Plans for the next patch (3.2.5)

Announcements made here about the game and the company.

Re: Plans for the next patch (3.2.5)

Postby GreekGodSudura » Thu Jul 30, 2020 5:35 am

DFrenchBoi wrote:
GreekGodSudura wrote: Proportional winrates are for morons who demand participation trophies. Go home.

5/15 for all.


What are you talking about? A proportional win rate makes each faction member have to play well or they're going to lose. Giving 1 person an equal chance of winning as 13 people (NE is excluded since their win conditions are different) is simply ridiculous. Also, developers, PLEASE take note about how much people have played the game when you look at suggestions. There is a clear trend of more experienced players opposing these buffs, and those with less experience supporting them. That in itself should tell you something.


Yeah except this is a second account, you monkey.

I have WAY more games on my old one, JustSomeOtherGuy.

Congratulations, you played yourself.

Proportional win rates are the opposite of having to play well. It's roll town to win. It's roll NK to lose.

Playing well doesn't turn 1/15 chance into a win. You never HAD a chance.

Making it equal chances for all makes playing well turn it into a win. It LITERALLY openly means there's no RNG, it's whoever plays better.

You just demand FreeLO.
GreekGodSudura
Executioner
Executioner
 
Posts: 23
Joined: Sun Jul 19, 2020 1:52 pm

Re: Plans for the next patch (3.2.5)

Postby ydnnek » Thu Jul 30, 2020 6:37 am

Correct me if I'm wrong, but the discussion of "the win-rates for each faction" should depend on a few assumptions. For example, assume that a "normal" player whose skill is neither above-average nor below-average has a long-run average ELO gain of zero (this player seems to not be special at all, and never seems to neither decrease nor decrese their ran) then if we allow their win-rate to be 50% (provided ELO gains are tweaked to be a flat and equal increase/decrease), then setting (among an infinite number of choices) the winrate for NK to be 10%, mafia to be 16%, and town to be 74% then this will satisfy the "normal player" assumption.

(Here's something I just typed up quickly for the above argument. I assumed for ease of argument that the chances of rolling NK is 2/15 instead of 1/15: https://www.overleaf.com/read/vzsvzfcdcjnh
Note: I'm not trying to suggest anything. I'm only trying to show that it's not easy to just pick a winrate for a faction, without making some underlying assumptions first).

I can't see why you can choose between a flat winrate for each faction or a proportional winrate for each faction without first assuming something on the ELO increases/decreases.
Last edited by ydnnek on Thu Jul 30, 2020 7:25 pm, edited 2 times in total.
ydnnek
Jester
Jester
 
Posts: 15
Joined: Thu Jul 07, 2016 1:14 am

Re: Plans for the next patch (3.2.5)

Postby Freechancer » Thu Jul 30, 2020 6:50 am

I'm pretty sure the main concern is that it would put more pressure on town to play right and much less so for nk. They just need to kill townies. That's why nk is most powerful in low elo (think sk in classic). All any is, say, average elo, they know they should be hanging people so they just vote with maf all the time and 4-man maf wins at least 50% of games.
I think that nk should be fine in high elo, reaching 6v6 or 5v5 by d3 has been about declared the go-to win condition for evils. And it's town or nk that should come out the winner more often than not. Nk would be hanged (and maf won't vote nk if they feel threatened) or exed only if town takes the gamble to have escort or tp save the game for them (does this actually happen often?).
Freechancer
Jester
Jester
 
Posts: 12
Joined: Tue Jan 16, 2018 1:59 pm

Re: Plans for the next patch (3.2.5)

Postby GreekGodSudura » Thu Jul 30, 2020 7:52 am

ydnnek wrote:Correct me if I'm wrong, but the discussion of "the win-rates for each faction" should depend on a few assumptions. For example, assume that a "normal" player whose skill is neither above-average nor below-average has a long-run average ELO gain of zero (this player seems to not be special at all, and never seems to neither decrease nor decrese their ran) then if we allow their win-rate to be 50% (provided ELO gains are tweaked to be a flat and equal increase/decrease), then setting (among an infinite number of choices) the winrate for NK to be 10%, mafia to be 16%, and town to be 74% then this will satisfy the "normal player" assumption.

