Machine learning in trading: theory, models, practice and algo-trading - page 770
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 1st order autoregressive coefficient coincides with the 1st order autocorrelation coefficient, but further they do not seem to coincide
Well, in the figure on the left you can see, but the charts themselves are different for some reason.
By the way tried to figure out ARIMA algorithm on Python, and got tired of watching how it is calculated by jumping around a dozen or so files plugged in.
Maybe you can somehow catch the whole algorithm, how it all works there?
By the way, I tried to understand the ARIMA algorithm in Python, and got tired of watching how it is calculated by jumping through a dozen or so plug-in files.
Maybe somehow you can catch the whole algorithm, how it all works there?
I don't know, I haven't coded it myself. I've been told there's no fish there so I'm not interested :)
Another thing - it is useless to use akf on non-stationary charts
But why is it useless? You can see the trend on it) For this purpose only a smoothed aper can do.
Well, why is it useless, the trend on it can be seen) Only for this purpose and smoothed atre can work.
or a regular makdak :)
Well, why is it useless, you can see the trend on it) Only for this purpose and smoothed atp can be suitable.
If the residuals are not autocorrelated, then the noise is left, it means that the model is normal
in other views it is useless as an indicator
Maybe read the theory better, I don't know, I didn't code it myself - I was told there are no fish, so I'm not going in :)
This I have already heard, and from you, too. Only after reviewing all the theories and videos, came to the conclusion that one and the same, as a verse memorized say, like parrots.
And in practice no one can make and show an algorithm) an opinion that many do not understand what they say.
If you want to check it, you may load it all into the tester and check it. In this case you'll get tired while calculating every bar with R and Python, plus you'll have to "play" with parameters.
I have already heard it from you too. Only after watching all the theories and videos, I came to the conclusion that one and the same, as a verse memorized say, like parrots.
And in practice no one can make and show an algorithm) an opinion that many do not understand what they say.
If you want to check it, you may load it all into the tester and check it. And so calculating each bar on R and Python you get tired, plus the parameters "play".
Well, yes, but if people say so I believe them... ))) I think they understand
arima works for forecasting periodic cycles, seasonal sales and other stuff. The market has a priori non-periodic cycles (maybe on some constantly growing indexes there will be some effect).
Arima works for forecasting periodic cycles, seasonal sales and other stuff. The market cycles are a priori non-periodic (well, maybe on some constantly growing indexes there will be some effect).
But, again, this is what "people said".) I don't believe it until I check it myself, plus I will make some additions of my own, but for this I need to understand how everything is arranged.
Also by the way there are a lot of examples of machine learning on cryptocurrencies. Why, because a priori there is a trend in them, that's why it works there too.
As soon as the trend is broken all this "intelligent learning" will stop working. But, there must be some truth in MO, you have to look for it....
My understanding of ARIMA was enough to start by calculating AIC with different settings. If I understand correctly, the function of maximum likelihood must be obtained in order to calculate AIC. But, something I have it with R does not work a bit.
Are there any source codes on the pluses over there? to copy and then figure it out as you go along
I'm exhausted with my own topic, until I found a normal source code with explanations, and it is not completely finished.