Total Posts:36|Showing Posts:1-30|Last Page
Jump to topic:

Mafia ELO

Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
11/4/2013 10:16:32 PM
Posted: 3 years ago
This has been a project for months in the making (namely because I'm a slacker), but I would like to thank all who have helped with it, with a special thanks to TUF and Drafterman.

I'm personally happy with the results, the accuracy and what it allows us to do when Mafia regrows to its former glory.
"Wanting Red Rhino Pill to have gender"
Noumena
Posts: 6,047
Add as Friend
Challenge to a Debate
Send a Message
11/4/2013 10:34:14 PM
Posted: 3 years ago
At 11/4/2013 10:27:58 PM, F-16_Fighting_Falcon wrote:
And the results are ... where?
: At 5/13/2014 7:05:20 PM, Crescendo wrote:
: The difference is that the gay movement is currently pushing their will on Churches, as shown in the link to gay marriage in Denmark. Meanwhile, the Inquisition ended several centuries ago.
TUF
Posts: 21,309
Add as Friend
Challenge to a Debate
Send a Message
11/5/2013 1:36:30 AM
Posted: 3 years ago
At 11/4/2013 10:34:14 PM, Noumena wrote:
At 11/4/2013 10:27:58 PM, F-16_Fighting_Falcon wrote:
And the results are ... where?
"I've got to go and grab a shirt" ~ Airmax1227
TUF
Posts: 21,309
Add as Friend
Challenge to a Debate
Send a Message
11/5/2013 2:07:56 AM
Posted: 3 years ago
I would feel bad taking any credit for this outside of the idea. Drafterman did all of the extremely hard work, and OreEle processed all of the data. I was just on the sidelines cheering them on. Thanks guys for helping with this. Very special thanks to drafterman. you da man.
"I've got to go and grab a shirt" ~ Airmax1227
TheAntidoter
Posts: 4,323
Add as Friend
Challenge to a Debate
Send a Message
11/5/2013 7:17:35 AM
Posted: 3 years ago
At 11/5/2013 3:01:33 AM, JonMilne wrote:
At 11/4/2013 10:27:58 PM, F-16_Fighting_Falcon wrote:
And the results are ... where?
Affinity: Fire
Class: Human
Abilities: ????

Nac.

WOAH, COLORED FONT!
drafterman
Posts: 18,870
Add as Friend
Challenge to a Debate
Send a Message
11/5/2013 8:24:25 AM
Posted: 3 years ago
LOL. Why would we want to taint the pristine glory of virginal data by letting it be processed by the unwashed mashes?

Seriously though, where are the results?
Yraelz
Posts: 4,056
Add as Friend
Challenge to a Debate
Send a Message
11/5/2013 3:10:02 PM
Posted: 3 years ago
At 11/5/2013 1:58:46 PM, Ore_Ele wrote:
The results are at home, I'll post them when I get home after work.

Lmao, K.
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
11/5/2013 9:15:44 PM
Posted: 3 years ago
Okay, let me first note that the Mafia ELO (or MELO for short) is much more streaky than the debates ELO. Much like in football, 3 wins in a row or 3 losses in a row can completely change someone's ranking. Both ELO and MELO have players starting at 2,000 and going up or down from there. Let me also note that this is only taking into account the 160 games in the mafia archives. I am fully aware that there are older games and I do have it set up so that if those older games ever get compiled, I can add them in. But for now, we just have the 160.

So, without further ado, here is our top 20 list

Bluesteel - 2872
TUF - 2804
Ore_Ele - 2684
Askbob - 2601
Blackvoid - 2512
royalpaladin - 2436
Leafrod - 2425
caveat - 2385
BullDiesel - 2381
Johnicle - 2357
Mestari - 2322
M.Torres - 2308
drafterman - 2303
IFLYHIGH - 2294
Spritle - 2280
Belle - 2278
Hardcore.Pwnography - 2273
Annhasle - 2266
Tvellalott - 2245
Viper-King - 2244

Please note that experimental games, beginner games, flash games, games that collapse due to mods stopping mid-game, etc were taken out.

One thing to show how streaky it is, Bluesteel started with winning his first 6 games (again, only looking at the archives) and started 17 - 1 (and had a ridiculous MELO). He ended up winning on 2 of his next 10 games (bringing his record down to 19 - 9), and then went 6 - 8 for his closing before he left for a final record of 25 - 17.

Since Mafia is a team game, the odds of winning or losing are based not just one how good a single player is, but how good the team is. So one's score goes up or down based on how your team's MELO compares to the other team's MELO.

Another thing to note is that MELO does consider experience. Someone that is 20 - 20 is rated as better than someone that is 2 - 2. This is one of the things that has helped TUF, who has done over 60 games. And with winning his last 3 games in a row, that turned into about a 285 MELO point jump, but that is a lot when the top level is only 872 points above ground.
"Wanting Red Rhino Pill to have gender"
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
11/5/2013 9:16:52 PM
Posted: 3 years ago
I should also note, that we did test the MELO predictions against the games. Using the first 10 games to let it get its barrings, it went 28 - 12 for the next 40 games (stopped wasting time counting after that).
"Wanting Red Rhino Pill to have gender"
Andromeda_Z
Posts: 4,151
Add as Friend
Challenge to a Debate
Send a Message
11/5/2013 9:24:03 PM
Posted: 3 years ago
Is there a huge list somewhere or something? I haven't played mafia in multiple forevers so I'm somewhat curious to see if I'm even on it lol.
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
11/5/2013 9:33:17 PM
Posted: 3 years ago
I have a list of 176 different players that have played in the games. Including 10 that started games and were replaced on the winning team (if you stop playing and get replaced on the winning team, you get no points)
"Wanting Red Rhino Pill to have gender"
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
11/5/2013 9:39:39 PM
Posted: 3 years ago
At 11/5/2013 9:24:03 PM, Andromeda_Z wrote:
Is there a huge list somewhere or something? I haven't played mafia in multiple forevers so I'm somewhat curious to see if I'm even on it lol.

