Machine learning in trading: theory, models, practice and algo-trading - page 1886

 
Maxim Dmitrievsky:

It's not crap, it's candy.

It's supposed to run like clockwork, you're doing something wrong.

But let's close the public discussion on this one. Because I imagined the cash flowing into my pockets after such an ingenious discovery. I don't want to share it with anyone.

)))) let it be so)

 

Continued at

On the left are errors on the tray and validations from epochs. At the center is the net output on the tray and the benchmark. On the right, the network output on the validations and the benchmark.

Relu, data are normalized to the range 0-1

linear (linear/no activation function), normalized to the range 0-1

linear (linear/no activation function), normalized to +-1

sigmoid, normalized to 0-1 range

sigmoid, normalized to +-1 range


 
Rorschach:
looks good on relu, sweat the real market
 
nate-benton90/Tutorials-1
nate-benton90/Tutorials-1
  • nate-benton90
  • github.com
Code for all my tutorials. Contribute to nate-benton90/Tutorials-1 development by creating an account on GitHub.
 

It's a good MA, but it falls short of the SMA... there is a lot to strive for.

LOL

 
Evgeny Dyuka:
looks good on relu, sweat the real market

I like tanh the best so far.

 
Maxim Dmitrievsky:

It's a good MA, but not quite as good as the SMA... there's a lot to look forward to

lol

This is a reference to this post


In nsdt there was a fitness function to maximize profits, how would one do that in keras? Isn't that training without a teacher?

 
Rorschach:

This is a reference to this post


In nsdt there was a fitness function profit maximization, how would one be done in keras? Isn't that learning without a teacher?

Well do it that way, count instead of the standard

 

Hey machinists, I just want to say, don't you ever get tired of bullshitting? I stopped getting paid for my posts, so I'm allowed to have a good time. Koolie, if the barn burns down, so does the house.... :-) Do we have any interesting ideas or are you going to keep pushing bullshit to young people?????

So..... just curious....


Max your theme last interesting only for BO, when you are armed cooler than the organizers of this, but no more. So maybe we can wipe them out on the sly, so they do not show off. What do you say? I'm just the same theme was poking around and change was one important limitation when working with BO. What do you think?

 
Mihail Marchukajtes:

Hey machinists, I just want to say, don't you ever get tired of bullshitting? I stopped getting paid for my posts, so I'm allowed to have a good time. Koolie, if the barn burns down, so does the house.... :-) Do we have any interesting ideas or are you going to keep pushing bullshit to young people?????

So..... just curious....


Max your theme last interesting only for BO, when you are armed cooler than the organizers of this, but no more. So maybe we can wipe them out on the sly, that they do not show off. What do you say? I'm just the same theme was poking around and change was one important limitation when working with BO. What do you think?

What topic? Nothing works, checked already

You're going to learn python, so go learn it. This is useful. But do not rely on profits from neural networks.