I personally do not think that the ELO ranking is the more appropriate. It would be very good (reliable) if all games would be FFA, but most are not FFA.
So a good player can lose a game where his weaker allies lose badly and the good player gets base invaded while he was doing good in his front. That's just an example but it shows how average players can win over a very highly rated ELO player.
I propose to use the ranking that www.freechess.org (FICS) uses. The free internet chess server uses a ranking different from ELO with some advantages over the ELO. Like for example it is time dependent. By that I mean that someone that hasn't played within 1 year gets a less reliable ranking and his "points" will fluctuate much more easily than a regular player, until he plays some amount of games withing a rather short period of time.
If you're interested in this rating system (which does not differ that much from ELO), let me know and I'll investigate on FICS the exact mathematical formula to calculate the ratings. But from my own experience, it's pretty reliable.