[120] [Tradestation engine] Fuzzy strategies problem is still there

This is the historical data which I use: https://drive.google.com/open?id=1pwt9VsY_ko1YgXlvp5f9GYRrx0wYPaKP

If you need any information, please let me know.

I made some tests today and noticed that Multicharts uses a different way of calculating higher timeframe bars than Tradestation.

The only way how to do a backtest in SQ that will match the results from MC is that you export the data from MC, import it into SQ and use SelectedTF precision in Data settings and NoSession in Trading Options settings. That way I was able to get the same results.

There were also some functions missing in MC - for example ADX. These have been added into SQ_MC.pla file.

We are thinking about adding a new engine type for MultiCharts in later releases

You said MultiChart uses higher timeframes, what would be? Can you clarify ?

Do you mean H1 timeframe? I always use M60 instead of H1 in Multicharts, I think you can also change the TF to be M60 instead of H1 in SQX

You said "We are thinking about adding a new engine type for MultiCharts"When??

We have been waiting for SQX fully support MultiCharts for 6 months since Oct, 2019.

We can can't wait for it any longer!

Furthermore, please check every function and indicator for MultiCharts, please don't be lack of them or unable to compile in next version.

It really makes users very tried of it.

I was checking the fuzzy logic processing both in Tradestation and Multicharts and it worked fine. I got the same results as in SQ. So the fuzzy logic should not be the cause of the problems.

I meant that when you import M1 data and want to do a backtest on H1 TF, the bars are different in TS and MC. Maybe M60 is not the case, I haven't tried that.

To get matching results between SQ and MC, you should export the data you are running your study on and use that data when backtesting in SQ to be sure the data are exactly equal.

Next thing is to use SelectedTF and NoSession. Session is already taken in account in the data itself.

We are thinking about adding a new engine just to simplify the backtesting process for our users. And to enable reliable backtesting even if using lower timeframe based data. 

I have already added missing functions and fixed compilation errors. We are also testing various kinds of strategies generated by SQ and comparing results between platforms. 

I believe it will work much better in the upcoming build.

I tried the method as you said, but the results are even worse in Multicharts. It didn't work out.

I exported M20 data from Multicharts, and then imported to SQX. I even deleted the session, and selected "No Session".

I'm sure my settings are all correct, but the results still do not matched between Multicharts and SQX.

In my experience, B119 is more reliable than B120.

Furthermore, EOD and EOF strategies in B120 are still not reliable, results don't match with each other.

Hopefully all bugs can be free in B121.

I attach screenshots of my config.

New build should be available next week. If you have any problems in the new build, let me know.

And what's your MC version? My version is newest version as screenshot.

If you need, you can teamviewer to my PC to check everything.

Here is all historical data of M1 : https://drive.google.com/open?id=1xciiAdlKMDi63Ert3EhIDL9fpgM1Amhj

I tried to do a backtest with your settings - OOS is not needed here, it doesn't make any difference. I must have set the commission to a half value in SQ, but it worked also.

Ok, I can log in to your PC and check it out, it's not a problem.