Machine learning in trading: theory, models, practice and algo-trading - page 1394

New comment
 
Maxim Dmitrievsky:

When it comes to results - I haven't seen anything similar to mine in the thread, not even close

The only results I've seen are from fxsaber, and not from the MO in the full sense of the word.

i don't need to remind you about the backtests on the napkin.

I don't take it as a criticism, I'm just saying that it's a very complex approach and I'm amused by statements like "I'll do it for a couple of weeks and everything will be OK.

You have megalomania. Only your approaches and tests are correct, the rest is wimpy and bullshit, etc.
I think my tests are no less correct, but plus, even more informative. And I advise you to switch to a napkin, and give up the religious belief in the divine origin and infallibility of the MT tester).
 
Maxim Dmitrievsky:

Even about such a seemingly simple thing, no one wrote here, as well as in general about RL, alglib forests, etc., until I brought up the topic

so what are we talking about... So you see only that "Random Targeting" and you can't think of anything more complicated to attach to it, because it's always easy to see what's ready and say it's easy, but to improve...

Before your articles I thought it was something very complicated. But now I can make something up and modify it if necessary. And thank you for that!
 
Yuriy Asaulenko:
You have megalomania. Only your approaches and tests are correct, the rest is on the napkin and bullshit, etc.
I consider my tests to be no less correct, but plus, even more informative. And I advise you to switch to a napkin, and part with the religious belief about the divine origin and infallibility of the MT tester).

I don't see any tests from you, the dispersion spot on the chart is a complete random))

 
elibrarius:
Before your articles, I thought it was something complicated. But now, if necessary, you can think of something and modify it. And thank you for that!

this is the basis, how to complete the model without a teacher, then there is a lot of wisdom on how to make something work with the new data

e.g., how to properly sample these examples, from which distributions, at what frequency, how to validate, etc.
 
Maxim Dmitrievsky:

Asaulenko served 20 returnees into the net and is happy... isn't that funny?

I need results, not complex approaches. Complex approaches must produce a new quality. Yes, you have complicated methods. But there are no qualitatively different results. So why all these difficulties?
Of course, you can think of 2x2 as a sophisticated way of stating how cool we are, and how cool our idea was. It makes an impression on some people,
That's it, I'm gone.

 
Maxim Dmitrievsky:

I don't see any tests from you, then the scatter spot on the graph is a complete random ))

random is round)

And there is an ellipse.

 
Yuriy Asaulenko:
I need results, not complex approaches. Complex approaches should produce a new quality. You do have complex modd elsewhere. But there are no qualitatively different results. So why all these complications?
That's it, gone to Auchan.

)) The quality is that I get a handful of good models in 5 minutes (the result). Complexity is needed initially so that then do not sit for months and not pick up models when the old broke down

 
elibrarius:

randome is circular)

And there's an ellipse.

Okay, an ellipse is on the edge of randomness, closer to a circle than a straight line.

 
Maxim Dmitrievsky:

OK, the ellipse is on the edge of randoms, closer to a circle than to a straight line.

As Yuri suggested if I trade with predictions > 2.5 and <-2.5 - in reality most trades will be profitable:

You see that amount of mistakes is 15-20%

 
elibrarius:

As Yuri suggested, if we trade with predictions > 2.5 and <-2.5 - then in reality most trades will be in profit:

You can see that the error rate is 15-20%.

it doesn't work like that, there is a 45g slope line through the cloud and not horizontal

from don't count the deviations... even suggested not to know what ))
1...138713881389139013911392139313941395139613971398139914001401...3516
New comment
Reason: