Interesting and Humour - page 4035

[Дмитрий]

[khorosh]

Дмитрий:

Neural networks are a great cheat in general and just a pretty toy in particular.

If any other regression or classification method can be applied, even at the expense of prediction accuracy, they should be used.

And NS is either for fun or if nothing else.

It remains to understand how Alexander Topchilo (Better) won the ATC championship in 2007 using his nice toy - neural network Expert Advisor. And later he earned good money at different brokerage companies. I wonder what helped him to fool and cheat everybody).

[Дмитрий]

khorosh:
It remains to be seen how Aleksandr Topchilo (Better) won the ATS Championship in 2007 with the help of his beautiful toy, a neural network advisor. Later he earned good money at different brokerage companies. I wonder what helped him to fool and cheat everybody).

Alexander Topchilo became a millionaire thanks to this advisor?

[khorosh]

Дмитрий:

Did Alexander Topchilo become a millionaire thanks to this advisor?

Is answering a question with a question your signature style?

[Дмитрий]

khorosh:
Is answering a question with a question your signature style?

Judging by your answer - OUR.

[khorosh]

Дмитрий:

Judging by your answer - OUR answer.

And if he didn't become a millionaire, but just became a wealthy man, then what, we can conclude that neural networks are the great hoax? Where is your logic?)

[Дмитрий]

khorosh:
And if he didn't become a millionaire, but just a wealthy man, then what, we can conclude that neural networks are the great fraud. Where is your logic).

Never mind, never mind - I wrote faa and he understood what he was talking about.

[khorosh]

Дмитрий:

Never mind, never mind - I wrote to faa and he understood what he was talking about.

Well, if you can't explain why your opinion about neural networks does not coincide with the facts I gave, then maybe faa can explain this paradox, since he knows what he's talking about).

[Vasiliy Sokolov]

khorosh:
Well, if you can't explain why your opinion about neural networks doesn't fit with the facts I gave you, then maybe Faa can explain this paradox, since he knows what I mean).

For a neural network to show results, it needs a good teacher. But if there is a good teacher, there is no need for a neural network - you already know how to trade.

[khorosh]

Vasiliy Sokolov:

For a neural network to show results, it needs a good teacher. But if you have a good teacher, you do not need a neural network - you already know how to trade.

So a neuronet is an unnecessary intermediate tool? But here is what Better says:

You cannot just say that neuronet is more effective than other methods. It is not a panacea. This is just one of approaches to trading systems development. I tried many approaches and once tried all popular Expert Advisors based on standard indicators, and only neuronet based EAs produced acceptable results.

So, what, he just didn't find the right approach ?