So I haven't really read this thread but usually team or player ratings for real sports have an average-results-based system that both requires a few matches to be played to get the rating up to a competitive level and phases out old results over time. This stops a team from being top of the latter just by playing tons of matches, or by winning one or two and leading with a 'perfect' victory record.
I would advocate a points system over a plain win ratio, because it's easier to balance and can take into account the how close the results are. For a best-of-three battle, an absolute steamroll could give, say, 50 points to one team and 10 to the other, while a close result would give 40:20 (or maybe even as low as 35:25), and a tie 30:30. Naturally these numbers could be adjusted as needed. The clan's ladder rating would be the scaled average of their match scores. Old results would phase out over a given period (maybe two months/a season), with the option of penalties for inactivity and the like.
All that's really needed for the scaling would be some kind of multiplier that starts low but reaches a fairly stable value after about 3 matches. Based on current clan activity I would start with something like
1 - (0.75 / no. of active matches)
which would mean that at least two matches need to be played to make any kind of impact on the ladder.
I guess the main downside to this method is that individual round results would need to be taken into account instead of just overall match results, but I've been messing around with potential outcomes and I think it's at least well-balanced.
P.S. 19 new replies what the hell
You must have been typing that for hours, because the thread's dead now, we're just waiting for the moderators to lock it.
Poll
