OK, so the Capital variable does not influence drawdowns? As long as I apply the square root rule and optimalf for margin allocation and the Capital variable is more than the required capital without reinvestment I am on the safe side (at least in theory)?

Sorry, but if so, I think it contradicts the definition I quoted above of capital required when testing with reinvesting, which is the capital required to avoid a margin call on the test period.
Unless it is that applying the square root rule and optimalf for margin allocation already takes care of avoiding a margin call but then I don't understand why I get that the capital required when investing 1000 is 4119 if I applied optimalf and the square root rule for training factors...i am puzzled.