Machine learning in trading: theory, models, practice and algo-trading - page 2075
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
Many people have tried it. Me included. Simple MA can be done, and complex bandpass filters can be done.
All that can be built from bars the NS/forest can be easily replicated.
What do you mean by easy?
It's not so easy for me.
I just tried to reconstruct a CCI with period 14, built using only claws
I was getting clowes returns in a squared window of size 20
So returns suck))
But if I normalize the original (not returnees) into a 0-1 edge of each observation, it looks better
Anything that can be built from bars the NS/forest can easily replicate.
Maybe I'm wrong, but logic tells me that trees can't do calculations, which means they can't create a new object from numbers, but a network can transform a number series and create a new feature for learning.
Maybe I am mistaken, but logic suggests that trees cannot do calculations, which means that they cannot create a new object from numbers, but a network can transform a number series and create a new feature for learning.
Yes, that is true, simply put, networks are able to interpolate within themselves (to invent (in both good and bad sense))
and trees are only good at adjusting
The application of this task is not very useful, but as a result of the recovery we can judge how well we process/normalize the data and in general their quality, and it is not bad to know that the grid does not work not because it is dumb, but because we poorly represented the data, or vice versa
So there you go, a simple experiment, but you get more understanding ))
Maybe I'm wrong, but logic suggests that trees cannot do calculations, which means they cannot create a new object out of numbers, but a network can transform a number series and create a new feature for learning.
They can't compute, they're like a database with very fast access - they just remember.
They will remember that for such combination of inputs SSI or any other indicator had such and such value. If there is exactly no such combination, the tree will find the closest combination and as a result there will be a small error on CCI. If the forest, it will average several combinations, probably it will be more accurate than from one tree.
Once saw the expression, "neural network based databases". I think neural networks also only remember.
What do you mean by easy? Can you give me numbers or more details?
I don't have it easy
I just tried to reconstruct CCI with period 14, built using only claws
I was getting clowes returns in a squared window of 20
So returns suck))
But if I normalize the original (not returnees) to a 0-1 edge each observation, it looks better
In numbers or more detail I can't. About 2 years ago I did it on Darch neural network, with bandpass filters from Vladimir Perervenko's articles.
It's like a Zircon missile... the first to figure out how to do it is the winner.
We need to saw different classifiers, maybe someday we'll get lucky
It's like a Zircon missile... the first to figure out how to do it is the winner.
We need to saw different classifiers, maybe someday we'll get lucky.
you should write a random enumeration of everything and learn from it, select what is learned + genetics to shorten the search, and leave the computer for a couple of months....
It's like a Zircon missile... the first one to figure out how to do it is the winner.
we need to saw different classifiers, maybe someday we will be lucky
It seems to me that the developers themselves tried their new products on market quotes. If there was a significant advantage, it is unlikely they would put it in the public domain, but sell it to a bank or fund or make their own fund.
Almost all new advances in MO are grasped by the quants... instantly. It is already a question of where to apply
We are at least five years behind.)