Automatic Sensitivity Test for variables in a strategy to be used in montecarlo/optimization

One of the caveats of a blanket test such as +x%/-x% of all or random variables is that not all variables are equally sensitive. A 5% change of a SL in a strategy with other exit rules could have a low effect on the results. Some indicators will have variables more sensitive to a 5% change and others less sensitive. Strategies with more variables will have a higher effect to its results. This brings a not so talked about but distinct bias towards strategies with less sensitive variables and less variables. 

I propose a method to test sensitivity before invoking the optimization or montecarlo. We can measure the change of each individual variable has on draw down, number of trades etc.,  then take the max change % found in any of those categories and use that in an algorithm to determine sensitivity of each variable and calibrate the %change differently for each variable depending on total number of variables and their individual sensitivity. In the UI there could be estimated performance change in percent and number of steps, no need for anything else when using the automatic sensitivity test and calibration.

Anyways, just an idea, no hurry to see it implemented and not critical at all. Just thought I'd put it out there and see if anyone has anything to say or add.