MoGo - new genetic optimiser for NT 6.5

[-Swig-]

Hi folks,

I'm a relative newcomer to automated trading systems and recently discovered Pete S.'s genetic optimiser. There were a few additional features I thought would be handy, so I rewrote it to provide a nicer GUI and some more goodies. It is now available for free for anyone to use.

Main features added:

• Supports all standard & custom optimisation types available to NinjaTrader
• Abitrary user-specified dynamic constraints on strategy parameters
• Progress dialog showing iteration number & true remaining time (NinjaTrader's progress dialog has trouble here)
• Detects with high probability when it has exhaused the valid parameter space
• Saves last-used settings between sessions

Download and usage instructions available here:
http://mcraig.net/site/Development/N...9/Default.aspx

The source code should appear there soon. Comments welcome.


-Swig-

[whitmark]

Ladies and gentlemen, we've got talent! Outstanding GO contribution, in particular, how you are able to successfully overcome the little obstacles of operating under the current NT 6.5 framework with style and added functionality. Congratulations, you've got some serious mojo in your MoGo that would make even Austin proud.

Whitmark

[NinjaTrader_Ray]

Thanks for this contribution!

[Buttermilk]

Preliminary test looks great....

[bibulous]

Thanks! Sounds promising, but unfortunately it does not work for me.

I always get the following error:

Failed to call method 'Optimize' for optimizer 'MoGoNT': String was not recognized as a valid Boolean.

[laparker]

Originally Posted by bibulous

Thanks! Sounds promising, but unfortunately it does not work for me.

I always get the following error:

Failed to call method 'Optimize' for optimizer 'MoGoNT': String was not recognized as a valid Boolean.

Swig

I get the same error - any ideas?

-thanks

[-Swig-]

Originally Posted by whitmark

Ladies and gentlemen, we've got talent! Outstanding GO contribution, in particular, how you are able to successfully overcome the little obstacles of operating under the current NT 6.5 framework with style and added functionality. Congratulations, you've got some serious mojo in your MoGo that would make even Austin proud.

Whitmark

Cheers Mark; appreciate the comments

[-Swig-]

Originally Posted by laparker

Swig

I get the same error - any ideas?

-thanks

Ouch. Do you guys have any non double/int parameters in your strategies? Just occurred to me I haven't tested that scenario.

Are you both on the latest version of NT?


-Swig-

[Buttermilk]

Swig
Look Ok ..no problem in getting it to work but only a fractional increase in analyis speed on NT standard (brute) optimization. Any advice?

[-Swig-]

Originally Posted by Buttermilk

Swig
Look Ok ..no problem in getting it to work but only a fractional increase in analyis speed on NT standard (brute) optimization. Any advice?

How big is your parameter space?

For a small one (i.e. small number of parameters and/or very limited ranges for each parameter), brute-force optimisation is likely to perform equal to or better than any genetic optimiser.

Where GOs shine is when you have a large parameter space (i.e. many parameters and/or a wide range for each), since they hone in on good solutions while testing only a small fraction of possible combinations.

Keep in mind that the amount of time MoGo takes to run is determined by the population size and the number of generations. If you want it to run faster, reduce either/both of these values. Of course lowering them reduces the likelihood of finding the best solution, but there's always a trade off.

-Swig-

[bibulous]

Originally Posted by -Swig-

Ouch. Do you guys have any non double/int parameters in your strategies? Just occurred to me I haven't tested that scenario.

Are you both on the latest version of NT?


-Swig-

Yes, I do have boolean parameters, but those are not included in the optimization universe. I also run the latest version of NT.

Marc

[laparker]

Originally Posted by -Swig-

Ouch. Do you guys have any non double/int parameters in your strategies? Just occurred to me I haven't tested that scenario.

Are you both on the latest version of NT?


-Swig-

Swig

Yes, I have the latest NT 6.5.1000.9
I have bool params but I don't change them in the optimization process.

-thanks

[Elliott Wave]

Originally Posted by laparker

Swig

Yes, I have the latest NT 6.5.1000.9
I have bool params but I don't change them in the optimization process.

-thanks

6.5.1000.11 is the latest.

[laparker]

Originally Posted by Elliott Wave

6.5.1000.11 is the latest.

Thanks for the heads-up. There doesn't seem to be much in the bug fixes that might be relevant to this issue, but I will upgrade.

-thanks

[bibulous]

Originally Posted by laparker

Thanks for the heads-up. There doesn't seem to be much in the bug fixes that might be relevant to this issue, but I will upgrade.

-thanks

They are not - I am running .11.