Machine learning in trading: theory, models, practice and algo-trading - page 2625

New comment
 
Maxim Dmitrievsky #:
So far nothing comes to mind, in the context of symbolic regression 😀 we have chunks of building blocks like MAs, nadal minima and the like. The question is what's on the output.
The output is the rule if week min..... Next.... Price<week min.....next... ...Price>week min.....next.... .. Price>sma.....next.... .. Price==sma.... Buy



It doesn't matter what you put in there,
Put letters in there, make words, put words in there, write texts, put signs in there, create formulas, put rules in there, create TS.
 
mytarmailS #:
Output rule if week min..... Next.... Price<week min.....next... ... Price>week min.....next.... .. Price>sma.....next.... .. Price==sma.... Buy
So the buys won't always be in profit under this condition, or is this a found rule already?
 
Maxim Dmitrievsky #:
So purchases won't always be in profit under this condition, or it's already a found rule?
So it's already in the fitness function you write what you want from the rule... There already any blue


For example: if this rule worked, then we buy, in fitness we set requirements: minimum 80% of deals must be profitable and this rule must appear on the history at least 100 times... And fitness throws combinations until it does the commander's task
 
mytarmailS #:
So it's already in the fitness function that you write what you want from the rule... Anything blue there
So what are the tags, what are we teaching?
 
Maxim Dmitrievsky #:
So what are the marks? What are we teaching?
We teach: find a rule after which the market goes up and makes money...

A rule is an event, a pattern, a cluster, a TS, whatever you want to call it.
 
mytarmailS #:
We teach : find a rule after which the market goes up and makes a profit...

A rule is an event, a pattern, a cluster, a TS, whatever you want to call it.
I don't see how it's different from NS, apart from the fact that we can't see this rule. It's also a kind of fitting.
 
Maxim Dmitrievsky #:
I don't see how it differs from NS, apart from the fact that you can't see the rule. It's also a fitting of sorts.
Well, teach the network to find a rule like I described. With gaps in time, with logical relationships, and in general, how do you know that the network gave you this rule and not your own, how to calculate statistics for the rule? And if you do not know the recurrence, then how can you call it a pattern? Itd.....
 
mytarmailS #:
Well teach the network to find such a rule as I described. With gaps in time, with logical relationships, and in general, how do you know that the network gave you this rule and not something of your own, how to calculate the statistics for the rule? And if you do not know the repeatability, then how can you call it a pattern? ttd.....
In general it's done by brute force in mt5 optimizer with the same success. And the more complex the condition, the higher the probability of fitting
The main hitch is choosing the right situations for which conditions are being built. You need to make initial assumptions here, or it can take a long time to find
 
Maxim Dmitrievsky #:
In general, this is done by brute forcing through genetics in the mt5 optimiser with the same success. And the more complex the condition, the more likely the fit
Okay... I'm not gonna convince you of anything.
 
mytarmailS #:
OK... I'm not going to convince you of anything.
Just show me if it worked, I'll send you my results on the method later, I didn't have time to finish it. Found the mistakes, didn't fix them.
1...261826192620262126222623262426252626262726282629263026312632...3629
New comment