I found the following formula for calculating ELO
Bonus = K * (V - 1 / [1 + 10**((EloOpp - EloOld) / 400)]) where EloOld = Your elo EloOpp = Elo of your opponent V = (1 for a win, 0.5 for a draw, 0 for a loss) K is the development coefficientI used 30 for K.
I then rearranged some pre-existing code to apply this formula to all the archived games I have, arranged in game number order. The complete results of this program are available below, as is an ELO ordered list of active Firetop Mountain users.
K * (v -1/[1 + B ** ((EloOpp - EloOld) / S])) where K ~ 25, B = 2, and S = 100Ratings for new users would start at 1000.
This would make an average user have a rating of 1000. A difference of 100 points between users would indicate the higher rated user should be expected to win twice as many games.
Yaron Racah suggested a K of 40 and I have re-run the ratings. The order has changed a bit from the earlier run as users with fewer games are getting moved farther with the larger K value in use.
A running report as ELO ratings were generated
ELO ratings for active users
I'm going to try to look at Glicko's, suggested by Yaron Racah and documented at http://www.glicko.com/glicko.doc.glicko.html. If I can figure it out I will post sample ratings here.