(Here's something I just typed up quickly for the above argument. I assumed for ease of argument that the chances of rolling NK is 2/15 instead of 1/15: https://www.overleaf.com/read/vzsvzfcdcjnh).

I can't see why you can choose between a flat winrate for each faction or a proportional winrate for each faction without first assuming something on the ELO increases/decreases.


If all factions have the same winrate, ELO would be the same - and it'd be actually skill based. Not the current system which is just that Mafia is the only faction that makes a dent.
GreekGodSudura
Executioner
Executioner
 
Posts: 23
Joined: Sun Jul 19, 2020 1:52 pm

Re: Plans for the next patch (3.2.5)

Postby Brilliand » Thu Jul 30, 2020 11:21 am

ydnnek wrote:Correct me if I'm wrong, but the discussion of "the win-rates for each faction" should depend on a few assumptions. For example, assume that a "normal" player whose skill is neither above-average nor below-average has a long-run average ELO gain of zero (this player seems to not be special at all, and never seems to neither decrease nor decrese their ran) then if we allow their win-rate to be 50% (provided ELO gains are tweaked to be a flat and equal increase/decrease), then setting (among an infinite number of choices) the winrate for NK to be 10%, mafia to be 16%, and town to be 74% then this will satisfy the "normal player" assumption.

(Here's something I just typed up quickly for the above argument. I assumed for ease of argument that the chances of rolling NK is 2/15 instead of 1/15: https://www.overleaf.com/read/vzsvzfcdcjnh).

I can't see why you can choose between a flat winrate for each faction or a proportional winrate for each faction without first assuming something on the ELO increases/decreases.


I don't think it's necessary to intertwine the winrate discussion with the ELO discussion, in fact I'd prefer not to.

I'd rather balance the winrates to something that feels "fair" regardless of ELO, and also (in parallel) set up the ELO system in such a way that it will give appropriate amounts of ELO regardless of what the winrates are. Setting things up that way is more robust than using a "flat an equal" ELO gain/loss, and trying to set the winrates to values that don't cause the ELO system to be broken.
My best role ideas:
Conqueror
Apocalypse Survivor
User avatar
Brilliand
Disguiser
Disguiser
 
Posts: 1028
Joined: Sat Nov 29, 2014 8:34 pm

Re: Plans for the next patch (3.2.5)

Postby BasicFourLife » Thu Jul 30, 2020 12:05 pm

There is a reason why no one has suggested this before or if anyone has, their ideas have all been thrown to the garbage. A double kill by SK, will guarantee them being exposed even more so as before.

The Arsonist buffs is flawed because of visitation roles work.

Werewolf changes are irrelevant cause rampage shouldn’t exist in the first place.

And Jugg is neutral, cause it is flawed upon a flaw upon a flaw.

Edit: Aye aye can’t believe psychic, a role which is 100% rng and player has no control over, didn’t get reworked instead.

Lookout changes are something something fm tried some time ago but 3 players is basically unlimited anyways and it will have no effect in 4/5 games with a LO.
I am kindly telling you to fuck off mate.

FM / TG: 5 - 10 - 8
BasicFourLife
[Forum Mafia XVI] Winner
[Forum Mafia XVI] Winner
 
Posts: 1566
Joined: Sat Dec 17, 2016 6:47 am
Location: UTC +2/+3 - Somewhere hidden

Re: Plans for the next patch (3.2.5)

Postby FerretBandit » Thu Jul 30, 2020 2:52 pm

Let me just say that Serial Killer change is REALLY bad. Escorts will just start outting who they're going to RB at the end of the day and when their will is cleaned they will be confirmed.

Instead, make Serial Killer Clean his victims, whether they were a role blocker or not. If he is RBed then he will only kill the RBer, but ANYONE the SK kills will be cleaned so only SK can see their role + will.

This means that Escorts dying to SK will no longer confirm who SK is even if they end up getting stabbed + cleaned that night, since it's unclear if SK attacked them directly of was roleblocked.
FerretBandit
Executioner
Executioner
 
Posts: 21
Joined: Sun Apr 26, 2015 10:49 am

Re: Plans for the next patch (3.2.5)

Postby GreekGodSudura » Thu Jul 30, 2020 4:23 pm

BasicFourLife wrote:Edit: Aye aye can’t believe psychic, a role which is 100% rng and player has no control over, didn’t get reworked instead.


