Well, I gave it a shot with my four most played decks and... now I feel a little bad.
Ezuri, Elfball Combo: 12.57
Scion, Dragon Combo: 7.21
Mayael, Big Boi Tribal: 7
Thrasios and Silas Renn, Myr Combo:...13.43
So... I wasn't expecting that. The Scion score surprises me a little, but I think it's mostly due to the unorthodox nature of the deck. I didn't count reanimation in any way, and I only counted Scion for 2 in the tutors slot, and he's the only tutor in the deck. I feel the deck would score higher if I interpreted it a little differently, and probably be more reflective of its actual power level.
On the other hand, I think I might have to stop calling the myr combo deck silly. And maybe apologize to the people I've played with it. It is pretty consistent and it does have a lot of layered routes, but wow I was not expecting it to beat Ezuri.
Overall, I think the formula does a decent job. It needs to have variations for graveyard decks. I know the original post mentioned ways to adapt it, but there should be more concrete solutions. Leaving every graveyard strategy open to interpretation means the formula can be applied asymmetrically with those decks, leading to improper analysis. I also think the removal portion could be separated into more variables, lending different weight to stax effects, targeted removal, and board wipes. I would even go as far as to say that certain kinds of board wipes should be given higher value than others (symmetrical v asymmetrical, types of permanent hit, destroy/
exile/bounce). Still, I'm pretty happy with this and look forward to the eventual, glorious day we can somewhat quantitatively talk about deck power levels.