Machine learning in trading: theory, models, practice and algo-trading - page 379
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
https://habrahabr.ru/post/243211/
Here's a good post about retraining of NS, I reproduced the results, my ARIMA predictions matched, but NS gave me something wrong :)
The cited link is just a killer for NS, because ARIMA is a model with very limited use in financial markets.
The article didn't test ARCH, so it's impossible to say how this model will behave in the future.
We should do GARCH. And then to refine the initial data these models should be applied together with the machine learning.
We have to do GARCH. And then apply these models together with machine learning to refine the initial data.
Yes, the train of my mind seems to be moving towards understanding this
The cited link is simply killer for NS, because ARIMA is a model with very limited application to financial markets.
The article did not test for ARCH, so it is absolutely impossible to say how this model will behave in the future.
We should do GARCH. And then to refine the initial data these models should be applied together with the machine learning.
Have you used azure machine learning studio?
The trained model is stored in the cloud and the results can be retrieved via web service, bypassing all sorts of dlls. The initial version is free. It supports R.
Have you used azure machine learning studio yet? I think it's pretty cool.
The trained model is stored in the cloud and the results can be retrieved via web service, bypassing all sorts of dlls. The initial version is free. R support.
No, I haven't.
I am extremely skeptical of innovations. I always try to answer one question: if I follow the progress, will my profits increase at least in my dreams.
And the cloud... where is it? So the Internet is off, but the computer is still on and you can work.
No, I haven't.
I am extremely skeptical of innovations. I always try to answer one question: if I follow the progress, will my profits increase at least in my dreams.
And the cloud... where is it? So the Internet is down, but the computer is left and you can work.
You can't work without the Internet :) I've almost completely switched to the cloud... the system crashed or the computer died - I restored it and no problems... I even went somewhere without my laptop, went to someone else's place and took what I needed from the cloud. And your own neural network in the cloud is a real buzz, you can sell connections to it, people will be connected by bots and trade. Plus the learning curve there is very fast.
Yes, the train of my consciousness seems to be moving towards understanding it
My thought goes in a circle: I switched from TA to ARIMA. I did not get any practical results. Financial rows that ARIMA can cope with are extremely rare. Then I moved on to machine learning. Here I managed to get practical results, but I'm not satisfied. Now I returned to GARCH and was surprised to find that there are a great number of publications on application of GARCH models to the financial markets, including Forex. This is against the background of practical absence of similar publications for MO.
Why would that be?
My thought goes in a circle: I left TA for ARIMA. I did not get any practical results. Financial rows that ARIMA can handle are extremely rare. Then I moved on to machine learning. Here I managed to get practical results, but I'm not satisfied. Now I returned to GARCH and was surprised to find that there are a great number of publications on application of GARCH models to the financial markets, including Forex. This is against the background of practical absence of similar publications for MO.
Why would that be?
Well because GARCH is a ready-made meaningful model and MO is just MO. I bought a higher school textbook, there is garch in it.)
My thought goes in a circle: I went from TA to ARIMA. I didn't get any practical results. Financial rows that ARIMA can handle are extremely rare. Then I switched to machine learning. Here I managed to get practical results, but I'm not satisfied. Now I returned to GARCH and was surprised to find that there are a great number of publications on application of GARCH models to the financial markets, including Forex. This is against the background of practical absence of similar publications for MO.
Why would that be?
Here's something I found.
http://www.quantalgos.ru/?p=2037
Here's something I found.
http://www.quantalgos.ru/?p=2037
Google garch, GARCH, rugarch, ARCH test, APARCH and others. The use of garch in the financial markets and other variations. Huge lists.
Pinned a list of arches. Any of the list you score and you get up to your eyebrows
PS
Currently trying to master the APARCH beveled-student model from the rugarch package
Google garch, GARCH, rugarch, ARCH test, APARCH and others. Applications of garch to financial markets and other variations. Huge lists.
Pinned a list of arches. Any of the list you score and you get up to your eyebrows
PS
Currently trying to master the APARCH model with the beveled stander from the rugarch package
and the forecast will be on "Hooray!"
By the way, they say that "YES" it's good as long as volatility is low.
give an example of a hedge fund that was based on the application of GARCH, but went down the drain as soon as the market took a tumble...