Machine learning in trading: theory, models, practice and algo-trading - page 366
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'll be honest and frank - I made my NS diagnosis for myself a couple of years ago and abandoned this method.
So how exactly for NS - it is difficult for me to say. Maybe there is something in the NS, which allows you to shovel into the network everything that is at hand without pre-selection.
For all DM methods I have outlined the approach.
Then the second option is to stick everything in the NS. But there are two BUTs:
1. Hope that uncorrelated variables do not degrade the quality of the model (there is such a thing for regression).
2. sacrifice dimensionality and time.
I'll mull over the new information entered... )
Because correlation of inputs and outputs doesn't matter at all when the model is looking for patterns in a set of predictors... It's a contradiction - remove correlated inputs but look for correlated inputs to outputs... )) I.e. at least we will have one input correlated with the output, then all other inputs we will have to remove, since they too are correlated with the output, and consequently with the other inputs... cool, right?
I would not remove correlated with the output (and correlated to each other), unless there will be more than 5 - 10.
The problem is different - what indicators (at least at 0.6) are correlated with the zigzag, for example? I haven't found any yet (
I would not remove correlated with the output (and correlated with each other), unless there are more than 5 - 10 of them.
The problem is different - what indicators (at least at 0.6) are correlated with the zigzag, for example? I have not found any yet (
I don't know them either, and I wouldn't worry about it at all... and I wouldn't feed the zigzag output
I don't know them either, and in general I wouldn't bother about it... And I wouldn't use a zigzag output either.
And what to serve as a teacher on the way out? Do I have to do the markup manually myself? By the way, are there any scripts to automate manual markup?
Not necessarily markup, increments of the same indicators or prices... You can prepare the data for training in many different ways.
I'll start experimenting with it soon, while I'm preparing the material what exactly I want from the NS
I don't know these either,
By the way, you have a good balance graph (a few pages ago), so you still use this one, (perhaps you just do not know that it is correlated with the target)
I don't have a target there at all, it's without a teacher
I have no target there at all, it has no teacher
Do you use TrendLinearReg.mq5 for input?
Two of them with different periods, and one psi.
But I want an ensemble of self-training NSs or one but very cool one... constantly using an optimizer and hoping for a miracle is not an option
If complex - you can place it on Google or azure servers, train it there, then just send requests to a server and take the result... and that's it, no stress on your computer or cpu... the idea is a millionfold. I.e. we train a normal neural network in a normal cloud for that purpose, and the terminal is just used for trading and getting results.