Machine learning in trading: theory, models, practice and algo-trading - page 2089
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
the crowd interprets the news in different ways, and the quote is generally in its favor
All of this is not the purpose of the study.
All of this is not the purpose of the study.
I saw a professional study on this subject, I think it was a branch
there were screenshots on the forum
sorry I do not remember when
either on 4-rka or here
most likely on the 4-pc.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.
Exactly right - short for Low, Medium and High Impact Expected. There's also an N for Non-Economic - holidays and time transfers.
Perhaps it is possible to simplify, but you need further research of the issue - for example it may turn out that Medium in relation to the dollar is stronger than High for another currency.
Another problem - one day there can be a lot of news about one country - different in time, strength and direction.Somehow it all has to be carefully "brought to a common denominator.)Explain why please, I will kill all the information about trends and large fluctuations
Look at this file, there is a price and 8 PCA components, if you add them all together you get almost a price, I wonder if in this decomposition the characteristics of the series are also floating like in the normal price or not
The price has such a spectrum; if we take Fourier from the price the maximum amplitude will always be for the slow fluctuations.
The increments have such a spectrum, the constant component and frequencies that are out of the window are automatically removed.
Probably float, that 2d picture is from matlab, I took it down long ago.
The price has such a spectrum, if you take the Fourier from the price, the maximum amplitude will always be in the slow oscillations.
The increments have such a spectrum, the constant component and frequencies, which are outside the boundaries of the window, are automatically removed.
Probably float, that 2d picture is from matlab, I tore it down long ago.
Norm
how do you use the spectrum?
Exactly right - short for Low, Medium and High Impact Expected. There is also N from Non-Economic - holidays and the time translation.
Simplification is probably possible, but you need additional research - for example, it may turn out that Medium for the dollar is stronger than High for another currency.
Another problem - in one day may be a lot of news for one country - different in time, strength and direction.Somehow it all has to be carefully "reduced to a common denominator.)Everything cannot be realized at once, of course. Besides, the choice of the correct TF scale is also a problem. What can be seen on the hour, is difficult to see on the minutes, and vice versa. We need the logic of a sequence of simple studies. The task cannot be called simple.
norms
how do you use the spectrum of increments?
Meditating))
By eye it is not clear what to do with it, perhaps the MO will find
The price has such a spectrum, if you take the Fourier from the price, the maximum amplitude will always be in the slow oscillations.
The increments have such a spectrum, the constant component and frequencies that are outside the window are automatically removed.
Kind of like price increments have always been flicker noise (pink noise).
Obviously, you've discarded the most necessary thing about the price - its memory. Hehe...
Can MO cash in on price increments? Yes and no. The constant expectation bias for any sample gets in the way.
Kind of like price increments have always been flicker noise (pink noise).
Obviously, you have discarded the most necessary thing in price - its memory. Hehe...
Can MO cash in on price increments? Yes and no. The constant expectation bias for any sample gets in the way.
Didn't notice a 1/f decline on the incremental spectrum.
I only threw out the constant component and the trend.
In this case the increments are needed only for the cutoff frequencies.
I forgot you can also subtract the mean from the gradients.
If someone makes a sample with news - I'm ready to run in CatBoost with different parameters - this topic is interesting to me.
I tried to change type of trawl depending on expected news - the effect was good, but somehow sample collection was not automated.
And yes, according to my research waiting for news better works than its coming out - smoother that way...