Yes, you are on it. Since you started 15 - 10, but had a rough end run of 3 - 7 you are listed at just above even at 2032.5
"Wanting Red Rhino Pill to have gender"
Logic_on_rails
Posts: 2,445
Add as Friend
Challenge to a Debate
Send a Message
11/6/2013 2:56:05 AM
Posted: 3 years ago
I knew I was only half the player that Viper was...

It probably won't mean anything, but... what's my ELO?
"Tis not in mortals to command success
But we"ll do more, Sempronius, we"ll deserve it
TUF
Posts: 21,309
Add as Friend
Challenge to a Debate
Send a Message
11/6/2013 4:31:49 AM
Posted: 3 years ago
I guess this means I play too much mafia. On a more serious note; how should this new list be moderated? Shoukd we post it publicly to allow anyone to edit it? I just fear someone might tamper with it to increase their score. Is there a way to restrict access to the list but allow people to view it only?
"I've got to go and grab a shirt" ~ Airmax1227
Noumena
Posts: 6,047
Add as Friend
Challenge to a Debate
Send a Message
11/6/2013 4:43:50 AM
Posted: 3 years ago
Is there any way to separate out the team factor? It just seems off that HCP and Viper made the top 20 whereas some of the best players (Danielle, F-383, FT, and Logic) didn't.
: At 5/13/2014 7:05:20 PM, Crescendo wrote:
: The difference is that the gay movement is currently pushing their will on Churches, as shown in the link to gay marriage in Denmark. Meanwhile, the Inquisition ended several centuries ago.
TheAntidoter
Posts: 4,323
Add as Friend
Challenge to a Debate
Send a Message
11/6/2013 7:38:54 AM
Posted: 3 years ago
I'll get on the top 20, although it will kill me in the end.
Affinity: Fire
Class: Human
Abilities: ????

Nac.

WOAH, COLORED FONT!
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
11/10/2013 1:19:28 AM
Posted: 3 years ago
At 11/6/2013 4:43:50 AM, Noumena wrote:
Is there any way to separate out the team factor? It just seems off that HCP and Viper made the top 20 whereas some of the best players (Danielle, F-383, FT, and Logic) didn't.

Part of the reason is because no algorithm can be made to fully compensate for all the complexities. We have methods for going forward, but nothing that can go back. Danielle is the perfect example. In the end (or run Mafia days) she was 100% game planned against, NP1 SOP was target Danielle. Cop, tracker, watcher, doc, role blocker, they were all on her. So if she was mafia, she was found right away. If she was town, town was wasting their roles on her and mafia was free to do as they will. This stuck her with a bad losing streak, and, as I said, the MELO is streaky. So because of that, her MELO suffered unjustly.

Though, for going forward, we have mod adjustment factors (which we can't apply to all the past games, only games going forward) so that players that do well are not hurt as hard for a loss and are benefited more in a win. It recognizes that although mafia is a team game, individual players contribute greatly and can make the difference between a win and a loss.
"Wanting Red Rhino Pill to have gender"
TUF
Posts: 21,309
Add as Friend
Challenge to a Debate
Send a Message
11/10/2013 2:37:19 AM
Posted: 3 years ago
When is the official list coming? Also how should this new list be moderated? Should we post it publicly to allow anyone to edit it? I just fear someone might tamper with it to increase their score. Is there a way to restrict access to the list but allow people to view it only?
"I've got to go and grab a shirt" ~ Airmax1227
FourTrouble
Posts: 12,759
Add as Friend
Challenge to a Debate
Send a Message
11/11/2013 12:04:42 AM
Posted: 3 years ago
Hmm, definitely a flawed way to rank mafia players, and I say that mostly because F-16 is missing, not because myself or Yrealz or Danielle is missing.

I don't know how the system works but I think, in the case of measuring of town, it is must less accurate. It would be interesting to see an ELO list for only mafia, as that would be slightly more accurate, although still far from perfect.

I imagine, in the case of F-16, maybe the fact he replaces into so many games skews his results, because he really should be towards the top of the list, given the number of games he plays in, and his strong play and influence as both town and mafia. I dunno how replacements are being calculated, but that may be something to look into...
FourTrouble
Posts: 12,759
Add as Friend
Challenge to a Debate
Send a Message
11/11/2013 12:06:49 AM
Posted: 3 years ago
I'd be curious to see the ELO formula that Riot implements in League of Legends applied here. I know it's quite advanced, and fine-tuned for team games, so it might make the outcome slightly more accurate here.
Yraelz
Posts: 4,056
Add as Friend
Challenge to a Debate
Send a Message
11/12/2013 1:21:48 PM
Posted: 3 years ago
At 11/11/2013 12:06:49 AM, FourTrouble wrote:
I'd be curious to see the ELO formula that Riot implements in League of Legends applied here. I know it's quite advanced, and fine-tuned for team games, so it might make the outcome slightly more accurate here.

I'll do that. Though the oddities of lopsided teams means it's not directly applicable.