Rebooting Weekly Tourneys <Brainstorm>

[Water]

I'm thinking about rebooting the weekly tourneys. I decided to make it more simpler so that it's more easy to manage by those running the tourney. Your feedback would definitely be appreciated.

Plan
Tourney Host will decide on the specific roles and their limitations and will play the tournament. The tournament is composed of three consecutive games at a designated time on the US server.

Winning
The winner will be the one who accumulates the most points during the three games. Rather than using an overly complicated scoring system, the points will be awarded based on their role. Generally, the harder the role, the more points they are rewarded for winning or losing. There won't be individual scoring or anything -- it is simply a results-based approach. This is to alleviate administrative headache.

Point distribution [Win/Loss]
Town: 100/0
Mafia/Triad: 100/0
Survivor: 150/50
Jester: 100/0
Executioner: 100/0
Amnesiac: 100/0
Serial Killer: 150/50
Arsonist: 200/50
MM: 150/50
Judge: 150/50
Witch: 200/50
Auditor: 200/50

Award
To be determined, but probably around 5-10 dollar range.


So, what do you guys and gals think of this approach?

[Cryptonic]

I don't see how that above system works.
Not every role has the same chance to win every game. So, shouldn't be weighted as such imo.

[Water]

I don't see how that above system works.
Not every role has the same chance to win every game. So, shouldn't be weighted as such imo.

I wish I had the data for that. What I ideally wanted to do is factor the points by the likely expectation of the role winning, given that all else is equal. The above is my crude attempt to try to create that metric, which obviously didn't end up so well.

I was considering making the number of games higher, say 5 or 7, but it would take too long.

[secondpassing]

I support this idea in general.

I do not agree with some of the point loss compensations. Losing as Judge, Witch, or Auditor seems like a win. Perhaps reduce it to 25?

You can count me waiting on the first tournament.

[Cryptonic]

I wish the balance weightings were still shown at the end of the game

[Water]

I wish the balance weightings were still shown at the end of the game

So do you think that, if the points were proportional to the likelihood of winning, the score system outlined above is fair? If that's the case, the focus should then fix to the point allocation. If you think the idea is still shitty in general, then I need to revise my approach and go back to the drawing board.

[Cryptonic]

I think it's good to just allocate points based on wins. Losses I don't really see a point to. But the system like this prevent bias and personal opinion, which is always good.

[Water]

Town: 100
Mafia/Triad: 100
Survivor: 150
Jester: 100
Executioner: 100
Amnesiac: 100
Serial Killer: 150
Arsonist: 200
MM: 150
Judge: 150
Witch: 200
Auditor: 200

So, I'll remove the points for losing, what do you guys and gals recommend I should adjust, and by how much? Reasons?

[Cryptonic]

Probably not fair, but it's simple
If you had time and patience, you could apply every random slot (like town killing, ect) a value. Mafia/Triad being higher value, and evil neutral even higher.
Then a players points is =(total valur of all players who lost)/(total value of all players who won).
Cult would complicate this, though. But i think it works because the more opponents you had, and the more strong power roles they had, you get more points if you win. And the weaker your team was in comparison, the more points as well. Also allows random any to be in the game while retaining a balance in points.

[Water]

I honestly think cult/mason is too overpowered and would not want to put them in a cash-game tourney. I guess I will try to adjust points as I experiment more. I, for now, need to figure out the logistics to spread awareness. I will try to start this up within a week or two.

[Sen]

I wish I had the data for that. What I ideally wanted to do is factor the points by the likely expectation of the role winning, given that all else is equal. The above is my crude attempt to try to create that metric, which obviously didn't end up so well.

I was considering making the number of games higher, say 5 or 7, but it would take too long.

Stats from 206 games I kept track of a few years ago. These are end-screen wins, so the listed Neutral wins are solo victories.



Mafia
122
39.87%


Town
121
39.54%


Cult
24
7.84%


Serial Killer
19
6.21%


Arsonist
13
4.25%


Witch
4
1.31%


Jester
2
0.65%


Mass Murderer
1
0.33%



As for your points distribution, I think Evil and Benign Neutrals other than Jester and Executioner should get more points; given that people will be competing for points, it's fair to assume that the winning faction will deny as many victories as possible (ie; lynching a confirmed Survivor or the Mafia killing the Witch).

Also, like Cryptonic said, points should also depend on specific roles. With only 3 games and the current points distribution, there's a big enough chance that many player will be tied on the first place.

[Water]

Thank you Sen, that chart really helps give some insight on the probability of winning.
Maybe I should impose a penalty for killing a neutral, either by lynch or murder. For benign, both mafia and town get point reduction, for evil, only mafia gets a point reduction.

I would need to figure out how to differentiate between town/mafia without requiring too much administrative workload. I feel inclined to not use points for specific roles for town/mafia, because it detracts from "skill" to put more weight on "luck." Maybe I could reward points for successful mafia identification/lynches(mayor/marshal)/executions, but that adds to the administrative burden. Alternatively, I could just split payment among the winners, or use a death match to end draws. Maybe night immune arson, jester, and non-night immune GF, for example.

[DarknessB]

Maybe I should impose a penalty for killing a neutral, either by lynch or murder. For benign, both mafia and town get point reduction, for evil, only mafia gets a point reduction.

