Machine learning in trading: theory, models, practice and algo-trading - page 1443

 
Igor Makanu:

I don't remember all higher mathematics, although I always had my deserved 5 ((((, but I'm saying that no matter how you "twist" any TF, even ticks, even alphabets have arrived, but statistically there will always be such a picture:

All this forum is full of such graphs and it doesn't matter what and how authors of their methods searched, and any use of statistics for TS will have a gain for the market ... Well somewhere a bit more than 50/50 - depends on the time of testing, a year, two...10 years - all profitable TS come out in profit only due to successfully selected capital management and some fortunate coincidence ))))

Taleb has an interesting discussion on a similar theme. He divides people's areas of activity into two types - those with normally distributed income and those with fat-tailed income. What is interesting is how he describes this, not from a scientific point of view, but from a simple human point of view.

 
Aleksey Nikolayev:

Taleb has an interesting discussion on a similar topic. He divides people's fields of activity into two types--those with normally distributed income and those with fat-tailed income. What's interesting is the way he describes this, not from a scientific point of view, but from a simple human point of view.

Taleb just got lucky a couple of times and now everyone reads his whiny memoirs about how scary everything is, no one takes him seriously on Wallstreet, well no more seriously than our clown Maxim Denisenko and his "dopamine" staggering guys here.

 
Kesha Rutov:

Taleb just got lucky a couple of times and now everyone reads his whiny memoirs about how scary everything is, nobody takes him seriously on Wallstreet, well no more seriously than our clown Maxim Denisenko and his "dopamine".

Kesha, how do you trade, do you have a signal or monitoring?

I don't have anything on my mind, it just doesn't feel right.

I agree, of course, that Maxim, to put it mildly, is "no angel" in communication, but still, the question was asked.
 

Yes, "get it first" is a double-edged weapon that requires precision.)

 

Information for thought - I make a new herbarium, selected 93 sheets to buy on the data futures Si, decided to check whether these patterns work on other tickets - conducted tests for the period 2014-2019, it turned out 19 sheets, which according to the tests bring income for this period on additional tickets.



The question arises, if 20% of the sheets turned out to be resistant to a complete change of the tool, then it is necessary to leave them only, or we can talk about randomness here?

 
Renat Akhtyamov:

Kesha, your trade is going well, do you have a signal or monitoring?

He just does not have anything behind his back, it is not humanly possible.

I agree, of course, that Maxim, to put it mildly, "not an angel" in communication, but still, the question arose.

I'm not selling anything and I'm not condemning Maxim's way of talking but his attempt at a scam, for which he should be ashamed.

 
First you have to think of a problem, and then actively solve it.
 
Aleksey Vyazmikin:

Information for thought - I make a new herbarium, selected 93 sheets to buy on the data futures Si, decided to check whether these patterns work on other tickets - conducted tests for the period 2014-2019, it turned out 19 sheets, which, according to the tests bring income for this period on additional tickets.



The question arises, if 20% of the sheets were resistant to a complete change of the tool, should we keep only them, or we can talk about randomness here?

It's the same as usual with half of the indicator code - "Well, let's see what I have, what do you think?

everybody's like, uh... okay, a table with numbers, the kid's got a good idea

 
Maxim Dmitrievsky:

As usual, as with half of the code of the indicator - "here I lay out, look what I can, what do you think?

everybody's like, uh... okay, a table with numbers, the kid's got a good point.

It's not the code that's important here, what's important is that the selected rules in the sheets work partially on other instruments without any modification.

This is to the question of the possibility of sifting out false hypotheses of regularities - i.e. to the possibility of controlling the results of learning.

Unfortunately, I don't have models that work on many instruments, so it is logical that there are many overtrained leaves in a complex.

 
Aleksey Vyazmikin:

It is not the code that is important here, what is important is that the selected rules in the sheets partially work on other instruments without any modification.

This is to the question of the possibility of sifting out false hypotheses about regularities - i.e., to the possibility of controlling the results of training.

Unfortunately, I do not have models that work on many instruments, it is logical that there are many overtrained leaves in the complex.

Well, look how the tools correlate, why there is a coincidence

what can we say about these numbers?