Not to argue with the rest but I'm fairly sure they've said Coven roles won't be changed because it's paid. Which is some bulllllllllshit if you ask me.
GreekGodSudura
Executioner
Executioner
 
Posts: 23
Joined: Sun Jul 19, 2020 1:52 pm

Re: Plans for the next patch (3.2.5)

Postby MysticMismagius » Thu Jul 30, 2020 4:46 pm

The OP literally confirms otherwise, they're changing Jugg

So either you remembered wrong, or they changed their minds, either way "Coven roles won't be changed" isn't true
Image
User avatar
MysticMismagius
Blackmailer
Blackmailer
 
Posts: 1139
Joined: Sun Apr 30, 2017 4:46 pm

Re: Plans for the next patch (3.2.5)

Postby Soulshade55r » Thu Jul 30, 2020 4:50 pm

Didn't they change Necromancer and coven leader last patch lol?
\O.O/ \O.O/
User avatar
Soulshade55r
Vigilante
Vigilante
 
Posts: 614
Joined: Fri Apr 21, 2017 10:45 pm
Location: Uk

Re: Plans for the next patch (3.2.5)

Postby MysticMismagius » Thu Jul 30, 2020 4:52 pm

Lowkey I'm thinking somewhere along the game of telephone "Coven roles won't be brought into Classic" (which actually has been said) mutated into "Coven roles won't be changed" and that's how Greek heard that
Image
User avatar
MysticMismagius
Blackmailer
Blackmailer
 
Posts: 1139
Joined: Sun Apr 30, 2017 4:46 pm

Re: Plans for the next patch (3.2.5)

Postby GreekGodSudura » Thu Jul 30, 2020 5:52 pm

Soulshade55r wrote:Didn't they change Necromancer and coven leader last patch lol?


Necro was a parity change - I mean they refuse to remove or rework any Coven roles.
GreekGodSudura
Executioner
Executioner
 
Posts: 23
Joined: Sun Jul 19, 2020 1:52 pm

Re: Plans for the next patch (3.2.5)

Postby ydnnek » Thu Jul 30, 2020 5:58 pm

Brilliand wrote:
ydnnek wrote:Correct me if I'm wrong, but the discussion of "the win-rates for each faction" should depend on a few assumptions. For example, assume that a "normal" player whose skill is neither above-average nor below-average has a long-run average ELO gain of zero (this player seems to not be special at all, and never seems to neither decrease nor decrese their ran) then if we allow their win-rate to be 50% (provided ELO gains are tweaked to be a flat and equal increase/decrease), then setting (among an infinite number of choices) the winrate for NK to be 10%, mafia to be 16%, and town to be 74% then this will satisfy the "normal player" assumption.

(Here's something I just typed up quickly for the above argument. I assumed for ease of argument that the chances of rolling NK is 2/15 instead of 1/15: https://www.overleaf.com/read/vzsvzfcdcjnh).

I can't see why you can choose between a flat winrate for each faction or a proportional winrate for each faction without first assuming something on the ELO increases/decreases.


I don't think it's necessary to intertwine the winrate discussion with the ELO discussion, in fact I'd prefer not to.

I'd rather balance the winrates to something that feels "fair" regardless of ELO, and also (in parallel) set up the ELO system in such a way that it will give appropriate amounts of ELO regardless of what the winrates are. Setting things up that way is more robust than using a "flat an equal" ELO gain/loss, and trying to set the winrates to values that don't cause the ELO system to be broken.


Those were just assumptions made, based on "setting a win rate, dependent on ELO" such that there is no ELO inflation/deflation because you're more likely to land on Town.

I do agree that it should be balanced to make it feel fair (e.g. landing on NK, you should have a relatively fair chance of winning, assuming everyone is of equal skill/deduction level).

