Machine learning in trading: theory, models, practice and algo-trading - page 2541
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
But most manual traders don't have any stable theories though?
And in general it's a bit like the process of searching for a theory to describe the whole world around us.
After all, a time series can be anything and describe any process, just like the market can be almost any situation.
It would be much harder to fully describe and theorize everything than to just extract some kind of profit with some kind of stability.
Theory and practice are two legs that must be moved in turn and equally vigorously to get to the desired goal.
Practice is the criterion of truth, which theory tries to find.
Practice is the criterion of the truth that theory tries to find.
There is no arguing with this. I just wanted to say that the pain of failing to apply a particular theory to a particular practice should stimulate the search for a new particular theory (rather than generating a useless stream of complaints and scolding).
But most manual traders don't have any stable theories though?
And in general it's a bit like the process of searching for a theory to describe the whole world around us.
After all, a time series can be anything and describe any process, just like the market can be almost any situation.
It would be much harder to fully describe and theorize everything than to just extract some kind of profit with some kind of stability.
And you're smarter than your predecessor...
You're right, you need a theory, followed by a market model...
There are two kinds of participants on the market + a certain global trend + three kinds of levels
If we imagine the market from the viewpoint of waves, then the market participants are a fast wave, the trend is a slow wave, the levels are sudden disruptions of the wave trajectory...
The market is a sum of all these events (if we consider the model as additive) or interference...
And you are smarter than your predecessor...
You're right, you need a theory, followed by a market model...
There are two kinds of market participants + a certain global tendency + levels of three kinds
If we imagine the market from the viewpoint of waves, then the market participants are a fast wave, the trend is a slow wave, the levels are sharp breaks of the wave trajectory...
The market is a sum of all these events (if we consider the model as additive), or an interference ...
Or it is possible to imagine fast waves as panics, and slow waves as trends.
One can only wonder why there are so few successful algorithmic traders with such a simple theory and the simplest classical strategies for it, like trend following.
Maybe visualizing data as waves in non-stationary time series makes them delusional, and locks them in an endlessly apophenic hell from which they search for a way out for years?
Actually I meant that for a neural network (at least a human brain), in general, theory is not necessary, because many people feel that certain situations are good or not, but they can't explain why. But this is not a good method for algorithmic trading (in my opinion), I am not sure that computer neural network can compete with human.
Actually I meant that for a neural network (at least a human brain), in general, theory is not necessary, because many people feel that certain situations are good or not, but they can't explain why. But it is not a good method for algorithmic trading (in my opinion), I am not sure that computer neural network can compete with human.
Bravo! If the system is "complete" and closed and shows "robust" results, it is a very good method ;-) Yes, it can ;-)
I'll keep the conversation going:
What is the result of learning? What is a neural network? In simple terms this is a formula of a polynomial with unique weight coefficients found during training. Hence, training is a search for a mathematical law effective in the market at the moment. The unique law known ONLY to you and this is why it works in general, because if the current law becomes available to ALL market participants, it immediately loses its relevance.
Since we do not know the law displayed in the data, we are training to find our own, unique law formula, which will be known only to us. So it goes like this...
I'll keep the conversation going:
What is the result of learning? What is a neural network? In simple terms this is a formula of a polynomial with unique weight coefficients found during training. Hence, training is a search for a mathematical law effective in the market at the moment. The unique law known ONLY to you and this is why it works in general, because if the current law becomes available to ALL market participants, it immediately loses its relevance.
Since we do not know the law displayed in the data, we are training to find our own, unique law formula, which will be known only to us. So it's like this...
Well... With an artificial neural network and an infinite number of features we can generate as many laws as we want, but it's useless. I thought the point was to exploit existing patterns, the seasonality of different periods. But the funny thing is that it is impossible to isolate them, because there are hundreds or thousands of people like me, and they use the same or more sophisticated tools)))
Well... With an artificial neural network and an infinite number of features, you can generate as many laws as you want, but it's useless. I thought the point was to exploit existing patterns, the seasonality of different periods. But the funny thing is that it is impossible to isolate them, because there are hundreds or thousands of people like me, and they use the same or more sophisticated tools))).
What is the basis for the opinion that these patterns exist and even exist?