Strategy optimizer, always another result.

[Frederic]

Hello all,

I try to optimize my strategy with Genetic(original part of ninjatrader 7).
this is how i optimize my strategy:
-choose dates, lets say: 01.01.2012 - 25.05.2012
- lets take a strategy, in this example named: TOTO
-lets give parameters for TOTo: initial stop loss, break even point, autotrail.
and now lets run it. we have a result: 241trades, PNL +1050usd and so on...

now I press again Optimize and lets wait for the result. Do not forget we do not change any parameters, or date, all stay the same....
we have result:
238 trades, PNL: -5625usd and so on....

how is it possible that everytimes, strategy optimizer give me different results for same strategy, same period of time??

i connect using CQG demo account, and i can say that right now, i am blocked!

Please help to understand this extraordinary phenomenom.

Thanks!

Regards,
Frederic

[NinjaTrader_Jason]

Hello Frederic,

I am not sure why you experience different results. Are you able to clarify the difference if you analyze the trades that happened?

Please double check if the exact same settings are used.

Are you able to reproduce the issue when testing the SampleMACrossOver strategy?

[koganam]

Hello all,

I try to optimize my strategy with Genetic(original part of ninjatrader 7).
this is how i optimize my strategy:
-choose dates, lets say: 01.01.2012 - 25.05.2012
- lets take a strategy, in this example named: TOTO
-lets give parameters for TOTo: initial stop loss, break even point, autotrail.
and now lets run it. we have a result: 241trades, PNL +1050usd and so on...

now I press again Optimize and lets wait for the result. Do not forget we do not change any parameters, or date, all stay the same....
we have result:
238 trades, PNL: -5625usd and so on....

how is it possible that everytimes, strategy optimizer give me different results for same strategy, same period of time??

i connect using CQG demo account, and i can say that right now, i am blocked!

Please help to understand this extraordinary phenomenom.

Thanks!

Regards,
Frederic

A genetic optimizer gets its speed of testing precisely because it is not exhaustive. A genetic optimizer does not test every possibility: instead it uses principles of genetic inheritance to try to test the most likely outcomes. Thus it is often prone to finding local optimum points and even sometimes getting stuck optimizing around them. As it uses pseudo-random seeds to find its initial iteration points, it may find different optimal values in different runs.