Machine learning in trading: theory, models, practice and algo-trading - page 960
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
Well I didn't know that all this has been in use for a long time while I was inventing...
they've been using everything discussed here for half a century now.
now only more advanced models like diplerning are being added
observation - there is not a single idea that I came up with and then did not find almost exactly the same one on the Internet. (For example, I recently posted an example about fuzzy logic and NS. First I thought up and then I found exactly the same idea 1 in 1, though the article is rather fresh there) And there is no MO model which hasn't already been tried in the market before you :) Mostly English-language resources, of course ... in runet in general complete cluelessness
have been using everything discussed here for more than half a century.
now only more advanced models like diplerning are being added
observation - there is not a single idea that I came up with and then did not find almost exactly the same one on the Internet. (For example, I recently posted an example about fuzzy logic and NS. First I thought up and then I found exactly the same idea 1 in 1, though the article is rather fresh there) And there is no MO model which hasn't already been tried in the market before you :) Mainly English-language resources, of course, in runet in general complete cluelessness
Boring :)
Boring :)
I dunno what to do next, no one throws up ideas so far, too lazy to think
There is a model, trained steadily well in different modifications, some at 100% and more from trains work out on the AOS, as here ... (4 months of training 10 months of OOS) then shit
I don't see the point in testing the demos because everything is already clear.
I don't know when the system will break down in the future :D I made some kind of semigraals and now I sit and stare at it blankly, 50k have already been offered
I`ve got to read 500-page books in English again...
I do not know what to do next, no one throws up ideas so far, too lazy to think
There is a model, trained steadily well in different modifications, some at 100% and more from trains work out on OOS, as here... (4 months of training 10 months of OOS) then shit
I don't see the point of testing the demos because everything is already clear.
I don't know when the system will break down in the future :D I made some kind of semigraals and now I sit and stare at it blankly, 50k have already been offered
I've already been offered 50k to read 500+ pages of English books again...
Maybe we'll start swapping fetches?
"garbage in - garbage out" is a correct and at the same time important for understanding the thesis, worthy of fundamental study. Of course, it does not exhaust all the possibilities of modeling and does not take into account the infinity of ways to select the input data for the study. Everyone knows that the choice of data is determined by the characteristics of the object under study or the nature of its mathematical model, if it is known. On the other hand, any data should be considered at a certain level of their abstraction relative to the set of "absolute" factors determining the market behavior. Not having these benchmarks, we can only make a comparative estimate, which will be purely local. Personally, from my experience, I am convinced that a thoughtful approach to the choice of initial data increases the performance of numerical modeling.
Can we start switching chips?
I just have entry prices, I don't suffer from chips :) the main thing is the selection of targets
"garbage in - garbage out" is a correct and at the same time important for understanding the thesis, worthy of fundamental study. Of course, it does not exhaust all the possibilities of modeling and does not take into account the infinity of ways to select the input data for the study. Everyone knows that the choice of data is determined by the characteristics of the object under study or the nature of its mathematical model, if it is known. On the other hand, any data should be considered at a certain level of abstraction relative to the set of "absolute" factors determining the market behavior. Not having these benchmarks, we can only make a comparative estimate, which will be purely local. Personally, from my experience, I have found that a thoughtful approach to the choice of input data increases the performance of numerical modeling.
I think that terver+MO, there are no other options. It turns out as if scientifically and tastefully
I don't have much experience with terver, I need to check it out.
I just have input prices, I don't suffer from chips :) the main thing is the selection of targets
Then all this should work only while historical prices are repeating...
I think terver+MO, there are no other options. It turns out both scientifically and tastefully
I don't know about terver, I'll have to study it.
Exactly so.
A certain Asaulenko does just that. Although he is trying to wiggle like a hare, he is still a physicist and I have confidence in his model.
And it is as follows - it looks if the price has left the level of confidence probability, and the NS additionally gives the permission/decline to enter the trade. I have the same thing, only instead of NS I use Pearson's asymmetry coefficient. I want to use Pearson's asymmetry coefficient.