Using neural networks in trading - page 21
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
Leonid, I am my own, bourgeois.
A neural network does not in any way mimic the workings of the brain. It only allows you to simulate its vegetative ("vegetable") processes.
A neural net? It's elementary, Watson.
And it is required to solve the simplest and the most pleasant task - to buy cheap, and sell expensive.
Neuronet is a toy for children, grown-ups do not deal with such bullshit.
The good thing about the neural network is that it is, if I may say so, a "universal regressor". Well, almost universal. For example, I use them in some, so to speak, "intermediate" cases, when it is necessary to understand whether there is a dependence, what kind of it and all that, and there is no time and effort to do it, it is necessary to move on. But to trade with it - that's me against it.
Sometimes it seems to me (more and more often) that working with neural networks reminds me of experiments of an obsessed alchemist. A lot of cones, books, reagents... Something is mixed, tested, looked at, and finally a philosopher's stone should come out, but no... It is the same with neural networks, only in a modern way. Several monitors with a bunch of windows, tables, graphs, diagrams. About a dozen of programs are hanging, counting something. The system is designed in such a way that the MSEs are ready, graphs are shaped in the right way, diagrams promise rivers of money. Tests, checks, fine tuning begins, and... bam. Another bummer. You just sit there for a while, then smoke and think about where to add a neuron, what grid to redo, change the connections, and what is the error in general. You start to change something, rework it. Again a whole kitchen with data preparation. Fresh samples, optimization, rationing, tests, checks... Again running training, waiting... And so on in a circle. To be honest, you don't follow the market, you don't have time and energy. All in the hope that your "kitchen machine" equipped with neural networks will see everything and "understand" the market. This is what it is created for! After a few months you get tired. You just binge on some game. You give time for the inflamed brain to cool down, but at the same time somewhere in the recesses of consciousness a new "brilliant" plan is born. And everything starts all over again. Hope dies last....
You seem to be a stranger to the "scalpel of ocamma" :) And in vain.
I agree, I got into a lot of trouble.
But it doesn't work on a simple one.
Hope dies last....
the miraculous perseptron is only good for curing chicken pox
I agree, I gave up on the perseptron a long time ago, it's weak. Regression is simpler with the same result.
The Perseptron as an auxiliary piece in the system may still be useful, but it is easier to spend time "extracting" the "knowledge" from it (if it has any), and to regress it mathematically. There will not be any bother with training.
Sometimes it seems to me (more and more often) that working with neural networks reminds me of experiments of an obsessed alchemist. A lot of cones, books, reagents... Something is mixed, tested, looked at, and finally a philosopher's stone should come out, but no... It is the same with neural networks, only in a modern way. Several monitors with a bunch of windows, tables, graphs, diagrams. About a dozen of programs are hanging, counting something. The system is designed in such a way that the MSEs are ready, graphs are shaped in the right way, diagrams promise rivers of money. Tests, checks, fine tuning begins, and... bam. Another bummer. You just sit there for a while, then smoke and think about where to add a neuron, what grid to redo, change the connections, and what is the error in general. You start to change something, rework it. Again a whole kitchen with data preparation. Fresh samples, optimization, rationing, tests, checks... Again running training, waiting... And so on in a circle. To be honest, you don't follow the market, you don't have time and energy. All in the hope that your "kitchen machine" equipped with neural networks will see everything and "understand" the market. This is what it is created for! After a few months you get tired. You just binge on some game. You give time for the inflamed brain to cool down, but at the same time somewhere in the recesses of consciousness a new "brilliant" plan is born. And everything starts all over again. Hope dies last....