Machine learning in trading: theory, models, practice and algo-trading - page 780

 
Maxim Dmitrievsky:

These pictures are exactly what they're talking about.

about freshmen without a clue, apparently the local residents have not even reached the 1st

I showed exactly what the dull Vizard showed yesterday, and then got scared and deleted

You are welcome. Magic pencils work wonders...

 
Vizard_:

Please. Magic pencils work wonders...

Well, where's the grail?

fxsaber showed me what to do and I almost got it

 
Maxim Dmitrievsky:

Well, where is the grail?

Here fxsaber showed, and I even almost understood what to do

This is a tester...
 
Renat Akhtyamov:
it's a tester...

There's half a forward, it's written.

But I do not discuss other people's ideas, I respect the authorship, only my own.

 
Maxim Dmitrievsky:

Well, where is the grail?

I think Rena had it in the terminal)))

 
Maxim Dmitrievsky:

:D

And gets the optimum strategy on the first 10 games, further significant improvements do not happen

1st game with a random policy, and often it is the only negative, then immediately converges to the optimum... still do not understand why so quickly, probably just the algorithm so picked up

Can I have a picture from June 2017 to March 2018. It is interesting to see how well the model fits.

 
forexman77:

Can I have a picture from June 2017 to March 2018. Interesting to see how much of a fit the model is.

Strongly fitted, 1st varic only, on 3 oscillators with different periods

Put it in training for 4 years for fun, I'll throw it off later with OOS if I wait

 

Dear, advise, a way to classify if the numbers of bars for the inputs are known, but the reasons are not known.

What is the way to identify the patterns. Divide into two classes where to enter and where not?

 
Maxim Dmitrievsky:

Strongly fitted, 1st varic only, on 3 oscillators with different periods

Put in 4 years of training for fun, I'll send later if I can wait

Well, it is still interesting to see in the past will pour or not very much.

 
forexman77:

Well, it's still interesting to see in the past whether it pours heavily or not.

heavily overfitted, pouring on the os

I will train on a long interval and then show the OOS

you also need to change the fiches (environment description).

I just showed that you can learn it in 5-10 runs in a few months with NA, without the target, very fast... and what is it - is another question