1) Is there a guideline of minimum # of tests/run in the Monte Carlo optimizer, for a strategy with N (say N=5) parameters? Also, does this depend on the range of the parameters configured (i.e. the Start and Stop values) and the Step value configured for the parameters?

2) Do the Default and Step values in the parameters have any influence on the results of Monte-Carlo optimizer? i.e. does specifying a default value that's closer to the expected optimal value "assist" the optimizer find and generate better results?
- Also, what's the guideline for specifying the Step value to "assist" the optimizer?

Re: Step value. Quoting the User Guide > Strategy Window > Optimization > Full Optimization > Optimization Control:

The Monte Carlo process chooses random combinations of values that fall between the Start and Stop values defined for the parameters; the Step value is meaningless.

Thanks - pl also advise reg Question 1 -

1) Is there a guideline of minimum # of tests/run in the Monte Carlo optimizer, for a strategy with N (say N=5) parameters? Also, does this depend on the range of the parameters configured (i.e. the Start and Stop values) and the Step value configured for the parameters?

Also, the Default value has no meaning in the Monte-Carlo tests, correct? Or does it assist it with a better starting point that would lead to better results?

thanks
Kiran

QUOTE:
Also, the Default value has no meaning in the Monte-Carlo tests, correct? Or does it assist it with a better starting point that would lead to better results?

Correct, the Default value is not utilized by the algorithm and therefore has no meaning.

QUOTE:
Also, does this depend on the range of the parameters configured (i.e. the Start and Stop values)

Yes. The Start and the Stop values are used internally.

QUOTE:
1) Is there a guideline of minimum # of tests/run in the Monte Carlo optimizer, for a strategy with N (say N=5) parameters?

From what I can recall there is no written guideline besides the User Guide chapter where you could find this:

CODE:
Please log in to see this code.

The idea of MC optimization is to quickly find optimal ranges for a large number of parameter combinations and then zero in on them with Exhaustive optimizer.

Eugene,

It appears from what you are saying that the MC Optimizer is performing a version of Simulated Annealing. Is that correct?

Vince

While I've heard of SA before, I don't have any more experience with it than what I've read in Wikipedia. With that in mind, the MC optimizer is more basic (MC random vs. SA probabilistic), and I can't really explain how it works any better than what it already says in the User Guide where it describes Runs and Tests per Run. However, MC probably cannot guarantee putting you at a global maximum like SA can. It's likely to do so, but it might not.

Thanks Cone!