Machine learning in trading: theory, models, practice and algo-trading - page 1086
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
Minkowski or Hausdorff dimensionality gives only understanding of volatility inside fractal formation, and explication that fractal sets have fractional dimensionality, that they are like not line and not plane but something between :D or evaluation of how curve fills the plane
Then there is a notion of generalized Brownian motion, which can take different fractal dimension (from 0 to 1), it is more interesting there. Like the fractal structure of generalized Brownian motion is market memory (Hirst) and the fractal dimension of Minkowski is a random component (noise) that not only affects but also modifies the structure (in contrast to classical interpretation of noise in econometrics)
I thought about your opinion, probably you're right, the fractal dimension can only show some properties of the symbol on the history, but it does not define even the current state, i.e. with the help of fractal dimension you can only learn if it is worthwhile to try to conduct the research about the regularities for this time series or simply there are none
I think they are not worth attention as the fractal dimension can be estimated by the same ZigZag with less computational costs, moreover the codehttps://www.mql5.com/en/code/9676 implies the choice of period, the choice of analysis price (High,Low,or Clkose...) and you cannot decide which price is more important for determination of the fractal dimension
.... If you want to estimate the fractal dimension, you'll have to choose a period and go to Low or High, or Clkyose.
Learn from Misha how to breed)))
https://www.mql5.com/ru/forum/281573
Misha, I asked for .....
AHAHAHA))) fUrMulator of the task))
Where are you my black-eyed where in volgde where volgde where volgde volgde. TAAAAAAAmmmmmmm where is the forest polisade :-). I wiped away a stingy man's tear.... And passions do not cease in a branch about the MO :-)
п.1. Note the avatar, well, not me.
2. But, there is, there is.
Help about where to look - p.1, etc.
;)
Look at the avatar, not me, that's for sure.
the head from... the dawn.
the head from... the dawn.
Ohhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh easy Maxim, we are all cultured people. Let's keep a brand of intelligence that casual passerby looked looked what nice and well-mannered people and here we tell him "Fuck you" so he really shook his head. Otherwise you just shoot him in the face and no intrigue. Sorry :-)
Ohhhhhhhhhhhhhhhh easy Maxim, we are all cultured people. Let's keep a brand of intelligence that casual passerby looked looked what nice and well-mannered people and here we tell him "Fuck you" so he really shook his head. Otherwise you just shoot him in the face and no intrigue. Sad :-)
he's just offended.
apparently the euro has fallen...
I've been working on the neuron, it's understandable in principle.
But I don't like algorithms, all based on statistics and images...
It's all the same old thing.
thought about your opinion, probably you are right, fractal dimension can only show some properties of a symbol on history, but does not define even the current state, i.e. with the help of fractal dimension one can only learn whether over the given time series it is necessary to try to conduct research on regularities or they simply are not there
I think they are not worth attention as the fractal dimension can be estimated by the same ZigZag with less computational costs, moreover the codehttps://www.mql5.com/en/code/9676 implies the choice of period, the choice of analysis price (High,Low,or Clkose...) and you cannot decide which price is more important for determination of the fractal dimension
.... methinks, I've written a hundred times and probably will say it again, I haven't seen more useful than the standard indicators from MT4 delivery, the same RSI will be more informative than the analysis of fractal dimension
It is necessary to search for exactly how to work with scale invariance and fractal formations, how the "memory" works - it makes sense, imho. Even the interpretation of classical indicators in this context can take on a new meaning. The big problem will be the processing of non-periodic "cycles", whereas in the B-M f-iia, for example, they are periodic and it is not difficult to predict it even by a simple aryma. And dimensionality is just volatility, by and large.
Something can be learned here http://tpq.io/p/rough_volatility_with_python.html, but I haven't gone deeper as I don't have time yetAnd dimensionality is just volatility, by and large
There, I remembered ZigZag, you are volatility, but the point is the same, that even for statistical studies fractal dimension is meaningless
Got some ideas on statistics, basically this is going to be an intraday volatility study. I think we should still look for simple solutions, roughly how often the price reverses by updating the weekly high/low or monthly high/low? How many times a month the high/low is updated and the price crosses the opening price of the month? ....
it is necessary to search how exactly to work with scale invariance and fractal formations, how "memory" works - it makes sense......
Yes yes, that's where you should have started actually...
1) I tried with "DTW" but it's a very heavy construction and I didn't manage to finish the experiment, because I couldn't think out some conceptual things in the algorithm...
2) Maybe we should break the BP into fourier harmonics? there are three parameters: amplitude, frequency and phase. then we can translate these parameters not in a numerical form but in a proportional one, for example
We have amplitude (amplitude range of motion), take for example the first harmonic and the fifth one, the amplitude of the first one is 100p and of the fifth one 10p. We divide one by another and obtain a universal characteristic - proportional.
Those are not 100 and 10 amplitudes, but just the knowledge that the first amplitude of the first harmonic is 10 times greater than the fifth, and this is a universal measure that will be repeated in the future and is essentially objective. But of course it would be better if some experienced spetrator expressed his opinion on this matter.
What's going on with your networks? If you do not want to help me to translate my indicator into Metatrader, you may even write a robot there and get an indicator + new knowledge on how to use networks in trading, a working knowledge