WFO Results and NOT Consistant between tests of SAME Parameters

[rayko]

Hello CodyB, Thanks for the call. I appreciate the help.

[mrengleman]

Was this a bug that was fixed in any of the subsequent beta releases?

if the WFO isn't reliable..how do we know if the regular optimizers are reliable?

[rayko]

It was confirmed that it is a bug in NT7 and was a bug in NT8.

The issue was with Optimization when a strategy is using MTF Multiple Timeframes by deploying AddDataSeries with a secondary time signature different from the primary chart time signature.

So for example, if a strategy adds a 10 min dataseries and is then used with a 5 min primary chart series, the optimizer will not reliably calculate the same results when the strategy is executed multiple times using identical parameters.

This is why the WFO was not reliably producing the same results multiple times using identical parameters.

I was told then that NT8 is more robust and reliable, so I decided to begin moving off NT7 and onto NT8.

I later discovered the same issue with NT8 beta 8 and reported it to support who where able to accurately reproduce the behavior http://ninjatrader.com/support/forum...ad.php?t=82148.

With the release of NT8 beta 10, I can verify that the Optimization problem is now fixed.

The strategy that I use most often is MTF and I have been testing NT8 beta10 Optimization. Its true, NT8 optimization is far more robust and faster than NT7 and I prefer working in NT8.

After having said all this, I would also add that NT8 works best when it is used on a PC with plenty of horsepower and memory running at 64-bit.

My system is Win10 using an AMD FX-8350 8-Core processor and 16GB memory 64-bit.

NT8 beta10 is running really good and fast.

[mrengleman]

Thanks for the info.