Issue with Optimizations

[jared77]

Hello, I have purchased your software but recent update has issues. I am trying your hack/trick where you use control point test first to help defeat the slippage issue....but the question is - optimisations do we need to do anything? you see I optimise the EA, then when you retest one of the optimised sets the system shows two very different test results when you use the above hack/test - and quite simply I cannot trust which test is the correct one. Its a real issue, as I am sure many people are optimising and thinkgint the new optimised set is perfect yet when you do the hack test above, it really shows different results.

So, do optimisations work now? or is there a trick we need to do when we optimise?

[tickstory]

Hi Jared,

Re: Optimizations, we have double-checked to make sure spread is being correctly set in the back-tester during optimizations by using the following code:

Code: Select all

     int filePtr = FileOpen("Spread.txt", FILE_WRITE);
     FileWrite(filePtr, DoubleToStr(MarketInfo(Symbol(),MODE_SPREAD)));
     FileClose(filePtr);

All looks good, so I assume that there is some other reason for your differing results. If you haven't already, check that your input parmeters exactly match or try to do an optimization with a result set restricted to a single result you know works.

Hope this helps.

[jared77]

second test gives poor result after using hack method on optimised parameter

OK, so explain how please. I assume I can choose my EA and currency, run an optimisation test and have say 20 options at the end to choose. I choose one of them and set that new parameter, then what I am currently doing is running a test on the setting OPEN prices (as my EA opens only on a new bar)....

if I do a test without using the hack methods - my results i get is spot on correct matching the optimisation BUT when I try your trick hack (by pushing the control point, stopping then pushing the Open prices test) it gives me a VERY different result that the optimisations and poor results?

[jared77]

but if i do a test right after optimisation, using the open prices method - the test runs true to the optimisation...maybe I dont have to do the hack method on this EA? it seems weird though how its retaining the poor results after I run that hack trick.

[tickstory]

Hi Jared,

Just to firstly explain the "trick": All this does is essentially work-around an MT4 issue where spread is not being set when using Tickstory tick data. To be using Tickstory tick data, you need to be using the "Every tick" modelling method. Given that you're trying to use the "Open prices" method (and not Tickstory tick data), you should not need to do the workaround.

That said, it's curious as to why your results are changing when you switch from Open Prices to Control points and back again to Open Prices. Have you tried to do this sequence without using the optimiser? Do your results change? If so, this seems to indicate another issue with the Strategy Tester.

For the moment, if you're working with Open prices you should not need to do any workaround which should solve your issue.

Regards.