But for the purpose of ELO rank, I do think that the amount should be adjusted for these "fair" winrates, so that a "normal" player deserves a rank which doesn't deviate from the default (to avoid any ELO inflation/deflation). I understand this is probably quite difficult though, and unrealistic.
ydnnek
Jester
Jester
 
Posts: 15
Joined: Thu Jul 07, 2016 1:14 am

Re: Plans for the next patch (3.2.5)

Postby Brilliand » Thu Jul 30, 2020 6:31 pm

ydnnek wrote:But for the purpose of ELO rank, I do think that the amount should be adjusted for these "fair" winrates, so that a "normal" player deserves a rank which doesn't deviate from the default (to avoid any ELO inflation/deflation). I understand this is probably quite difficult though, and unrealistic.


It's definitely unrealistic to pick a fixed number that will always be appropriate for the current winrate over time.

What is realistic is to come up with an algorithm that will continually adjust the ELO gain/loss rates to be appropriate for whatever the current win/loss ratio happens to be.

The easiest way to accomplish this is what Superalex posted earlier: Set things up so the total amount of ELO gained by the winners and the total amount of ELO lost by the losers are exactly equal for every game.
(This creates issues for the NE though, so maybe it wouldn't be so easy once the NE issues are ironed out.)
My best role ideas:
Conqueror
Apocalypse Survivor
User avatar
Brilliand
Disguiser
Disguiser
 
Posts: 1028
Joined: Sat Nov 29, 2014 8:34 pm

Re: Plans for the next patch (3.2.5)

Postby Superalex11 » Fri Jul 31, 2020 12:09 am

Am currently working on a full-scale analysis of Trials reports and I've just pulled this from all reports not guiltied for gamethrowing in range 2225730-3000000, duplicates excluded:
(Edit: Removed, fixed in future post; stats shown here are still approx. correct)

Lots to go over here, but on the point of winrates I think it's useful to have the actual data. So here are some quick stats:
In ranked-rolelist games, town wins 54%, mafia 33%, and nk 11%. The rest are draws.
In ranked-rolelist games, NK winrate order is ww, arso, sk.
All-any is played about 25% as much as ranked-rolelist games (36% including CAA).
Last edited by Superalex11 on Fri Jul 31, 2020 6:01 am, edited 1 time in total.
After-Unity™

luls:
Spoiler: Image
Image
Image
Image
Image
Image
Image
User avatar
Superalex11
Investigator
Investigator
 
Posts: 267
Joined: Fri Jan 08, 2016 10:11 pm

Re: Plans for the next patch (3.2.5)

Postby HAWAIIANpikachu » Fri Jul 31, 2020 12:17 am

Interesting. Very interesting data.

I mean it's from reports so we might want to take cheating into some account as well but this is interesting to have as a base line
Town of salem User name: HAWAIIANpikachu
Town of salem ingame name: Perfect town/Below Is Afk
User avatar
HAWAIIANpikachu
Survivor
Survivor
 
Posts: 36
Joined: Sat Apr 04, 2015 10:49 am
Location: USA

Re: Plans for the next patch (3.2.5)

Postby Brilliand » Fri Jul 31, 2020 12:26 am

There's probably some skew from the "only games in which a report occurred" selection bias, but it's probably fair to assume that that's not too severe.

Superalex11 wrote:In ranked-rolelist games, town wins 54%, mafia 33%, and nk 11%. The rest are draws.


That's a lot closer to my "ideal" numbers than I expected. Just needs a small boost to NK winrate at the expense of Town.

Superalex11 wrote:In ranked-rolelist games, NK winrate order is ww, arso, sk.


...and that's definitely not what the folks over in Role Ideas were expecting. (I think the most commonly-guessed order was SK>WW>Arso.)
My best role ideas:
Conqueror
Apocalypse Survivor
User avatar
Brilliand
Disguiser
Disguiser
 
Posts: 1028
Joined: Sat Nov 29, 2014 8:34 pm

Re: Plans for the next patch (3.2.5)

Postby Superalex11 » Fri Jul 31, 2020 12:29 am

HAWAIIANpikachu wrote:I mean it's from reports so we might want to take cheating into some account as well but this is interesting to have as a base line

The above, but also excluding guilty cheating reports: (Edit: Removed, fixed in future post)
Last edited by Superalex11 on Fri Jul 31, 2020 6:01 am, edited 1 time in total.
After-Unity™

luls:
Spoiler: Image
Image
Image
Image
Image
Image
Image
User avatar
Superalex11
Investigator
Investigator
 
