One difference between Zorro versions is that the test date range is fixed in the script and is only updated in subsequent versions by the zorro team. For this reason when you test the same Z script in different versions of Zorro you are testing over a different time period that you can't change. So the results must be different. Note you can't retrain a Z script yourself as the dates are fixed and you will always get same params. You can't change anything to change the params.

Thanks for posting your results. I wanted to look straight at your USD/CAD as my Z12 has also misplayed this pair recently. Z was long the way down from 1.38 and then short the recent bounce from 1.20 to 1.25. It suffered big losses. So I can confirm same results from my side.

Furthermore I foresee the downside of Zorro's WF optimization. A subsequent release of Zorro Z12 will re-optimize usd/cad over this time period and find params that were profitable. Hense this drubbing won't show up in reported back-tests. It would be nice to take an old version of Zorro Z12 like 160 and test it on data that came after it was released. But unless I am mistaken about the above you can't do that.

All this said I don't see this Zorro's fault per se. Zorro was optimized on the past and then market did its own thing. To be expected frankly and what the game is about. We should be developing our own scripts then we know exactly what has happened and what could be done for a better result.

later