define MAXPOW2 20 // MAXPOW*2

extern intPow=3; //rank of polynomial
extern inttern HistLength=24; //history length


double xx[MAXPOW]; //coefficients of polynomial


void CalcPc() //calculate polynomial coefficients
{
int i,j,k,N,H;
double a[MAXPOW,MAXPOW], b[MAXPOW], c[MAXPOW2];
double x,y,f,hd;

for (i=0;i<MAXPOW;i++) //zero the arrays
{
b[i]=0; c[i]=0; c[i+MAXPOW]=0; xx[i]=0;
for (j=1;j<MAXPOW;j++) a[i,j]=0;
}
N=Pow+1; H=HistLength;
for (i=1;i<=H;i++) {
f=1.0; x=i-1;
y=(High[i-1]+Low[i-1]+Close[i-1]+Open[i-1])/4; //входные данные
for (j=1;j<=(2*N-1);j++) {
if (j<=N) {
b[j]=b[j]+y; y=y*x;
}
c[j]=c[j]+f; f=f*x;
}
}
for (i=1;i<=N;i++) {
k=i;
for (j=1;j<=N;j++) {
a[i,j]=c[k]; k=k+1;
}
}
for (i=1; i<N; i++) {
for (j=i+1; j<=N; j++) {
a[j,i]=-a[j,i]/a[i,i];
for (k=i+1; k<=N; k++) a[j,k]=a[j,k]+a[j,i]*a[i,k];
b[j]=b[j]+a[j,i]*b[i];
}
}
xx[Pow]=b[N]/a[N,N];
for (i=N-1; i>=1; i--) {
hd=b[i];
for (j=i+1; j<=N; j++) hd=hd-xx[j-1]*a[i,j];
xx[i-1]=hd/a[i,i];
}
}

double CalcdYdX() //derivative
{
CalcPc();
return(xx[1]);
}


A snippet from the program. Polynomial of any order (theoretically, in practice it's better to limit it to 10-times). Will this do?

 
shar:
Prival:
I need a procedure that calculates the coefficients of a and b in the equation y(x)=a*x+b using MQL. Then maybe I'll be able to build some curve ACF algorithm in MQL again
..... Will it do?

Thanks.