Machine learning in trading: theory, models, practice and algo-trading - page 2088
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
What if you think about it?
A bandpass filter is a filter that passes a given range of frequencies (bandwidth)
to perform such a filter, you must first at least perform the Fourier transform
the physical meaning of MACD is a delta between two price averaging devices, each of which is focused on a different time interval, and they are compared now
oscillator? So, it's a streaky one.
On the vertical frequency, there can be several peaks at one moment, if there are for example 5, you need 4 LF filters to separate them from each other. With each bar these peaks shift in frequency or disappear at all.
A FLEXIBLE model is one that is FIXED by simple rules from a simple one.)
Simplicity should not kill the effectiveness of the model. To our rams. Signs should include consideration of unidirectional price movement and the supposed effect on the price of the news, whether the forecast came true or not. The simple news price movement is not informative enough.
Simplicity should not kill the effectiveness of the model. To our rams. The signs should take into account the unidirectional movement of the price and the supposed influence of the news on the price, and whether the forecast has come true or not. The mere presence of news is not sufficiently informative in terms of price movement.
Then maybe the signs should be based on the zigzag - the length of the knees and the time of the tops, for example. This will immediately remove the dependence on useless volatility jumps.
Do Fourier from the increments, you can even do without SSA.
Explain why please, I will kill all the information about trends and large fluctuations
Vertical frequency, at one moment there can be several peaks, if there are 5, for example, you need 4 LF filters to separate them from each other. With each bar, these peaks shift in frequency or disappear at all.
Look at this file, there is a price and 8 PCA components, if you add them all together you get almost the price, I wonder if in this decomposition the row characteristics are also floating like in the normal price or not
Then the signs should probably be based on the zigzag - knee lengths and top times, for example. This would immediately remove the dependence on useless volatility spikes.
Where the news triggered in the opposite direction, yes, but when the unidirectional impact we will have false indications of the beginning of the zigzag. In the direction of the change in speed I think we need to look. The averaging channel of small zigzags. We cannot perform a one-step study. It is necessary to select the cases of the price impact and lack of impact. And we should determine the optimal timeframe for the zigzag.
oscillator? So it's a bandwagon.
No.
What kind of bandpass is it?
Bandpass filters use resonance of the desired bandwidth
Like a neural network where you set the coefficients and get the desired signal
But it is possible to filter without itWhere the news triggered in the opposite direction, yes, but when the unidirectional impact we will have false data of the beginning of the zigzag. In the direction of the change in speed I think we need to look. The averaging channel of small zigzags. We cannot perform a one-step study. It is necessary to select the cases of the price impact and lack of impact. The optimal timeframe for the zigzag would also be better understood.
In any case, the opposite ones will lengthen and the opposite ones will shorten. But in general, yes, we should calculate it; it will not be possible to do it by guessing.
In spite of the fact that there are only four variants of news action (L, M, H and no), taking into account that we have pairs - we have 16=4*4 variants.
In any case, the opposite ones will have to lengthen and the opposite ones will have to shorten. But in general, yes, you have to count, you can't figure it out speculatively.
In spite of the fact that there are only four variants of news (L, M, H and no), taking into account the fact that we have pairs, we have 16=4*4 variants.
Pairs, yes, you have to take news from both countries. I don't get the action options. Strong, medium, small? If so, maybe simplify is or is not. Then there are 4 options.
Couples, yes, you have to take the news from both countries. I don't understand the action options. Strong, medium, small? If so, maybe simplify is or is not. Then there are 4 options.