Walk forward discrepancy

[ldissinger]

I ran a walk forward test. Everything looks good until I click on each walk forward test.

All but two of the walk forward test report identical results. But the two that do not are way off.

Back testing with the parameters selected gives a picture very similar to the second image, although not exact.

Leroy

P.S. First one is correct, the second image shows the discrepancy.

[NinjaTrader_CalH]

Hi Idissinger,

Thank you for your post.

I will need to see the strategy and test this on my end to see if I can reproduce the results.
You may send me your strategy to support[at]ninjatrader[dot]com with ATTN Cal in the subject and a reference to this thread in the body.

Thank you for your patience.

[ldissinger]

That will be difficult and I will try to perform the same test with some other strategy to see if I can duplicate. The strategy uses many indicators and a custom bar type.


Leroy

[ldissinger]

You stragegy....your bars

It did not even work here, the first result is not accurate. You can see that the first result net profit does not equal the net trade profit in the lower screen.

I ran it again, and it is now correct, so don't just run it one time. It appears to me there is a bug in there somewhere.

The I extended the time to start at 4/7/2013 and multiple of them were not correct.

I can't afford to do multiple passes of a walk forward optimization that takes 6 hours to run if the results are off.

Leroy

[NinjaTrader_CalH]

Hi Idissinger,

Thank you for the update.

Yes, we are aware of this issue. A use case has been submitted for the next major release of NinjaTrader.

Let me know if I can be of further assistance.

[ldissinger]

Then why in the world did you ask me for more information if you already knew there was an issue? I have better things to do .

So basically, walk forward testing is useless until the next major release. Please don't flaunt it if you can do it.

You should put out a note to all developers of systems that it doesn't work so don't rely on it or even use it.

I'm not being nasty, just realistic. I can't even imaging how many hours I've put into walk forward testing to find it was all a waste of time.

Leroy