The idea that I want to try for MP balancing is simply calculating Elo ratings for every mech.
It would get way too complicated for custom mechs, but would likely be helpful for stock matches. Figuring out the actual math of it is slightly over my head and would take some research if I was doing it, but my statistician brother claims that it's fairly straightforward conceptually. If you have a way of computing the expected win probability for a given match, then Elo simply adjusts the weightings towards the observed results... incrementally improving the accuracy of the expected win probability.
It would get way too complicated for custom mechs, but would likely be helpful for stock matches. Figuring out the actual math of it is slightly over my head and would take some research if I was doing it, but my statistician brother claims that it's fairly straightforward conceptually. If you have a way of computing the expected win probability for a given match, then Elo simply adjusts the weightings towards the observed results... incrementally improving the accuracy of the expected win probability.