The idea is to evaluate peoples' play per the game's normal rules, not change how they play the game though? I don't understand how lynching or killing a neutral is necessarily a sign of bad play and should be penalized?

In particular, there are a number of Neutral Evils that cannot win with the Mafia (Cultist, Witch Doctor, Serial Killer, Arsonist, Mass Murderer) so you'd essentially be penalizing the Mafia for doing their job. Also, merely because certain Neutral Evil roles (Auditor, Judge, Witch) can win with the Mafia doesn't mean they're always going to work together and in fact, they often don't due to a lack of recognition of each other.

Lastly, it can very much be advantageous to remove Neutral Benigns from the game who might be inclined to side with the other faction. For example, Amnesiacs, Executioners, and Survivors, who are going to be more concerned about their particular win condition than helping the Mafia or Town prevail.

I understand your motivation in wanting to give Neutrals a better shot in these games, but I don't think trying to influence how people play through penalties is the right idea.

[Cryptonic]

Yea, I would be against anything that tries to prevent someone from killing other players. Adds to much meta into the game. Scum will use to advantage and claim Benign so that Town doesn't lynch at risk of losing points, ect. Has too much affect on the game play.

Also, Benigns are pieces of shit and should be killed.




The problem w/ Sen's chart is that it doesn't take into account games where it's impossible for a team to win due to absence from game. A SK can't win if there is no SK in the game. You can see that this is the case as the total % is 100%.

[Water]

Yea, I would be against anything that tries to prevent someone from killing other players. Adds to much meta into the game. Scum will use to advantage and claim Benign so that Town doesn't lynch at risk of losing points, ect. Has too much affect on the game play.

Also, Benigns are pieces of shit and should be killed.




The problem w/ Sen's chart is that it doesn't take into account games where it's impossible for a team to win due to absence from game. A SK can't win if there is no SK in the game. You can see that this is the case as the total % is 100%.

Yes, I retract my previous idea about losing points for killing neutrals -- it was a terrible idea.
However, there's some underlying premises that I would like to clear up before scheming:
The meta inevitably will change because the incentives are different. For normal mafia, several parties can be winners, but there can only be one winner in the tourney. This almost always happens under a tourney structure. Even in SC2, people play tournaments differently, using more risk. In a normal ladder game, it's about risk-averse play because you want to maximize long-run wins. In tourneys, players tend to do all-in strategies, or deviate from normal play, because it's all about taking down the opponent to stay alive in the tournament. So basically, tourney structures will change the way people play, I think some kind of protection scheme could help mitigate the changed meta from the tourney structure.

Therefore, neutrals will likely be targeted once the outcome is clear. In terms of meta, without a neutral protection rule, rational neutrals should always side with town given ambiguity, because mafia always has the incentive to kill non-mafias before taking the win. I can think of two solutions:

a. Abandon attempts to incentivize and give points to neutrals for losing, sort of as a bye, because being neutral is practically an autoloss.

b. Try to change the incentive structure by rewarding extra points for keeping certain neutrals alive. I think in this scenario, there's an incentive to keep neutrals alive, which gives neutrals a chance, making it a more fun game for the neutrals. Here, I'm trying to undo the damage against the neutrals because of the tourney structure.

Whatever the choice, I'll defer to you guys since you all likely have more experience than me.


I understand that Sen's chart isn't useful when scrutinized in specificity, but it gives me a rough guideline, e.g., neutrals win about 20% of the time. I was also trying to make Sen feel good :)

[Sen]

I was also trying to make Sen feel good :)
I have vodka for that.

The best would be to simply give extra points to some and let everyone play as usual. Another thing to consider is setting the poins distribution after designing the setup, since the probability of a faction winning depends on that, and the rewards should reflect it.

[Cryptonic]

I think Sen's chart can work if you combine all non-Mafia/Town alignments into one category.
So, 20.59% of the time, neither Town nor Mafia wins. 200+ games is a decent sample size.
How'd you distribute points from that to Non-killing Evil Neutral roles, though, I'm not sure lmao.

Edit:

Maybe the average?
IE Witch = (39.87 +20.59)%/2 = 30.23%
So, that'd be a guess on the Witches chance of winning. Not sure how fair that is, but I think it's getting somewhere lol

[Sen]

Maybe the average?
IE Witch = (39.87 +20.59)%/2 = 30.23%
So, that'd be a guess on the Witches chance of winning. Not sure how fair that is, but I think it's getting somewhere lol
I think it all depends on the setup; the possible roles along with the smallest changes in a single role's settings would heavily change the chance of a particular role or faction ending with a victory.
ie; a Witch win in a setup with a Neutral Evil, no notification of witching, a Janitor, and few TPRs would be immensely easier than one in a game with a confirmed Witch along with a couple Veterans and Sheriffs.

Another thing is that it's fair to surmise that roles which are hard to win with in public games will be even more difficult in a private game filled with decent players; in a pub you can very well be a confirmed Sheriff as an Arsonist, and pull out a mislynch on a revealed Mayor, because they're dumb like that, but I doubt we'll be seeing anything like that here, so this should be taken into account for the points distribution and especially when designing the setup.