"Miracle", "digital" "group" movement indicator - page 12

New comment
 
and silence.....
 

Why delete ..... I try to talk to people who may have touched much wider areas of analysis than multicurrency analysis.

what's wrong with that .... it is easier to discuss with people than alone ...

 
Deleted what? I have everything as it was... or was it your posts deleted?
 
I do not understand why such aggression, I do not force anyone to respond to my posts, but I am trying to find like-minded people who are also interested in different ways of analysis.
 
I will try to gather my thoughts later and write about the direction I am trying to analyse myself....
 
There is no aggression. There is fantasy. There is a misunderstanding of the issues.
 
trol222:
I will try to gather my thoughts later and write about the direction I am trying to analyse myself....

We will be waiting. Just a bit more constructive.
 
The author of this thread has disappeared. I'll try to put in a few words of my own.
 
Who thinks about the following: if you build for example 4 tick-bars from ticks, in the multicurrency analysis it is not quite correct, because in every pair of ticks come in a different interval in time and this interval changes, I counted as the speed of angle-intensity change - i.e. for example there is the Euro-buck and Euro-wen and Dollaren of each of these pairs, for example for Minutes the square root of the sum of squares of ticks and square of change in pips, then we divide the change in pips (rise with + or fall with -) by the square root value - we get the cosine and the cosine of Euro-wen subtract the cosine of Dollaren, then it is ahead or behind the cosine of Euro-bucks thus it is a kind of advance We proceed the same way with other pairs that include euro and dollar (for eur\usd it's gbr\usd-eur\usd, eur\chf-usd\chf....). And it doesn't matter who is leading, it's important who is currently leading the series of pairs that make up the Eurobucks (maybe the Eurobucks will be ahead of the gbr\usd-eur\usd, or maybe eur\chf-usd/chf, it doesn't matter what the lead is). However, the EUR and USD exchange rate cannot be the same, because in pairs it is not clear which currency is pulling the other one unless you look at it against some benchmark which may be going against all currencies at the same time. 2- What is the best, if possible, to do it with gold (in pairs it is hard to understand which currency pulls the others, as all currencies cannot move against each other and gold may move up or down against all currencies simultaneously).
 
trol222:


Prival 07.10.2010 23:18


You contradict yourself. On the one hand you say that you know how to get crosses from majors. On the other hand, you don't know enough about the majors to get the crosses from the majors.

But you need a complete matrix to identify currency movements.

Judging by the answer, I understand you say that if a more complete matrix is important, it is not the prices that are calculated using the crosses, but the speed of reaching these prices that can be observed.

through the tick volume, i.e. through the frequency of arrival of price data.... please correct me if I'm wrong.

Yes, very similar. Close by almost. I don't want to correct, I want to clarify. As much as some here claim that they only need the majors. They will calculate the rest. Practice shows that they are silent about one thing their calculation accuracy is +-lapot (to the accuracy of spread/2). Here is an example https://www.mql5.com/ru/forum/132599/page2

Now about speeds. At school we often solved problems (remember) where there was a path, velocity and acceleration. Knowing these parameters, say, for a car, we can assume with a probability not 50/50, but a little more, let it be 85 to 15, that the car rushing with great speed and acceleration, is more likely to continue the way, than turn around. Am I right....a if so, then we need to find analogues of these concepts in forex (just imagine that this is not a graph of euro/dollar, but how a car (plane or let's say fly is flying) moves). Calculate and build a forecast...

At first glance the task is simple, but when you start going deeper into it, you understand that everything is not so simple. You have to remember, when you digitise any analogue value there is noise. This noise is called noise of sampling and quantization. So they exist (noise calculation formulae are available somewhere here), so before you need to get rid of them. It can be done sequentially, we can build a filter and pass quotes through it. And then analyze it. Or we may make a joint analysis (take into account the presence of noise), I chose this way. I describe the movement with stochastic differential equations and it means Kalman filtering here in details. https://www.mql5.com/ru/forum/105740/page15

I always try to forecast price and price is (asc+bid)/2. The more accurate the forecast the better and more perfect TS you can build, there is research here on the forum, look for how the forecast affects the quality of TS, it enters in the profit in the 4th degree if I remember correctly.

And the third - we must not forget that we do not see the net movement of euro or dollar, but can only observe the projection of this movement on the plane euro/dollar-time or the plane euro/pound-time etc. so I build an index of currency movements of euro, dollar, pound, etc. And it's very important to close the space to have the whole matrix (all projections), if you calculate them through the majors then shit happens (although this is understandable, mathematics is an exact science, it's in statistics there are confidence intervals there you can +-spread count...) in mathematics can not.

Well, in short as that so ...

1...56789101112131415161718
New comment