Machine learning in trading: theory, models, practice and algo-trading - page 3055

New comment
 
Maxim Dmitrievsky #:

At the matstat level I guess. If on average several models are wrong in predicting the same thing on new data (on validation subsample), then it is unpredictable at all and is moved to "do not trade"

You can take a specific time and corresponding values of signs/signals as the "same thing".

Like BestInterval?

 
fxsaber #:

Like BestInterval?

Looks like
 
Maxim Dmitrievsky #:

3 buy/sell/no trade classes

I just slightly modified the approach from the last article to make it less confusing to understand.

You can close on reverse signals or stop and take.

I'm just assuming a large drawdown on that chart. That's why I'm wondering how exactly closing is done there.

 
Aleksey Vyazmikin #:

I'm just assuming a large drawdown on that chart. That's why I'm wondering how exactly the closing is going on there.

Do you have anything in essence? There are already a lot of comments with different advice and questions off-topic :) I am not interested in answering them.
For example: coded, checked - it works/doesn't work. I propose such and such improvements, checked - it became even cooler/all is rubbish.
 
Maxim Dmitrievsky #:
Anything on the subject? There are already a lot of comments with different advice and questions off-topic :) I'm not interested in answering them.
.
For example: coded, checked - it works/doesn't work. I suggest such and such improvements, checked - it became even cooler/all is rubbish.

If there are 50 trades a year with reversals on reverse signals, the position can hang around for weeks, that's why I'm asking about it. This affects the result of the method's success rate.
Use Seed Overkill - create many models and combine them in order to get more signals. What other direction to think in...

Didn't work out with the leaves?

 
Aleksey Vyazmikin #:

If there are 50 trades a year with reversals on reverse signals, the position can hang around for weeks, that's why I'm asking about it. This affects the result of the method's success rate.
Use Seed overshoot - create many models and combine them in order to get more signals. What other direction to think in...

Didn't work out with the leaves?

It is with leaves and even with buds.
You can't do that with those leaves.
Well, forget it, if there's nothing else to think about.
 
mytarmailS #:
It looks cool, but get another OOS but bigger.
Still, did you double-check or not?
 
mytarmailS #:
Still, double-checked or not?
I don't have time to run all your errands. They are meaningless in the context of the topic at hand.
 
Maxim Dmitrievsky #:
I don't have time to run all your errands. They are meaningless in the context of the topic raised.
It's one line of code.

And you should have done it yourself, just for yourself, so you don't have to engage in self-deception....
So it's strange that you still don't know about it, or do you??? :) but we don't need to know about it.)

 
mytarmailS #:
It's one line of code.

And you should have done it yourself, just for yourself, so that you don't engage in self-deception....
So it's strange that you still don't know, or do you???? :) but we do not need to know about it:)

Anything on the subject? It's funny to read.
No - well, thank God
1...304830493050305130523053305430553056305730583059306030613062...3526
New comment
Reason: