Using neural networks in trading - page 23
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
A single layer perceptron can only compete with the mash-ups in terms of 'power'. But trading at the intersection of mash-ups is nonsense... IHMO.
You don't need power there, you just need the right fit
You don't need strength there, you just need the right fit
How many neurons are there in the layer? Can you at least find out approximately?
Debugger:
там не нужна сила, нужно правильно пристроиться
That's right. If you have a single-layer perceptron there, it is simply because the correct data pre-preparation, which is always the point, makes 70% of all sense there, all subsequent processing, as a rule, only "finishes" the problem. If this is the case, I can assure you that you can do without NS. Something else (depends on the type of problem to be solved) - PCA, regression polynomials of reference vector machine, but NS is not necessary. Just everything else is usually much more robust than NS
I'm not going to argue with you, as Uncle Zadornov says, you're all fine apart...
and I'm sick of everything... everything works.
You just have to get it right, or rather ask yourself the right questions.
By the way, data preparation plays no role in this case. I don't have it at all.
The data comes in raw.
What, are you serving prices in pure form?
By the way, data preparation plays no role in this case. I don't have it at all.
The data is coming in raw.
Excuse me, but I find this hard to believe.
What kind of stuff I was using, I used up to 200 000 neurons in committees of neural networks with total number of inputs on the initial level up to 2000-50 000 thousand. With pre-processing, post-processing. But the market still "twisted" out of the patterns found and went its own way.
a layer of 125 neurons only
Or maybe someone can tell me where else to go with neural networks?