Issue with using UPI and UI as Fitness Metrics

I’ve been trying to use UPI (User Point Index) and UI (User Index) as part of my fitness function, but unfortunately, it’s not feasible. While these metrics are quite interesting, favoring growth over depth and accounting for drawdown duration, they penalize performance inconsistently.
The problem is that UI penalizes early drawdowns more than those later on, as it compares against different net gain levels over time. This inconsistency makes it unreliable for a fitness function. Since UPI depends on UI, it inherits the same issue.
To fix this, I believe it would require recalculating metrics year by year, treating each year as starting from zero. That way, we’d get a clearer, consistent view of UI/UPI annually and globally. Unfortunately, as it stands, I can’t integrate them into my fitness function, even though they’d have been great to use.

Hopefully, someone can come up with an interesting solution to fix this issue because, in my opinion, these metrics are even better than the Sharpe Ratio, better than return over drawdown, and possibly the best measure of risk-reward. Unfortunately, I can’t use them to integrate into my fitness function to guide my strategies toward the desired behavior.