Search:

You should definitely aim for:

-[COLOR="#ADD8E6"]Neutral killing[/COLOR] -> 15% to 20%

-[COLOR="#00FF00"]Town [/COLOR]-> 40%

-[COLOR="#FF0000"]Mafia[/COLOR] -> 40%.

That's roughly the numbers a fair setup should have in a "9T-3M-3N" type of structure. The NK should not have the same odds of winning than the two other factions given that he is alone.

EDIT: I sampled 80 mod mafia games 2 months ago, and those were the results: Town (49/80 = 61%), Mafia (24/80 = 30%), NK (7/80 = 8%).
This is data is, however, somewhat erroneous because a) not all of the games tracked were of the 9-3-3 type (In some setups, there wasn't even a mafia), and b) I am a good player and roll town most of the time since I -blacklist half of the mafia roles, so the mafia tends to lose more games than usual in my games, thus invalidating my sampling.

It does tell us that a setup, fair or unfair, does not necessarily warrant the winrates to be reasonable. RNG and players are factors that make it hard for any setup to have the desired outcome percentage.

You should definitely aim for:

-Neutral killing -> 15% to 20%

-Town -> 40%

-Mafia -> 40%.

That's roughly the numbers a fair setup should have in a "9T-3M-3N" type of structure. The NK should not have the same odds of winning than the two other factions given that he is alone.

EDIT: I sampled 80 mod mafia games 2 months ago, and those were the results: Town (49/80 = 61%), Mafia (24/80 = 30%), NK (7/80 = 8%).
This is data is, however, somewhat erroneous because a) not all of the games tracked were of the 9-3-3 type (In some setups, there wasn't even a mafia), and b) I am a good player and roll town most of the time since I -blacklist half of the mafia roles, so the mafia tends to lose more games than usual in my games, thus invalidating my sampling.

It does tell us that a setup, fair or unfair, does not necessarily warrant the winrates to be reasonable. RNG and players are factors that make it hard for any setup to have the desired outcome percentage.