Machine learning in trading: theory, models, practice and algo-trading - page 3591
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
Imho... fundamental analysis is lagged analysis, in my paradigm it is SMA analysis. Real trades move the market, and fundamental is distant indirect triggers with incomprehensible power and incomprehensible lag. Why alanlise it if there is a price? In general my opinion (which nobody asked) I do not believe in the successful application of fundamentals
Imho, it is better not to contrast them, but to try to combine them in a meaningful and useful way. Fundamentals and the theories around them are also some kind of price information. Imho, it's better to use that information directly rather than hoping the MO algorithm will pull it out of prices on its own.
Also no more than an opinion. Well and also from a psychological point of view it is useful for feeling that you partially understand the logic of the market structure) Anything is better than conspiracy nonsense about rockfellers-reptiloids, your brains will be safer).
Somewhere long ago I saw a signal, it was written that it was based on fundamental analysis, by hand. One trade per month. The growth was something in the region of 10000% for several years, all deals were in plus.
I don't think there are any now. I searched the signals by the word "fundamental". There are about 30 signals in total, more than half of them are minus, those that are plus with a recovery factor of about one. Two or three relatively decent ones.
STUMPY:
https://github.com/TDAmeritrade/stumpy
Motivated by
An ultimatum masst hav f-y, thanks to which the model is never retrained. But don't expect any grail backtests either.
The number of clusters corresponds to the number of patterns into which you want to divide all examples and calculate the average label for each cluster.
The input is a labelled dataset with features, and the output is spit out with corrected labels that prevent overtraining.
Theme development/variations on a theme. It is possible to adjust the number of clusters whose labels will be corrected and the rest will remain unchanged.
This allows to seek a balance between overlearning and underlearning.
Example: if you fix labels for all clusters, the model will not be overtrained, but the balance graph will often be as in this post.
If you fix the labels in different ways, you can get smoother curves that pass OOS.
You can also program (rewrite) for your own language/ask code questions there. You can ask me about the philosophy of the approach :)
I found such a resource with collected fundamental data https://ru.tradingeconomics.com .
It works through its API, but they don't give me the key - I don't know the reason - try it, maybe they will give it to you, then write here.
APIshould berequested here https://tradingeconomics.com/api/pricing.aspx?source=/api/
Found such a resource with collected fundamental data https://ru.tradingeconomics.com .
It works through its API, but they don't give me the key - I don't know the reason - try it, maybe they will give it to you, then write here.
APIshould berequested here https://tradingeconomics.com/api/pricing.aspx?source=/api/
What do you mean Consensus there?
what do you mean ?
I don't know, but probably polling data.
Found such a resource with collected fundamental data https://ru.tradingeconomics.com .
It works through its API, but they don't give me the key - I don't know the reason - try it, maybe they will give it to you, then write here.
APIshould berequested here https://tradingeconomics.com/api/pricing.aspx?source=/api/
Data is good.
I would also like a good theory for meaningful conversion of such data into attributes, which would complement well the technical attributes obtained from the price.