I like the idea of the lookback or trending forecasting model as opposed to just a set 2-year snapshot. Ultimately, you want programs to move up to a next level because their PROGRAM is ascending and not just because a couple of teams did so in a couple of years.
The main reason that I suspect that most people who think that 2-year is too short and argue for 4-year isn't specifically because it's accurate, but because 1) it's better than 2-year in looking at program ascendancy as opposed to just limited team success and 2) it's easy to implement, as well as explain, than the lookback or trending modeling might be.
Edit: BTW, Scecina is a school that got bumped and didn't win a state final in doing so ... 1A to 2A with a pair of red rings. Happened in the first implementation of SF bump ups.