Machine learning in trading: theory, models, practice and algo-trading - page 1995

 
Evgeniy Chumakov:


Thank you Maxim!

Actually, I forgot to say, it is enough that the model predicted the next sign (+-) of the series, target + plus value. And the specific accuracy is not important.

Sample length n = 54 , covers all possible values.

The sign is predicted perfectly.

>>> regr.score(X_train, y_train)
1.0
>>> regr.score(X_test, y_test)
0.9938251543711407


 
Maxim Dmitrievsky:

The sign is predicted perfectly.



I wonder if it's the Grail.

 
Evgeniy Chumakov:


Interesting. grail or what?

A good stable stretch)

 
Valeriy Yastremskiy:

))) good steady stretch)


I guess ))

Here is the file with GBPUSD. It is interesting to check the same trained model using data of another pair.

Files:
 

Beware of accidents with perfect results, they allow you to overly believe in the workability of your TS, and there should always be some doubt, and if all options are checked, then you can relax.

To check the efficiency of the method you should perform at least 100 optimizations with different sliding window, and only then draw conclusions.

 
Evgeniy Chumakov:


I wonder if it's the Grail.

Yes
 
Evgeniy Chumakov:


I wonder if it's the Grail.

Maxim Dmitrievsky:
Yes

Guys! Throw the Grail, for the sake of Christ, into the public domain! I would eat the earth just to see Him ...

 
Evgeniy Chumakov:

Here is the file with GBPUSD. It is interesting to check the same trained model on the data of another pair.

>>> regr.score(X_train, y_train)

0.8616703952274422

>>> regr.score(X_test, y_test)

0.8647036898993664


Worse, but still closer to 90%. It was trained on EURUSD.
 
Alexander_K:

Guys! Throw the Grail, for Christ's sake, into the public domain! I'm willing to eat dirt just to see Him...

Alexander, if the Grail is found, the meaning of life will disappear and drag, sex and rock 'n' roll will begin. But is it really necessary? )

 
Maxim Dmitrievsky:

Alexander, if the grail is found, then the meaning of life will disappear and drag, sex, and rock 'n' roll will begin. But is it really necessary? )

I don't know, Max.

But I do know that there are profitable TCs, and people use them. Are they happy? I don't know. We must try it ourselves - it's never too late to quit :)))