Did you include prince
Number of killers can be modeled as a Binomial distribution with success rate 3/15 = 0.2 (barring uniques muddling this slightly) and 10 trials (actually 9.5, but I assume these opposing effects roughly offset each other). Expected value of a Binomial distribution is success rate * trials = 2 BD killers spawn on average. (Having at least 1 spawn happens with roughly chance 0.9) (Having more than 3 spawn happens with roughly chance 0.3333). This doesnât take into account Butler/Chronomancer Vig/Noble Bounty. This would raise your number somewhat.
Taking the average of a minimum and maximum scenario also leads to a skewed view, because it doesnât take account the probabilities of those extremes happening, nor anything in between actually. I realize you have to try to estimate something, but that approach feels flawed.
Evils also start off with fewer members, so just making a KPN comparison by itself isnât enough when you donât take into account their initial positions. Itâs not like a 8vs8 race where they kill each other as quickly as possible. Unseen starts with 2.35, BD with 11.15, neutrals with 1.5 and NK with 1.
NK is also incentivized in their win condition to townside if scum gains the upper hand. Doesnât mean they are so problematic.
These are just some comments I would have on that. I donât think Mathematics can capture the complex interactions here fully. If I fucked up on something, itâs been too long, since Iâve had to apply probability
1 Like
Speaking as someone who did a very bad job of doing this, how good have NKs here historically been at targeting scum when they need to?
Scum rarely have the upper hand, so it occurs too infrequently for that to happen.
I was very good at it
https://wiki.mafiascum.net/index.php?title=The_EV_Project
This might give you an idea of how complicated it is. Now add multiball, conversion, variable spawning rates and all PRs and the fact that the former is merely theoretical and there is still a âsocial deduction factorâ that you canât really account for with anything other than simulation, which we canât do as for example ToL can.
Oh and varying player skill is also a big element, but thatâs kind of the determining element youâd want. It matters who rolls what.
2 Likes
11P 9 - 2
(2/11) 18.2% GOTO 9P: 8 - 1 (187/315 Town win)
(9/11) 81.8% GOTO 9P: 7 - 2 (94/315 Town win)(2/11187/315 + 9/1194/315 = 244/693) 35.2% Town win
(447/693) 64.8% Scum win
what the FUCK.
We can write down the facts, but a complete math formula for KP⌠is not helpful. Youâll see shortly why.
Kills on daytime: 0-1 (2 if Bounty used on a BD)
Unseen kills on nighttime: 0-2
Cult kills on nighttime: 0-2
Conversions (donât know how you would count them, but they are kinda like a doublekill, since you get someone on your side AND remove something from BD)
NK kills on nighttime 0-n (can theoretically kill a lot of players at once, but thatâs luckbased)
BD kills on nighttime: 0-n Prince, Knights, Archers, Hunters
Bleed kills: 0-n (accessable by all factions)
Attacks wonât work ifâŚ
- they get prevented by a jail, merc guard, CW etc.
- they get occupied
- they hit a deathimmune target
- the person was protected by a Knight, King, CW etc.
- the attack was healed
- Attacks can be redirected to a unwanted target by a Drunk, Chronomancer etc.
Neutrals have no kill power on themselves, but they can make kills more likely/unlikely.
Also, executing neutrals is basically a mislynch for every side.
(I probably have overlooked some possibilities tho, donât stone me 
)
Premise: At even numbers, Town is assumed to No Lynch to raise their lynch accuracy as well
2 Likes
Please update the premade layout when you click to create a thread.
Like, 5 Irl days of daytime.
3 Likes
I think he means when you press âCreate Topicâ in the forum games category it gives a premade setup template. Itâs outdated.
