#Ranked by πΉ.π. π·ππππ
1 messages Β· Page 1 of 1 (latest)
Welcome @obtuse cape! Any feedback from the community will show up here.
Please note that this thread discussion is not shared with SMG.
It is solely for the RISK Community to discuss ideas & provide constructive feedback!
BACKGROUND:Β The current Rank system uses an Elo-based system designed for chess.Β Elo results in severe point swings when players of different skill point levels compete (even within the same class).Β It assumes that all players are seeking to win first place and are playing on accounts that represent their actual skill level (vs. alternate accounts).
Elo rank changes are difficult for players to calculate because it requires knowing the skill point level of all players. This data is not readily available.Β With Elo, at high levels, it can take over 30 consecutive wins to recover from a single loss.
ALTERNATIVE:Β The numbers below are provided for discussion and can be adjusted. A simplified rating system might award 100 points for a win against an opponent in oneβs class.Β A player might receive 200 points for a win against an opponent above oneβs class.Β A player would only receive 25 points for a win against an opponent below oneβs class.
Conversely, a loss to someone in oneβs class would cost 100 points.Β A loss to someone above oneβs class would only cost 25 points.Β And a loss to someone below oneβs class would cost 200 points.Β In this system, a loss to a peer could be recovered with a single win against anyone in the same class.
This system incentivizes playing strong opponents (you risk 25 points to potentially gain 200). It disincentivizes playing weak opponents (you risk 200 points to potentially gain 25). Perhaps most importantly, anyone can determine their point increase/decrease simply by knowing the class of their opponent(s).
I don't think you can beat basic ELO, nor is there anything wrong with it.
There are variants to it, but none can solve your issue that an inflated rating tanks massively on a loss. It has to.