What happens with the others?

Sebastian Arsjo wrote:

Not a problem that 1 H difference makes such a huge impat on performance

Depends on how you look at it. Each of these is optimised for that time slot. Historically, that’s the crossing pair that worked best. you could say that it’s extra curve-fitted, as it works on a particular curve within the curve. Or (the way I see it) if you take them globally as 4 parts of 1 strategy, then you avoid the curve fit by running 4 variations simultaneously. Map the individual results on top of each other and usually, where one takes a loss, one of the others will compensate. I hope that’s the way it works. Certainly (in back test) it is superior to running one algo over the whole day without OTD.

So far, forward testing is  consistent with the back test results, but it’s early days.

Sorry, this is another instance of PRT not saving changes! Everything was moved from the optimisation box before I exported it.

The attached is what it should look like, with all values inserted into the code.

GraHal wrote:

Out of interest, after you moved values … did you save by using the X in top right hand corner and then clicked Yes to Save? Then exported?

That’s my normal procedure, so pretty sure that’s what I did. Lo and behold, everything went back into the opt box.

Because PRT. Live and demo often behave differently. If you look around the forum you will find this discussed in various places.