Machine learning in trading: theory, models, practice and algo-trading - page 669
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
I am finally convinced that trying to predict specific events with NS. That is, learning with a teacher gets me nowhere. I think to look at learning without a teacher. Self-organizing maps come to mind. They are able to break down the data into classes without assigning output classes. Once you get the classes, you can look at the stories to see what you can get out of them. Generally, this is a new area for me. In general, I have come to a trivial conclusion. TC construction begins with searching for patterns. The method I have suggested is one approach. You can simply study the history visually. This is how it has been done for a long time. That's why the heads and shoulders and so on appeared. But I'm still a programmer, and I want to find algorithms to find patterns. That is, to reduce everything to their software search. Attempts to find materials on this subject have yielded no results. It does not matter if it is neural networks or not. The essence of trading is profit and what methods are used for its obtaining is not important. But again, everything starts with the search for patterns.
Read Sutton, Barto - learning with reinforcement. Change your perspective a little bit.
I am finally convinced that trying to predict specific events with the help of NS. That is, learning with a teacher does not lead to anything. I think to look at learning without a teacher. Self-organizing maps come to mind. They are able to break down the data into classes without assigning output classes. Once you get the classes, you can look at the stories to see what you can get out of them. Generally, this is a new area for me. In general, I have come to a trivial conclusion. TC construction begins with searching for patterns. The method I have suggested is one approach. You can simply study the history visually. This is how it has been done for a long time. That's why the heads and shoulders and so on appeared. But I'm still a programmer, and I want to find algorithms to find patterns. That is, to reduce everything to their software search. Attempts to find materials on this subject have yielded no results. It does not matter if it is neural networks or not. The essence of trading is profit and what methods are used for its obtaining is not important. But again, everything starts with the search for patterns.
It took me a long time to select signals, activation functions, network training method, etc. to achieve similar results. Now the quality of the prediction is remarkable. (Dark is the signal, light is the forecast)
And also look at the quality of the teacher. It depends a lot on him.
It took me a long time to choose signals, activation functions, network training method, etc. to get similar results. Now the quality of the prediction is wonderful. (Dark is signal, light is forecast)
What is the backtest and forward of the bot?
And also look at the quality of the teacher. A lot depends on it.
It took me a long time to choose signals, activation functions, network training method, etc. to get similar results. Now the quality of the prediction is wonderful. (Dark is the signal, light is the forecast.)
Does anyone know why the indicator values in the left corner may blink when recalculating? annoying :) animated gif
Does anyone know why the indicator values in the left corner may blink when recalculating? annoying :) the animated gif
Show me the code.
Does anyone know why the indicator values in the left corner may blink when recalculating?
I had this problem when I took time series bars not one after the other but one after another.
In the 5-pc of course, in the 4-pc - normal.
)
I had this happen when I took the time series bars to calculate not one after the other, but one by one
)
How do you mean, one by one?
in one how?