Mafia Games ELO thread

[ketchup777]

town rating+mafia rating / 2

[Dannflor]

my problem with that is that i feel like that would just mean good scum players will rise above good town players since scum games grant more ELO than town games

I don’t think that’s true? Because people roll town way more often than they roll scum on average

[Gypyx]

yeah i tried to correct that but sadly scum still has more gains than town

3 town wins barely reaches the gains of 1 scum win

[Aureal]

Wow I liked your previous sampling of games where I was right in the middle of both sheets better than this sample that puts me firmly at the bottom

(Cee Mirage and Duelist Kage are my alts, so 0-3)

Not that I'm saying it's not more accurate (the first one made me lol at how backwards it seemed)

I honestly don't understand what being "tracked" means, all the info except the last column shows up for everyone so now everyone can see that I'm the worst town player by your rating regardless of whether I agreed to be tracked or not.

[Gypyx]

A game will only count if it has a tracked player in it basically

also if you're unconfortable with your rating being public i can hide it

[Aureal]

But what does "count" mean? It all shows up for everyone. You're comparing players who asked to be tracked to players who didn't and rating them all and displaying it all to everyone. Having the color of the boxes be grey instead of gold doesn't make it so the information is actually hidden. An opt-in system should only be showing those who actually opted in.

[Thestatusquo]

I was also confused as to what tracked meant.

It seems like it means "based on a complete sample of data" whereas untracked means "incomplete" over the time frame for which sampling was done.

[yessiree]

I believe "tracked" means having >3 games of data collected for them, and "not tracked" means you just happened to be in games with tracked players

Also worth noting this ELO rating is not a measure of skill/proficiency of any kind and is only a fancier way of tracking wins/losses

[KayJayQueue]

I believe "tracked" means having >3 games of data collected for them, and "not tracked" means you just happened to be in games with tracked players

Also worth noting this ELO rating is not a measure of skill/proficiency of any kind and is only a fancier way of tracking wins/losses

It measures my proficiency in losing!

[Aureal]

It's clearly not based on how many games have been input, Jackson is top of town games with 7 and isn't tracked.

I don't really wanna get into a debate on skill correlating with win rate other than to note that it's pretty obvious that it can be felt that it does, whether true or not. (hi Kay)

[KayJayQueue]

Hi love

[Alianna]

I believe "tracked" means having >3 games of data collected for them, and "not tracked" means you just happened to be in games with tracked players

Also worth noting this ELO rating is not a measure of skill/proficiency of any kind and is only a fancier way of tracking wins/losses

It measures my proficiency in losing!

I'm better at losing than you, get rekt noob.

[yessiree]

It's clearly not based on how many games have been input, Jackson is top of town games with 7 and isn't tracked.

I don't really wanna get into a debate on skill correlating with win rate other than to note that it's pretty obvious that it can be felt that it does, whether true or not. (hi Kay)

yeah you're right. we talked about obfuscating the usernames in some way. python suggested to use the user id instead of username

would that be better, or does that not really solve the underlying issue of having leaderboards?

[KayJayQueue]

I believe "tracked" means having >3 games of data collected for them, and "not tracked" means you just happened to be in games with tracked players

Also worth noting this ELO rating is not a measure of skill/proficiency of any kind and is only a fancier way of tracking wins/losses

It measures my proficiency in losing!

I'm better at losing than you, get rekt noob.

Not as scum, come at me!!!

[Alianna]

I believe "tracked" means having >3 games of data collected for them, and "not tracked" means you just happened to be in games with tracked players

Also worth noting this ELO rating is not a measure of skill/proficiency of any kind and is only a fancier way of tracking wins/losses

It measures my proficiency in losing!

I'm better at losing than you, get rekt noob.

Not as scum, come at me!!!

I think my inability to roll scum makes up for any lack of proficiency in losing at it.

[Alianna]

In seriousness, I think Aureal has made some good points re: opt-in not really being opt-in.

[Gypyx]

Yeah i do agree that this isn't really opt-in, but it's also not meant to be opt-in really

if anyone has issues with that, i'd be more than happy to hide their ELO

[humaneatingmonkey]

if you want to hide your ELO, you'll need to pay gypyx

[Gypyx]

You come to ask me to hide your ELO, on the day of my daughter's wedding?

[KayJayQueue]

if you want to hide your ELO, you'll need to pay gypyx

I hope she takes PookyBucks

[ORAM]

Look how they massacred my elo

[DragonEater70]

yeah i tried to correct that but sadly scum still has more gains than town

3 town wins barely reaches the gains of 1 scum win

I'm pretty sure that's supposed to work that way, since people are roughly 3 times more likely to roll town. So if you play 12 games, roll scum 3 times and win twice, roll town 9 times and win three times, your average performance per alignment is 50% (2/3 for scum, 1/3 for town, averages out to 3/6=1/2). So you would be the same overall skill level as a player who won 6 games as town (out of 9) and only one as scum (out of 3). Isn't that logical?

[DragonEater70]

Also I very much agree that currently tracking ELO (and hiding it, or coloring it grey, or whatever) of players who have not opted in, makes this an opt-out system (or even a forced-participation system). So if you could have two sheets and have one that privately tracks ELO for everyone and one that publicly shows ELO for only those who actually want to be tracked, that could work. Another option which I think doesn't even make a big difference is to simply not calculate the ELO score of people who have not opted in at all. I mean, if I understand correctly, you calculate a team's ELO as the average of the ELOs of the individual players in it, right? So why not just... feed the algorithm only the scores of tracked players, and enter 1000 (I'm pretty sure that's the "base" value of your system) for every other player?

[Gypyx]

Also I very much agree that currently tracking ELO (and hiding it, or coloring it grey, or whatever) of players who have not opted in, makes this an opt-out system (or even a forced-participation system). So if you could have two sheets and have one that privately tracks ELO for everyone and one that publicly shows ELO for only those who actually want to be tracked, that could work. Another option which I think doesn't even make a big difference is to simply not calculate the ELO score of people who have not opted in at all. I mean, if I understand correctly, you calculate a team's ELO as the average of the ELOs of the individual players in it, right? So why not just... feed the algorithm only the scores of tracked players, and enter 1000 (I'm pretty sure that's the "base" value of your system) for every other player?

Well, in my philosophy, the ELO thing is barely a more complex thing than just tracking everyone's winrate (which folks on mafiauniverse do with absolutely no issues whatsoever as far as i'm aware) and like, i feel like the default stance is that you don't care?

It's similar to being meta'd on, obviously this facilitates things, but i feel like the average player doesn't mind, and there's not a lot that warrants making it such a private system

[humaneatingmonkey]

Approved Slot:

Empty!

Standard Slot:

DkKoba's srs gaming mountainous 12p (6 days left)
Note: This game has restrictions on replacing out. Only join if you are willing to commit for the full game.

1. humaneatingmonkey
   
2. Dannflor
   
3. Oxytocin
   
4. Tatiana
   
5. chidori
   
6. 
   
7. 
   
8. 
   
9. 
   
10. 
    
11. 
    

Mod queue: Empty!

it would be a SHAME if people won't join this game.