Strategic foresight systems - page 3

New comment
 

The script pumps up the history, but only if the appropriate history files have already been created, i.e. charts with the appropriate TF have been opened at least once and the terminal has been reloaded, I think that you, Sergei, will definitely have the script working correctly

All that's left now is to make the forecasts! :D

Files:
farnsworth.mq4  2 kb
 

to Mathemat

That's very interesting. Do you have any references?

It's just that I've been messing around with independence checks myself lately, i.e. not that far from Bayesian. And so far, too, on a single pair.

There's a course and examples on the web-site(http://www.bayesia.com/en/products/bayesialab/tutorial/chapter-1.php) and the first 100 Google hits on the query "Bayesian networks" :o). There are also books:

  • "Bayesian networks. the logical probabilistic approach", Tulupiev, Kolenko, Sirotkin.
  • "Bayesian Methods in Econometrics" by A. Zellner
  • "Introduction to Bayesian Networks."
  • "Introduction to Bayesian statistical inference methods."
  • "Elements of statistical decision theory (Bayesian and minimax solutionsa)"
  • "Control science. A Bayesian approach."
  • "Consistent control over incomplete data: a Bayesian approach"

needed?

 
IgorM:

The script pumps up the history, but only if the appropriate history files have already been created, i.e. charts with the appropriate TF have been opened at least once and the terminal has been reloaded, I think that you, Sergei, will definitely have the script working correctly

All that's left now is to make the forecasts! :D


Thank you very much, I will start exploiting the script from tomorrow.
 
Mathemat:


It's just that I've been messing around with independence checks myself lately, i.e. not that far from Bayesian. And so far, too, on a single pair.

How are you getting on, Alexey?
 

I'm struggling with the chi-square script. I'll get it done for sure.

There is almost finished dumb version (brute force), but it's too resource-intensive, takes a long time to count. I don't like it. As I'm not afraid of hardware upgrade in the near future, I will rely on software optimization...

The main problem is different: I have doubts about the system itself, for which, in fact, all began to be done.

But in any case, the script should prove useful in the future.

 

just in case you need a library:

EURJPY:

Recalculated with more detail in the options space (takes about 1.5 - 2 hours for one quote), seems to be the moment of truth. Theoretically it should go down. Will be clear by tomorrow evening. But it looks like this is the only quote for Monday that could change its local trend.

 
Mathemat:

The main problem is something else: I have my doubts about the system for which it was started.

God, I hope it's not because of me))))
 
Farnsworth:

  • Anything I recommend here, I will also be trading on an alpari demo account. I will post the password if I have to. I don't think it will be interesting and there is unlikely to be an explosive increase in the deposit, and for now that is not the most important thing. Quality control will now be more about other characteristics.
It would be interesting to see.
 
Mathemat:

I'm struggling with the chi-square script. I'll get it done for sure.

There is almost finished dumb version (brute force), but it's too resource-intensive, takes a long time to count. I don't like it. As I'm not afraid of hardware upgrade in the near future, I will rely on software optimization...

The main problem is different: I have doubts about the system itself, for which, in fact, all began to be done.

But in any case, the script should prove useful in the future.

What are your doubts, if not a secret? Regarding resource intensity, I solved a similar problem by introducing a genetic algorithm into the calculation (less data is counted than before), although the problem is not radically solved, but it is still sensitive. And you have to be careful with it. Maybe GA can help in your situation too?
 
Alexei, four more posts and you can celebrate))
12345678910...50
New comment
Reason: