Ward 6 - page 59

New comment
 
It's pathetic what this USE has done to me.
 
My point is that the question of constructing three lag-free filters from three graphs simultaneously is not the same as constructing three filters from three graphs separately. In the first case you have an additional coupling equation (linking the non-lagged filters as well as the prices), which is absent in the second. Thus, problem 1 is simpler than problem 2. It may be that problem 2 has no solution, while problem 1 has. Isn't it obvious?
 
Dr.Drain:
Landau L.D., Lifshits E.M. Course of theoretical physics, vol.1. Mechanics. Pages one and two. Concept of number of degrees of freedom. Sit and read.

In essence, in answer to the question from here:

https://www.mql5.com/ru/articles/1351

Of course, it would be more interesting to reduce currency movements to one or more independent variables - it would greatly simplify the task of restoring the market attractor and predicting quotes.

Note, the number of independent variables is always smaller than the number of graphs in question :-) it cannot be otherwise, because there are coupling equations. Thus, the task gets really simple.

 

I, like DmitiyN, don't understand either. Let's take two variables: x, y. And move them independently and randomly in time. At the same time we will calculate the 3rd dependent variable z=x/y. So, the doctor claims that somehow we can predict x and y.

Anticipate the response, "But x and y are correlated, not 100%, but at least somehow." If they were 100% correlated, then z=x/y would be a constant. Right? So we take x=EURUSD, y=GBPUSD (somehow correlated with each other) and get z=x/y=EURGBP. In the case of 100% correlation between EURUSD and GBPUSD, EURGBP was a horizontal straight line. But it is not. We may assume that EURGBP deviations from the flat straight is noise and trade on EURGBP returning to some constant value (or some smooth curve, whoever wants it), which dozens of scalpers such as commercial FAP TURBO already do. It works well if the spreads are small.

 
gpwr:

Dr claims that somehow we can predict x and y.

Absolutely not. There is no way to predict anything. Once again, read the above:

Dr.Drain:
I'm wrong. I thought I mentioned at this point that we don't have to build a linear filter, we will build a non-linear filter, but of course physically feasible, i.e. without peeking ahead.

I am in no way talking about peeking into the future. I'm talking about not lagging. It is not the same as being ahead of time. Not at all the same.

 
gpwr:

We can assume that EURGBP deviations from a flat straight line are noise and trade on EURGBP returning to some constant value (or some smooth curve, whoever wants it), which dozens of scalpers such as the commercial FAP TURBO are already doing.

One can assume that. And trade this way. However, not for a long time.
 
gpwr:

I, like DmitiyN, don't understand either. Let's take two variables: x, y. And move them independently and randomly in time. At the same time we will calculate the 3rd dependent variable z=x/y. So, dr claims that somehow we can predict x and y.

Woot, and if there was a dickfix here, he would correct that the equation should be z=x^a*y^b. But there are also some problems with prediction, or even non-delayedness.
 
DmitriyN:
The question arises: where can the additional information come from?
From where you have not one graph but two. How much is enough. There is as much information in the GBPUSD chart as in the EURUSD. When you display only EURUSD, you have 1 graph. When you examine the triangle EURUSD, GBPUSD and EURGBP, you have two independent charts (generalized coordinates, i.e., the ratio of the euro to the square root of the pound and the cube root of the pound to the dollar; however, there are two independent charts.) Enough about that. If it was obvious, I wouldn't allow myself to comment on it :-))) The score is going up, you say?
 
DmitriyN:
A MA built on the EURUSD chart using the values from the GBPUSD and EURGBP charts, what will not lag?
How you can twist a thought, it's amazing...
 
DmitriyN:
You are placing positions in the direction of the filter line, as far as I understand. But, how do you determine their number, the maximum lots of positions?
I don't put positions in the direction of the filter line. As many as you can. I'm putting positions in the direction of the filter return. I've already been asked about it and I was discussing the noisy operation of numerical differentiation in principle and the fact that the initial curve itself is not suitable for it because it's not smooth. I simply cannot know the "direction of the filter". It does not exist. The number of positions, lot size, pairs to open, when etc. - everything is out of the question. Does it matter? The market will average everything.
1...525354555657585960616263646566...78
New comment