Kind of sad when you think about it, it's like there is no standardised rating, just five thousand useless variables. Thankfully PR is still in its beta stage.
-----------------------------------------
Power Rating factors in Civ Win %, which currently look like this:
Spanish 48.09%
Germans 53.98%
British 40.89%
French 52.33%
Ports 49.32%
Russians 50.81%
Ottoman 53.41%
Dutch 50.08%
I'm sure many of you will agree that this is hardly reflective of the strength of the civs as they are at the moment, certainly with Spanish..
Its a stupid thing to factor in to a rating regardless - in AoM 1.06 Set was extremely strong at an expert level - but that doesn't mean Set is extremely strong at a noob level. So the real percentage could be 35% wins at noob level and 65% wins at expert level, but the average would be factored in for both.
The HC component of Power Rating works in tiers..
lvl 1 to lvl 10
lvl 10 to lvl 25
lvl 25 +
So games between lvl 25 + HCs means that the HC lvl has no effect (the way it should be).
However, from this information, easiest way to get a high Power Rating would be playing up Spanish HCs until they reach lvl 25. Its not like you need lvl 25+ cards for a simple pike rush anyways.
Just another reason to civ hop over to good ole Spanish.
I'm sorry if you know this already but I had only read less specific info.
(As a side note, Swinger said the main problem with ELO was that it couldnt handle team games... I've never heard anyone complaining about ELO not being able to handle team games and don't see the logic behind that statement... If there was, then theres a simple fix for it - splitting the rating into team and 1v1.)