AI Generate (Experimental) Does not follow Exit condition: Session Close

[raffu]

That's why only a single trade and 99 profit factor... from start date to end date, if uptrend or downtrend during trade test (Not optimize fitness max strength type problem I think)

[raffu]

See yellow highlight and trade

[NinjaTrader_ChrisL]

Hello raffu, thanks for your post.

If you take one of the generated strategies and run it on a chart you should see the exit on session close feature working on the historical backtest. Is this what you are seeing as well? I do not see the screenshot you posted, please re-upload this so I can have a look. The attachment button is at the very top right of the forum text editor.

I look forward to hearing from you.

[raffu]

Chris,

So the Performance shown (PF 99 and 1 trade) AND ALL results are NOT correct which is shown, see the previous and you have to run backtest with the strategy created to get real result.
(actually it was ok PF 2..25 and 41 trades after I added bool True = true; to strategy so it could be compiled)

How to get list of real performance then? (as optimizer has to calculate that somewhere although I have bad feelings that it use shown results (not the real one, exit at session close)

[raffu]

I tried to add it again to get you a point

[raffu]

Let's formulate that again,
How to get ONLY day trading strategies (exit at session close) from AI Generate as I think that box (Session Close) just mean that it add that possibility to the generator.

That would be one way to force to more trades (and not to pick that 1 trade PF 99 max strength strategies).

May I give a suggestion (RFE): Use that lower "Exit on Session Close" to that at some future version.

[NinjaTrader_ChrisL]

Hi raffu, thanks for your reply.

I'm not getting the same results in the same test. What happens when you set the Exit conditions to only the Exit On Session Close option?

[raffu]

Chris,

FYI, I had generations set 10 so it can be tested fast. And you might have to run several times as diff random seed, one more suggestion, possibility to set up random seed to get same results at every run (although that might cause some confusion to someone).

Anyway back to the real question and some tests.
Yes. if you ONLY choose "Exit on session close" (from Order handling) (No "Session close"), works as I expected, -> close at end of session, great!
But, if you choose "Session close" (the Exit condition) it is same as "Optimize" that (ignoring "Exit on session close" setup which might be confusing, it was for me) Actually that is totally Okay as I know that now.


So no problem at all, but this last post might be useful to someone.

Conclusion: Set Only "Exit on session close", NOT "Session close" to day trading strategies to get reasonable results.

[EB Worx]

I get the same issue.. I tried it on multiple futures, dates, timeframes..
But for me all strategies it creates only take 1-2 trades , have a profit factor of 99 and a performance of 99. (see attached screenshot)
Perhaps i'm doing something wrong ?

[EB Worx]

Here's another one.. different date and different timeframe.. Still just 2-3 trades and profit factor/performance of 99
I think ninjatrader could better filter out these fake results and show strategies which have a good balance of # of trades for the time period analyzed.
As its now.. its not very usefull, but again.. perhaps i'm doing something wrong?