Machine learning in trading: theory, models, practice and algo-trading - page 1168
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
checked the MLP from the alglib on the multiplication table for normalization, hmmm... no normalization needed, strange... very strange that it's that simple!
the alglib grid is quite good, of course there are no miracles, but even on forward it shows values with little to no logic.... very strange that it's so simple! )))
HH: #define k 1 you can twist, but have not noticed any changes, as for the input to supply from 0...1 that from 1...100, whatever works pretty well all the same
well, and the forest approximately the same will count, I threw the multiplication table for it too, only on large datasets will be the difference - the neural network is trained slower
Interested in your opinion on the use of the kagi chart for MO. How often is it used and how meaningful is it in your opinion?
Interested in your opinion on the use of the kagi chart for MOs. How often is it used and how meaningful is it in your opinion?
I think it is a common indicator of price change by a certain value, just like renko... in what situations it can be used: for example, to filter out "noise", but it can also be done through a muving, for example, with exactly the same lag. So if there is a need to get filtered data with lag - then maybe there is sense, but in practice most likely there is no such a task to get the strongly lagging market information.
I think it is a common indicator of price change by a certain value, just like Renko... in what situations it can be used: for example, to filter "noise", but it can also be done through muwings, for example, with exactly the same lag. So if there is a need to get lagged filtered data - then mb is the point, but in practice most likely there is no such a task as getting strongly lagged market information.
It's probably not about practice. Such graphs create some connection between the market and the "coin exercise".
Piecewise steady-state SB doesn't simulate a flat very well. An obvious development of the model is the piecewise homogeneous Markov chain. It is quite possible to use MO to train it. Surely someone did something similar, but I have not found anything.
Interested in your opinion on the use of the kagi chart for MOs. How often is it used and how meaningful is it in your opinion?
hmm, i thought i'd post a forest trained in renko charts in this thread today,
i read somewhere that some category of people have similar ideas)))
hmm, thought I'd post a forest of trained Renko charts in this thread today,
i read somewhere that a certain category of people think alike ))))
i think, if you add to your instuments possibility to optimize rengo settings, then we'll see
In general, you can use my example of bandit for Weierstrass plot, changing logit to forestTo switch to python without any fuss and to solve examples, I recommend google colaboratory
you can take conditional probabilities (continuity traces) as "changing a coin", as far as I understand it, which, in the limit, will converge to 50\50, but there may be variations in the moment. This is exactly what the Bayesian approach is about, as opposed to the frequency approach (the difference between the a priori and posterior distributions). This is sort of my humble view of what is going on in this book at the moment, since I am not very familiar with Bayesian statistics, although there is essentially one equation and all :)
But there is a suspicion in possibility of application of it for search of conditional regularities, if it can be called so. And a neural network is just for generalization
at least this approach I try to add to my RL bandits
correct me if I'm wrong.
I have Sagemath, which is based on python.
About 50/50 in the limit I agree, though there are different ways to formalize it. I want to find some simple model to distinguish our graphs from the symmetric SB at short intervals. Bandits seem complicated enough to me, as do hidden Markov models, to lead to overlearning.
Ignore the opposition of the frequency approach to the Bayesian approach-they coexist quite well in a theorist)
The problem with finding regularities, in my opinion, is all the same non-stationarity.
you can get 100% profit per month.
Zehr gut, Max! It's the Grail and it's hiding in the increments. It's just not so easy to see - but it's there, 100%.