Machine learning in trading: theory, models, practice and algo-trading - page 1043
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 agree!!!! But that's not all... It turns out that the second graph is better because there are vectors that are as close to the zero axes as possible. In this example it is not so obvious, but now we have such datasets where the component vectors coincide with the zero axes and divide the field into even 4 squares. In the first case, the component axes are scattered along the dioganals between the zeros, while in the second picture there are such component vectors that are as close to the zero lines as possible. Knowing the name of the predictor we train the optimizer as long as the inputs are those predictors that form the component vector closest to the zero axis and it doesn't matter in what direction. Again, this is my IMHO!!! That's why I wanted to clarify how right I am!!!!
Your optimizer is non-linear and PCA is linear, I don't know how you understand it
Your optimizer is nonlinear and PCA is linear, I don't know how you figure it out.
Very simple. From the PCA graph I find out which predictors lie closest to the zero axes. And I keep on running the model as long as it has exactly these predictors on its input. As a rule they drop out after 2-3 optimizations, such models will be robust to data and will maximally saw the current data area. IMHO of course :-)
Very simple. From the PCA graph I find out which predictors lie closest to the zero axes. And I keep on running the model as long as it has exactly these predictors on its input. As a rule they drop out after 2-3 optimizations, such models will be robust to data and will maximally saw the current data area. IMHO of course :-)
it's not so clear-cut there
It's not all that clear-cut.
It's a pity I never got a chance to record a video. It's September. On account, the plus was symbolic, and remained so :-( the main reason for the absence of the promised video. But all the same, I would like to say a few words. In other words, it is not logical, what works is not logical. That is, what is not logical works. It seems to me, what is the sense and purpose in it? There is no sense in such an input, and it shows the highest results. Allogicality comes within a millimeter of an error. That is, how to do things the way no one else is doing and at the same time not to make mistakes, both gross and hidden. That is what makes it interesting. I wanted to say a lot in that video, but the last three months of probation have kind of slipped my mind what I wanted to say.
If I'm asked to give trading advice, it's surprisingly blunt:"Get a good job.
You have no idea how cool it is to work at a job that you like and get NOT bad money there for a specialist of your level. Then it's easier to trade, too, that le....... You don't need it, reducing risk is the key to success!!!!
It's a pity I never got a chance to record a video. It's September. On account, the plus was symbolic, and remained so :-( the main reason for the absence of the promised video. But all the same, I would like to say a few words. In other words, it is not logical, what works is not logical. That is, what is not logical works. It seems to me, how this makes sense and what is the point? There is no sense in such an input, and it shows the highest results. Allogicality comes within a millimeter of an error. That is, how to do as no one else does and yet do not make mistakes, both gross and hidden. That is what makes it interesting. I wanted to say a lot in that video, but the last three months of probation have kind of slipped my mind what I wanted to say.
If I'm asked to give trading advice, it's surprisingly blunt:"Get a good job.
You have no idea how cool it is to work at a job that you like and get NOT bad money there for a specialist of your level. Then it's easier to trade, too, that le....... You don't have to trade, and reducing risk is the key to success!!!!
i don't know what it is. trading may not be your main occupation when you trade with other people's money, it's clear.
But no credit for the non-video :(Maestro, hello!
And the Teacher is here.
Well, at least now the branch will be alive.
Although... What am I talking about? Ah! Money, what else can I talk about?
So - without a stationary process at hand, it makes no sense to use a neural network.
This topic is interesting solely and only theoretically.
To make money with a neural network without an awesome transformation of BP to a stationary form is NOT possible, even if you die.
Is this transformation possible? Yes! But, I personally have not managed to find it - it is somewhere far away, at M5 and above. Not on ticks, definitely not.
Maestro, hello!
And the Teacher is here.
Well, at least now the branch will be alive.
Although... What am I talking about? Ah! Money, what else can I talk about?
So, without a stationary process at hand, it makes no sense to use a neural network.
This topic is interesting solely and only theoretically.
To make money with a neural network without an awesome transformation of BP to a stationary form is NOT possible, even if you die.
Is this transformation possible? Yes! But, I personally have not managed to find it - it is somewhere far away, at M5 and above. Not on ticks, for sure.
Why are you obsessing over this stationarity?
There is NO. NO. You won't find it on any TF.
But on the other hand, why the hell are you looking for it? Because your approach doesn't work without it? So your approach is not good for the problem. Change your approach. But in such a way that it is in no way tied to "stationarity", so that there is not even a mention of it.
By the way, neural networks are free from the limitation of "stationarity". This is their strong point. They have a different thing.
It's a pity I never got a chance to record a video. It's September. On account, the plus was symbolic, and remained so :-( the main reason for the absence of the promised video. But all the same, I would like to say a few words. In other words, it is not logical, what works is not logical. That is, what is not logical works. It seems to me, what is the sense and purpose in it? There is no sense in such an input, and it shows the highest results. Allogicality comes within a millimeter of an error. That is, how to do things the way no one else is doing and at the same time not to make mistakes, both gross and hidden. That is what makes it interesting. I wanted to say a lot in that video, but the last three months of probation have kind of slipped my mind what I wanted to say.
If I'm asked to give trading advice, it's surprisingly blunt:"Get a good job.
You have no idea how cool it is to work at a job that you like and get NOT bad money there for a specialist of your level. Then it's easier to trade, too, that le....... You don't have to trade anymore, and reducing risk is the key to success!!!!
Michael, you shouldn't mope around. You're the only one practicing in this thread!
Even if tired of fruitless research, PCA in R is indicative of lame, because neural network autoencoders are problematic there or pull to physicists masturbating to Fokker-Planck-Kolmogorov:)
So - without a stationary process at hand, it makes no sense to use a neural network.
This topic is interesting solely and only through theoretical research.
To make money with a neural network without an awesome transformation of BP to a stationary form is NOT possible, even if it kills you.
Is this transformation possible? Yes! But, I personally have not managed to find it - it is somewhere far away, at M5 and above. Not on ticks, definitely not.
Not pointless. Pseudostationarity can be found by brutforcing models in different dimensions, iteratively finding the best options on the automaton. This is partly what Roffild does, but somehow wrong or just not explained. I'm stumped on conversions myself - complicated work with multidimensional arrays is multistep, if I do handmade without any P and pythons.
It is not meaningless. Pseudostationarity can be found by bruteforcing models in different dimensions, iteratively finding the best options on the automaton. This is partly what Roffild does, but somehow wrong or just not explained. I'm stumped on conversions myself - complicated multidimensional arrays are multistep, if I do handmade without any P and pythons.
Explain in Russian what these words mean.