Machine learning in trading: theory, models, practice and algo-trading - page 1495
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
I checked the indicator fractional.mq5. The results were not significantly better than those obtained with returnees. Unrealistically good results using the RVI indicator series[i] = iRVI(NULL, 0, 14, MODE_MAIN, i+1) - iRVI(NULL, 0, 14, MODE_MAIN, i+2). The graph of equity growth is shown in the screenshot below. These results were obtained in the observation area, but what is surprising is that learning takes place without a teacher. Now it remains to check its performance in the tester. I have looked at it and concluded that ldhmm is a more thorough HMM package with many useful auxiliary functions for financial timeseries analysis.
I checked the indicator fractional.mq5. The results are not significantly better than those obtained with returnees. Unbelievably good results on the training sample using the RVI indicator series[i] = iRVI(NULL, 0, 14, MODE_MAIN, i+1) - iRVI(NULL, 0, 14, MODE_MAIN, i+2). These results were obtained in the observational domain, but what is surprising is that learning happens without a teacher. Now it remains to check its work in the tester. I have looked at it and concluded that ldhmm is more thorough package by HMM with a lot of useful auxiliary functions for financial timeseries analysis.
nice, i wish i had some more example with os, otherwise it could be a wild guess.
what fractional parameters did you use? did you try to reduce it to 0.1, for example? and treshhold 1e-05 is optimal?
in fact, they differ very little :) the first is more sensitive
Nice, I wish I had some more example with oos, otherwise it could be a wild fit.
and what parameters fractional did you use? tried to reduce it to 0.1, for example? and treshhold 1e-05 is optimal
Yes, tried changing degree and treshhold. The signal frequency at all settings was not high. Now I will see what it will show in ldhmm.
I'm posting an mq4-indicator for the ldhmm package.
And no one has tried anything on my data(( but how much empty stuff has been written(
It would be great if you also posted the code on the P
And no one has tried anything with my data((( but so much empty stuff is written(
It would be great if you also posted the code for P
i don't have time yet, i've downloaded some literature on circuits, but i haven't read it yet
Look closely at the source code
+ What's your dataset, some kind of numbers? Why sit the model to incomprehensible numbers to fit?
It would be great if you also posted the code on the P
There is an indicator with the code attached. The code is really not combed and archaic, but it works.
+ what kind of dataset you have there, some numbers.
There is a calculation of a certain event, which explains (and predicts) almost all the reversals in the market, if you compare dataset with the price it is clearly visible, but no one has done this, but found a bunch of retarded who began to write all sorts of nonsense like - "yes a moving average is better",
"it's all a peek into the future." yeah....
Why sit the model to incomprehensible numbers to fit?
And who asked to adjust something? All I asked is to do the same thing I just in a different package, for example from python. It takes four minutes and four lines of code.
But you started to read something, learn the theory, write about it and communicate.
As a result, you've written tons of pages and already forgot what you were asked to do), and you were asked to spend4 minutes on 4 lines of code
It's a shame...
There's an indicator with the code attached. The code is really not combed and archaic, but it works.
There is a calculation of some event, which explains (and predicts) almost all the reversals in the market, if you compare dataset with the price it is clearly visible, but no one has done this, but there was a bunch of retarded who started writing all sorts of nonsense like - "yes the moving average is better",
"it's all a peek into the future." Yeah....
And who asked to adjust something? All I asked was to do the same thing I only in a different package, for example with python. It takes four minutes and four lines of code.
But you started to read something, learn the theory, write about it and communicate.
As a result, you've written tons of pages and already forgot what you were asked to do), and you were asked to spend4 minutes on 4 lines of code
which is a shame...
I do not know μl, I tried to understand but there is a lot mixed up, it would be better if you put just the code and P, there are only a few lines of code and everything would be clear, what is where and where.because even that Python package doesn't work like that, you have to enter transition probability matrices, mean matrices and other stuff. Where do you get them from? I don't understand how it works on the fly, so I have to read the whole algorithm
and you really have two lines of code there\.
and anyway, just so you understand, state space models are divided into markov process and markov decision process. I've worked with the second one, and the first one is your case. And there are subspecies of algorithms there too.
Asaiulenka is on fire all the time, no matter what you say.
and you really have two lines of code there.
Well roughly speaking yes, here is the article where the basis of everythinghttp://gekkoquant.com/2014/09/07/hidden-markov-models-examples-in-r-part-3-of-4/ was taken from.
And here is a code snippet from the example in the article.
If you take away the data processing and visualization, yes, the code is three lines.