programmer with a minimum of garantee - page 9

New comment
 
delyus:

This is not the case, when the trader is looking for a programmer for Ma and Stoch, if there is an ocean of such EAs and two more drops in the free access, moreover, how can he not know that they give about 50% in initial testing with equal Take and Stop, no matter how you bend it, and 70% is required, otherwise everything is useless.

In another threadkomposter once commented on the quality of these ideas: having written over 300 EAs, he only once asked to use one for himself - and it quickly died. Integer, are your statistics better?

That's what I'm saying, that customers' own brilliant ideas cloud their minds and prevent them from thinking critically. A man learns from his mistakes...

But what about this contradiction: a programmer can evaluate the value of an idea in about 2 seconds, then what makes him produce code in 99% of cases (statistics) of unprofitable ideas of naive customers? :))

Of course, money. Why, has such an abomination thrown you into a sweat already?

 

Mathemat

tested over 2000 advisors - all are losing except one on a 7 year history.

 
The best option seems to be the following: the client and the programmer develop a profitable system and share the forex for two))
 
delyus писал (а): I have tested over 2000 EAs - all are losing except one on a 7 year history.
And imagine how many of them will become Grails after optimisation. ...
 

Mathemat

If an EA fails the initial test with equal take and stop, there is nothing to catch further, all the rest biking like cutting losses, extending profit, money management only allows to maximize profit if the system is initially profitable. It's all just to sell it to the client.

 
delyus:

tested over 2,000 advisors - all are losing except one on the Potik history for 7 years.

Offer them to the author of the 'Indicators' thread, he only has about 50 advisors.
 
I offered to give him the first test))
 
delyus:

tested over 2,000 EAs - all are losing except one on the Potik history for 7 years.

This action only made sense and had reliable results for EAs which are complete automatons.

And the vast majority of these 2000 EAs (maybe all of them) are just the means of small-scale mechanization. It is very possible that among them there are some that are built on a more or less sensible idea. However, they are essentially intended for close interaction with a trader: for one you need to be able to prepare data, for another - to be able to interpret the results, etc. It is impossible in principle to single out such potentially useful Expert Advisors by simple testing. So you are very lucky that you have found at least one.

 

Yurixx

They have a full automatic, semi-automatic, their strategy is unprofitable, it's the basis, the essence. 80% of them fail even by checkpoints. there are some that don't even last a month. who made them and why? no one knows.neural networks are a pain in the ass and the more layers, the more pain in the ass. No matter how you slice it, their input is random, hence the results are random, but there is enough freedom with optimization, you can optimize and "train the layers" till the end of the century, but in the best case you end up with flat ass from prolonged sitting.

 
Mathemat писал (а):

 Integer, are your statistics better?

I have this information under X_FILES :-)
12345678910
New comment