Thank you for the interesting and very complex question. What facilities do we already have in Stake Wizard that are related somehow to the requested drawdown threshold? The answer is risk distribution diagram: Stake Wizard -> Variants -> View reports -> Diagrams -> Risk distribution. Here is an example:
Although math expectation of Z is 1.24 (expected ROI = 100*(Z-1) = 24%) in this example, the range of possible values of Z is rather wide: from 0.1 to 4. Average number of betting rounds required to reach the target fund 10000 (starting from 1000) is 10.7: 1000 * (1.24)^10.7 = 1000 * 10 = 10000. By the way, complex bet with such parameters can be considered as very fortunate: on average we need just 10.7 betting rounds. The question is how the above diagram would look as a result of the whole betting season comprising those 10.7 rounds? The answer is unbelievable: the range of Z would be 0.1^10.7 ... 4^10.7 = 0.00000000002 ... 2767208. On the one hand less than 1 penny, on the other hand a billionaire! Those possibilities do really exist, but their probability approaches to zero. Real Z values are smeared between these extreme results with the mean value = 10.
I think it would be useful to implement 'whole season' diagrams in Stake Wizard in order to better understand possible risks of the long run betting.
To get higher level of a drawdown in a single round, we'll need higher value of x0 - amount to be set aside. This yields in greater number of required betting rounds which diminishes the whole season drawdown. This contradiction may be resolved by an accurate computations.
Anyway the full answer requires additional research. Also, we'll need to develop new bet optimization procedures with drawdown constraints.