Genetic Optimizer 2.0 for NinjaTrader 6.5

[jdfagan]

The Ulcer Index is on the list as well as the ability to specify your own fitness function. As far as using other algorithms, that is a large undertaking and as Dierk points out, best to wait until NT 7 Beta is released before evaluating.

Thanks for the response.

By the way, I was talking about UPI (Ulcer Performance Index) vs UI (Ulcer Index). They are slightly different meanings - UPI makes use of UI:

http://en.wikipedia.org/wiki/Ulcer_Index (search for UPI within this page - aka, Martin Ratio; a better Sharpe Ratio basically IMHO)

http://www.fasttrack.net/newhelp/ulcer.asp

Cheers,

JD

[whitmark]

Genetic Optimizer 2.1 for NinjaTrader 6.5

A new version of the wmGO is available at this thread:

http://www.ninjatrader-support2.com/vb/showthread.php?t=17802

[Anagoge]

NT7 will come with a GA which is based on Pete's work here: http://www.ninjatrader-support2.com/...ead.php?t=5932

This will be a great new feature. One suggestion I have is to eliminate the duplicate parameter sets from being displayed in the final optimizer results, when the parameters all match. For example, if you are optimizing a Fast/Slow SMA crossover, there is no reason to display more than one row of Fast=10 Slow=20 parameters in the results, as Pete's optimizer does. Also, it would be great if the parameter set duplicate "organisms" were not even re-run through the entire set of historical bars/trades, since the performance results would be identical to a previous run in the set of duplicates (the WMGO tool apparently includes this optimization). One of the best features of other optimizers I've used is automatic stagnation detection, where if the GO makes no progress in finding better solutions in a certain number of iterations, it automatically stops. This way, you can enter a large number of iterations, just in case, but it only uses as many as it needs. Another old request of mine is being able to add any number of custom performance metrics that could also be used as a custom optimizer fitness function (I'd like to have a monthly Sortino ratio with a configurable minimum acceptable return). A progress UI display of the current best parameter set and the optimizer max/min performance metric value for those params would be awesome and maybe even a graph of the min/max value over time as the optimizer runs. One can dream, right?

[NinjaTrader_Ben]

Hello,

I will pass this suggestion on to our development department for consideration. Thank you!

[piersh]

hey guys, I've made a bunch of changes to Pete's GAOptimizer, please see this thread for more info:
http://forum.bigmiketrading.com/showthread.php?p=512

[Elliott Wave]

hey guys, I've made a bunch of changes to Pete's GAOptimizer, please see this thread for more info:
http://forum.bigmiketrading.com/showthread.php?p=512

Some more information.

it uses a different way of extracting the tested strategy's score. it uses a hack in base Strategy.Dispose to get around a design flaw in NT - they don't expose the value being optimized to the optimizer . this allows you to use whichever scoring metric you like without modification. i removed the logging/threshold stuff since this it uses the same scoring as the analyzer. you can add this in your custom OptimizationTypes, if necessary.

• it uses an integral representation of the parameter space that allows it to ensure that all children are unique - reducing the number of backtest runs that are required.
• it uses a fitness weighting function to more naturally choose the parents of the next generation. the chance that an individual gets to procreate is on a probability curve, not a boolean function. this increases the richness of the gene pool by allowing worse-performing individuals to procreate (albeit with lower probablility than others)
• it makes an attempt to determine when a maximum has been found and the parameter space is stable (ie. subsequent generations are unlikely to produce any better individuals). it does this by checking the top 'stability size' % of the population - if they're the same between two generations, then it considers the parameter space to be stable. at this point it blows away (almost) the whole history and starts over with the top 'reset size' % of the previous lot, seeds the rest of the new generation with aliens and continues. if you find it's resetting too much, try increasing the 'stability size' percentage (a value of 100 means it'll never reset).
• between resets, it creates new generations from all previous generations. this is necessary since new children are now unique. i might add an aging function to reduce the probability of those crusty great-great-great-grandparents getting some action...
• it adds aliens. aliens are completely randomly generated (none of their genes are derived in any way from other individuals). a certain configurable percentage of every new generation are aliens, the rest are normal (possibly mutated) children of the previous generations. I added this because I found that while it slows convergence somewhat, it allows a broader search of the parameter space, helping to avoid local maximums.

[whitmark]

Whitmark Genetic Optimizer 2.2 for NT 6.5

GO aficionados,

There is a new version of the Whitmark Genetic Optimizer 2.2 referenced in this thread:

http://www.ninjatrader-support2.com/vb/showthread.php?t=18132

Whitmark

[piersh]

GO aficionados,

There is a new version of the Whitmark Genetic Optimizer 2.2 referenced in this thread:

http://www.ninjatrader-support2.com/vb/showthread.php?t=18132

Sorry, no aliens, just mutants

Whitmark

Phew, derision. Hopefully that means my code's not worth plagarising...

[whitmark]

Optimizing the Optimizer

Piers,

Derision? . . . no. In jest? . . . yes! The subjects of biologically inspired algorithms involving parents and children, bee swarms, ant colonies, and now aliens, makes me smile. Clearly, the introduction of random "aliens" is one of among several innovative approaches you are introducing.

We would all be delighted to have a developer of your caliber working to deliver an advanced GO. My efforts have been a stop gap since Pete S abandoned support of his initial work over a year ago (for reasons unknown) and in lieu of a production NT solution or viable alternative. Obviously, you've got the skill and ideas to go well beyond a canonical GA solution. Hopefully NT7 will make this easier to develop custom GO solutions going forward without all of the workarounds.

Thanks for the heads-up on your work. Good luck and please keep us posted.

Whitmark