Machine learning in trading: theory, models, practice and algo-trading - page 1699

New comment
 
Tag Konow:
I don't understand what's the problem with creating an NS that recognizes price patterns? A human can do it without any training. And they teach and teach ... Where is the NS that recognizes basic chart patterns? Why isn't there one in QB? There are a lot of algorithms, but there is no such network...
You forget that human beings are subject to emotions, which prevents them from having an advantage over algorithms. And as for patterns... When drawing up a training sample it is necessary to save the sample from volatile data, when the market reaction to the same pattern is diametrically opposed. And in this case neither the algorithm nor a human will be able to solve this problem infallibly.
 
mytarmailS:

no! it is not enough.

A neural network or other AMO is just a "multidimensional optimization" and that's it!

It is a tool to solve the problem, that's all!

And the problem has to be set!

And the problem has to be invented!

And the task must be selected from other tasks!

And the task is relevant!

etc... While it's all up to the man, while... These kind of creativity is called something like

I see. The network cannot define its task and select data for self-learning. It cannot identify the situation in which its "experience" is relevant and apply it as it sees fit... So it is far from being an AI.
 
Konow tag:
I see. The network can't identify its task and select data for self-learning. It can't identify the situation in which its "experience" is relevant and apply it as it sees fit... So it is far from being an AI.

And what is AI ? formulate a definition for starters

 
Mihail Marchukajtes:
You forget that human beings are subject to emotions, which do not allow their advantage over algorithms. And about the patterns... When drawing a training sample it is necessary to rid the sample of non-volatile data, when the market reaction to one and the same pattern is diametrically opposite. And in this case neither the algorithm, nor a human being will be able to solve this problem infallibly.
Yes, humans perceive patterns differently, but there are invariant forms, like "head-shoulders". Well, at least make a conditional, approximate recognition. "Pseudo-subjective" identification. At least something, for all that time...
 
ReTag Konow:
I see. The network cannot identify its task and select data for self-learning. It can't identify the situation in which its "experience" is relevant and apply it as it sees fit... So it is far from being an AI.
OOO here you are right. NS in this case is like a child and if she is not thoroughly explained what to do, she will be of no use, and do not talk about AI in the literal sense of the word. We live in an era of generalizable algorithms, which are one-celled compared to AI. But its mathematical advantage over humans makes it an indispensable tool, capable of seeing patterns in the data stream that a human would not be able to see in a lifetime by constantly observing the data.
 
mytarmailS:

And what is AI ? formulate a definition for starters

A working model of human intelligence possessing, as it does, versatility in terms of problem solving. An "absolute solver."
 
Reg Konow:
Yes, people perceive patterns differently, but there are invariant forms, like head-shoulders. But, make at least a conditional, approximate recognition. "Pseudo-subjective" identification. At least something, for all that time...
What is important is not the pattern itself, but the subsequent reaction of the market, which can be contradictory and then nothing will help. Only intuition is a person's ability to feel with his/her oops, but nobody has been able to describe it mathematically yet...
 
Mihail Marchukajtes:
You forget that a man is subject to emotions, which prevents him from having an advantage over algorithms. And about the patterns ... When drawing a training sample it is necessary to exclude the trivial data, when the market reaction to one and the same pattern is diametrically opposed. And in this case neither the algorithm nor a human will be able to solve this problem infallibly.

Right! Because the task was set by the man wrong, but he does not understand it, it's all dumb network, not me, no not me)))

 
mytarmailS:

Right! Because the task was set by the man wrong, but he does not understand it, it's all network stupid, not me, no not me)))

Not that it is not set correctly, and elementary negligence. Forgot to clean the inconsistency, got a bad training results and as a consequence of the work on the CB
 
mytarmailS:

Right! Because the task was set by the man wrong, but he does not understand it, it's all network stupid, not me, no not me)))

Guys, if several people are looking at the same chart and see the same pattern, why can't the network see it?
1...169216931694169516961697169816991700170117021703170417051706...3699
New comment