What is the optimum depth of history for identifying a useful signal? - page 20
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 see you haven't had a conversation with your son-in-law...
With an ex-son-in-law. He is not a bad psychiatrist, but he does not know anything about methods of dimensionality reduction of optimization problems.
I will tell you a terrible secret: with smooth target function the dimensionality of the problem of searching for optimal parameter values can be almost halved very easily. The most serious result of dimensionality reduction is achievable when the whole optimization interval is divided into many pieces with two intervals in each. This eventually results in drawing of a multidimensional cube with 2 intervals for each edge. Since the optimization problem always comes down to comparison of these intervals, the computational complexity of the algorithm is proportional to the perimeter of the multidimensional cube.
It is used in dynamic (on-the-fly) optimization of parameters of something; sometimes this process is called adaptation because a system that can optimize itself in real time is obviously adaptive.
Because the smallest perimeter has a square.
1-The degree of two in that post was clear, this degree is not only 32 but also 16 and 64, why is 32 was the optimal ratio of signal detection and computational costs. And on what timeframe such a compromise was achieved, certainly not on the 5-minute timeframe. Meanwhile, the author of that post suggested using a large timeframe since my count is long. He limits himself to 32 bars of history for the forecast stipulating that he counts half a second for 64 bars and he wonders how long it will take to get 720 bars. So what's so wild about it if he already uses a large TF (has not yet said which one).
2- There is such a thing the chpc method, implementations have both common and private bikes, and what...
3-Turning to your curves on tangents, there are graphical ways where you do not need to average anything, the curves may jump in absolute values, but there are zones where they do not change (a certain context) after leaving these zones, we look where the movement went, and drive, because in most cases it will continue, it is obvious that both methods reduce to about the same.
But if it is possible to fit your curvatures into a multicurrency cluster (at a stage after index construction), then I don't think it will be easier to interpret these curvatures in a simultaneous joint analysis. Although I don't know, they will simply draw entry points and that's all.
My curvatures have not been discussed. They have another purpose.
And this thread is about Kotelnikov's theorem. It's a bit rambling, but it's about it.
I don't think they were talking about my curves. They have a different purpose.
Your opinion of your exclusivity will ruin you)))). phew phew phew phew phew phew phew phew pow pow pow pow pow pow pow pow pow pow pow.
Asleep?