@
Marshmallow Marshall
I don't have a huge interest in the ladder personally because I don't think that you will get an accurate correlation between win rate and player skill until you have an extremely large number of FMs to sample from (larger then what would be realistic for a season) and I see it more of such a fun little side thing. As such, feel free to implement it however you think that the community will most enjoy it.
Having said that, this is not my first time seeing the proposed point attribution system (Slaol talked about it before his untimely departure) and I have the following issues with it...
1) There is a massive emphasis on games played, far more so then any performance-rating system that would ever be used for anything at all competitive. If player 1 and player 2 have identical records after 6 games, then player 1 stops playing and player 2 plays another 6 games with the exact same record, player 2 will end up with double the points of player 1. This is fine if you want the emphasis on games played being this large, but terrible from a competitive standpoint.
2) I understand the desire to heavily reward players who win as neutral killers where the odds are stacked against you. However, the proposed winner reward system being based off of the total number of winners is terrible. It can be assumed that for any given "normal" setup where the main scum faction is a single mafia, the mafia should have an expected win chance equal to that of the town (if this isn't the case, there is a high chance that the game was poorly balanced). Under the proposed system however, a mafia winner would get double to triple the points of a town winner. As such, players who roll mafia quite a lot will over time end up with a massive advantage in point potential, despite not having any disadvantage in theoretical win rate.
Issue 1 makes the system non-competitive, but is still fine if you want to heavily reward people for participating in every single FM. Issue 2 compromises the entire concept of a fair ladder.
Proposed alternate system
The formula for this is going to be a bit more complex, but it does fix the issues mentioned above.
(1-(C/(C+GP)))*(WR/ExpectedWR)
C = A constant. Raising it increases the emphasis on games played, lowering decreases. I'll suggest that the value is 2, but this can be modified.
GP = Number of games played/hosted
WR = Win rate
ExpectedWR = Average expected win rate for a player based off of all their roles. This would take a bit of work to standardize win rates. However, if for example it was determined that in a standard townVmafiaVneutralkilling, town and mafia should each have a 45% chance of winning and that neutral killing has a 10% chance of winning, if a player played three games, one as town one as mafia one as NK, their value for this would be (.45 + .45 + .1)/3 = 1/3.
Then also to illustrate the values for (1-(C/(C+GP))) if C = 2;
If GP = 1, this equals .33
If GP = 2, this equals .5
If GP = 4, this equals .67
If GP = 6, this equals .75
If GP = 8, this equals .8
etc.
This creates a middle ground where playing games is rewarded while also not making GP the centerpiece of earning points. If you do want GP to be a major centerpiece, you can make C = 1 and it will still be better then the previously proposed system.
I know that I put quite a lot of effort towards something that I just said I didn't care that much about. However, as I said, Slaol showed me the formula before this and it triggered me quite a bit that he, or anyone else, considered it competitive.