From theory to practice - page 1210
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I have been able to pinpoint extrema very accurately.
The only problem is that the price sometimes "flies away" further and very sharply.
The only question is that the price sometimes "flies" further and very sharply.
The question is how to determine it mathematically.
I understand... but I want to apply a tricky technique.
I don't want to regress with ANC to find the most optimal line. I want to use the opposite method. I need to build a line where MnC is set for example <=100 points.
Then the line will be adaptive and its period will fully depend on how the price behaves.
in this way i will be able to register only those moments in time, where the price is the most cyclical.
So if you're looking for a cyclic pattern, it means you want some kind of a flat channel. The approach is interesting, but I think it will be easier to choose a sampling period on automata taking into account that the incremental value for a regression line will be very small. It means that if we have a trend and built regression on a trend the increment value will be large, and on a flat period it will be almost zero. Then I think you will understand the point. Having the width of the channel you know the limits, knowing the limits you can calculate the risk....
Hmm, it means that if you are looking for a cyclic pattern, you want a flattened channel of some kind. The approach is certainly interesting, but I think it will be easier to select the sampling period on automata taking into account that the incremental value for the regression line will be very small. It means that if we have a trend and built regression on a trend the increment value will be large, and on a flat period it will be almost zero. Then I think you will understand the point. Having the width of the channel, you know the limits, knowing the limits you can calculate the risk....
the sampling period is a misleading direction of research, I've explored it up and down, there's nothing there.
because the sampling itself is not justified in any way.
sampling should be fully adaptive and based on market movements, not the other way around...
on what basis do people take a sample of 100-200 samples or 50?
There's no answer. Because they always do it off the top of their head.
A lot of statistical textbooks are written about validation.
and that can be avoided altogether. By doing it the way I suggested...
I have been able to pinpoint extrema very accurately.
The only problem is that the price sometimes "flies away" further and very sharply.
When the cycle is at least minimal, it gives more than 100% a month with very little slippage.
The question is how to determine it.
If the trades have reached their peak value (like a trend), the next move is to wait for the close of the trend, then the trend is over and you may work out a week of your flat.)
the sampling period is a false line of enquiry, I have explored it up and down and there is nothing there.
because the sampling itself is not justified in any way.
Sampling should be fully adaptive and based on market movements, not the other way around...
So you adjust the slope of the regression line to the period, I thought you could figure it out.
So you adjust the period to the slope of the regression line, I thought you could figure it out.
It's not that I figured it out, I implemented it two years ago.
I even calculated the degree of slope and it was useless.
I didn't just figure it out, I implemented it two years ago.
I even calculated the degree of inclination - it was useless.
Well, the degree is important from the observer's point of view, but in the algorithm it is much more correct the size of the increment per unit time.
Well, the degree is important from an observer's point of view, in the algorithm it is much more correct the size of the increment per unit time.
You're absolutely right. But it's wrong to do the regression line increment...
...it's all about sampling as without doing the adaptation we will always be lagging the signal.
yes, it might get lucky sometimes. but it's wrongWell, I'll work out an algorithm...
On what basis do people take a sample of 100-200 samples or 50?
There is no answer, because they always do it from scratch.
But these results work only for the optimization period. If you optimize for another period, you get different results.