Posts: 267
Joined: Fri Jan 08, 2016 10:11 pm

Re: Plans for the next patch (3.2.5)

Postby HAWAIIANpikachu » Fri Jul 31, 2020 12:36 am

Superalex11 wrote:
HAWAIIANpikachu wrote:I mean it's from reports so we might want to take cheating into some account as well but this is interesting to have as a base line

The above, but also excluding guilty cheating reports: https://i.gyazo.com/5f06166e46f4dc0d04f ... 58c767.png

This is probably the most reliable way we can use reports to check winrates without including other advantages/disadvantages using reports. Thank you madlad for doing this
Town of salem User name: HAWAIIANpikachu
Town of salem ingame name: Perfect town/Below Is Afk
User avatar
HAWAIIANpikachu
Survivor
Survivor
 
Posts: 36
Joined: Sat Apr 04, 2015 10:49 am
Location: USA

Re: Plans for the next patch (3.2.5)

Postby Superalex11 » Fri Jul 31, 2020 6:05 am

Ok so I did a dumb and didn't actually go through all P2P reports, just 2020's. Made the fix, and also excluded guilty MA+Leaving reports since they line up with the idea of cheating and GT:

Image
After-Unity™

luls:
Spoiler: Image
Image
Image
Image
Image
Image
Image
User avatar
Superalex11
Investigator
Investigator
 
Posts: 267
Joined: Fri Jan 08, 2016 10:11 pm

Re: Plans for the next patch (3.2.5)

Postby Royee » Fri Jul 31, 2020 11:35 am

witch's winrate appears to be very low in rp and ranked.
User avatar
Royee
Easter 2020 Winner
Easter 2020 Winner
 
Posts: 209
Joined: Wed Sep 30, 2015 1:11 pm
Location: UTC +3

Re: Plans for the next patch (3.2.5)

Postby OreCreeper » Fri Jul 31, 2020 12:48 pm

Royee wrote:witch's winrate appears to be very low in rp and ranked.

Yes because witch wins aren't really supposed to happen, they are supposed to win with other evils.
OreCreeper
Doctor
Doctor
 
Posts: 177
Joined: Sat Jun 17, 2017 3:52 pm

Re: Plans for the next patch (3.2.5)

Postby Royee » Fri Jul 31, 2020 12:54 pm

OreCreeper wrote:
Royee wrote:witch's winrate appears to be very low in rp and ranked.

Yes because witch wins aren't really supposed to happen, they are supposed to win with other evils.

oh I was mistaken
User avatar
Royee
Easter 2020 Winner
Easter 2020 Winner
 
Posts: 209
Joined: Wed Sep 30, 2015 1:11 pm
Location: UTC +3

Re: Plans for the next patch (3.2.5)

Postby HAWAIIANpikachu » Fri Jul 31, 2020 7:20 pm

Superalex11 wrote:Ok so I did a dumb and didn't actually go through all P2P reports, just 2020's. Made the fix, and also excluded guilty MA+Leaving reports since they line up with the idea of cheating and GT:

https://i.gyazo.com/6c181f295c20faa09ee ... 248422.png


Quick stats for simplicity.

Town has about ~52% win rate

Mafia has about ~36% win rate

NK in total has about ~10% win rate

Still about ~2% of games are draws/stalemates/solo NE wins
Town of salem User name: HAWAIIANpikachu
Town of salem ingame name: Perfect town/Below Is Afk
User avatar
HAWAIIANpikachu
Survivor
Survivor
 
Posts: 36
Joined: Sat Apr 04, 2015 10:49 am
Location: USA

Re: Plans for the next patch (3.2.5)

Postby cob709 » Fri Jul 31, 2020 10:02 pm

tbh the game shouldn't be about win rates.

it should be about player skill and strategy
each faction should have equal chances of winning
hi friends
User avatar
cob709
Bodyguard
Bodyguard
 
Posts: 221
Joined: Sat Aug 12, 2017 9:44 am

PreviousNext

Return to Announcements

Who is online

Users browsing this forum: No registered users and 3 guests