Mafia player rankings

[OceanWind]

Try me. Game list is on my wiki. I have no completed towngames on this site but I have three scumgames.

[Aronis]

Are you ranking just scum - or overall player performance?

I'm ranking everything, but the only normal/newbie/open hiplop has in the past 2 yrs is a single scum game and I haven't gotten around to adding theme games yet.

And updated it with Plotinus' stats

[Taly]

Are you ranking just scum - or overall player performance?

I'm ranking everything, but the only normal/newbie/open hiplop has in the past 2 yrs is a single scum game and I haven't gotten around to adding theme games yet.

And updated it with Plotinus' stats

Alright. I'll be patient - take your time, just notify me when you've gotten my stats. Those 6 games in my wiki are the ones where I haven't replaced in/out of - and proud of. LOL (But that's mostly all my games ever so far.)

[Taly]

ooh, my stats.

scumhunting is at the highest so far I don't lynch town too often <3

My OVR is low because I'm scumread often as town, and I die. LOL

At least my accuracy seems higher than survivability? ;P

[Aronis]

Added Taly's and OceanWind's(easy trumps post order).

Gonna do ika's and BROseidon's later tonight/tomorrow. And then I'll start adding the theme games to the stats tomorrow.

[Taly]

I'm curious though, what formula did you use to get the OVR with the other graded categories?

[Plotinus]

Thanks for doing my stats!

NKG is nightkilling. It's based on how good you are at shooting power roles at night. 50 is a good score, since shooting a power role every other night would be solid play. My overall leans more towards my survival stat since I counted that as being more important when averaging things.

I don't always kill people that I suspect are power roles; sometimes it's better to leave them alive and just roleblock them everynight and kill the people who are strong players who are right about their reads and have the charisma to get your teammates lynched. Sometimes killing the PR is the right play, sometimes it isn't.

[Aronis]

I'm curious though, what formula did you use to get the OVR with the other graded categories?

Okay, so I go through each game and take notes. So for town, I write down how long you survived and how many game phases there were. Then award you a score for TWN based on that. Then record your voting records and the lynches for the game, which I use to determine SHG and THG. And if applicable, who you used your PR on. Average those scores out to get your score for the game. Then average your scores for all the games to get your ovr.

The issue I ran into specifically with your ovr, was you got lynched D1 in one game. So you have an extremely small voting record and nothing to do with any other lynches and were VT, so no PR play. So your score for that game was zero. That game didnt effect your PR stats, and had hardly anything to do with everything except TWN. But your overall took a 10pt hit, because you had 4 game scores that were about 50 and then a score of 0.

I don't want to just calculate overall based just on your average in the 4 categories, bc you're only a power role maybe 25% of the time in normals or so and this current way accounts for that. But then I'm running into an issue where a single game can cause a huge dent to your overall that it prob shouldn't. Bc compare your town stats to mine, theyre fairly similar with mine being slightly better, yet the ovr is drastically different. So if I did do a no effort mafia type of thing, you wouldn't play near that bad.

[Taly]

Yeah, that's what I was thinking Aronis.

It seemed like the people who played more games for town - and had some more screw-ups of course - had lower in OVR.

We should find a way to work with that. o:

So, without the D1 lynch town game - would my OVR be like 44-49 now? XD

[Aronis]

Thanks for doing my stats!

NKG is nightkilling. It's based on how good you are at shooting power roles at night. 50 is a good score, since shooting a power role every other night would be solid play. My overall leans more towards my survival stat since I counted that as being more important when averaging things.

I don't always kill people that I suspect are power roles; sometimes it's better to leave them alive and just roleblock them everynight and kill the people who are strong players who are right about their reads and have the charisma to get your teammates lynched. Sometimes killing the PR is the right play, sometimes it isn't.

Yeah, I'm considering just ditching that category. I was trying to find extra categories for scum hence why I added it in the first place, but it just doesn't seem very crucial imo. And I could just go back to my original plans of having them just kill whoever they read as being the towniest, since that would prob be more accurate anyways.

Yeah, that's what I was thinking Aronis.

It seemed like the people who played more games for town - and had some more screw-ups of course - had lower in OVR.

We should find a way to work with that. o:

So, without the D1 lynch town game - would my OVR be like 44-49 now? XD

Take that game out of the equation and your ovr is 49.

[Ankamius]

/in

EDIT: I can get a list of all the theme games I've ever been in if it would help since I've been in quite a few by this point. Just let me know.

[Egg]

I'm curious how I stack up against some people

[wgeurts]

zMuffinman's winrate is probably horrible. Not because of anything he did, but because what happens after he's dead isn't under his control. The thing about using straight winrate is Mafia is a team game and a lot of the time weak towns get carried by the strong ones. Or town/scumsided setups. coughteammafiacough

Oi

[Hinduragi]

I think there are too many experimental uncertainties for this to ever become legitimate but...if I were to entertain your idea, I would take into account the players these people play with and calculate a rough "difficulty" category for that game in particular and how the odds are stacked for/against them based on the other players' alignment play. Obviously you can't do that until you have estimates for all of the players in the multiple games so this isn't really a suggestion but more of a general idea I'm throwing in. There are clearly more like "role and how it's used both in the daygame and nightgame" (i.e. cops not drawing a NK based on day play) and so forth. Food for thought.

[hiplop]

based on this one game (which one?) im fucking terrible

[Taly]

I feel a bit better about my performance now lol

Anyway Aronis; I suggest you point out the flaws to the system in the initial thread post - at least people will get an idea of the flaws of this ranking and people could work together to fix them while getting a better idea of their true ranking.

That's just my thoughts.

[Aronis]

based on this one game (which one?) im fucking terrible

Mini 1596. I haven't added themes yet, gonna do that later today. So I wouldn't worry about your stats based on a single game.

I feel a bit better about my performance now lol

Anyway Aronis; I suggest you point out the flaws to the system in the initial thread post - at least people will get an idea of the flaws of this ranking and people could work together to fix them while getting a better idea of their true ranking.

That's just my thoughts.

Yeah, I'll try to reorganize the main post later. I'm wanting to get a larger sample size of things first, so I can know exactly what all needs fixing and then go from there.

[Ythan]

My only experience with things like this is that people who care about them are infinitely less fun to play with.

[Taly]

VOTE: Ythan

[Untrod Tripod]

basically I want advanced statistics that back up what I've known all along which is that I'm the best mafia player in the world

[Alchemist21]

/in

[Aronis]

Okay. So update.

For scum:
I'm planning to remove the NKG stat since I don't rlly feel like it's very crucial. Then modify the way I score SVL and mainly use it for scum ovr, with PRP composing the other 5-15% of scum ovr.

For town I'm pretty happy with things. I am considering combining SHG and THG, since the way they're scored is very similar and one stat for both might work better.

And then I'm probably going revamp the OVR to hopefully make it more accurate.

I haven't rlly made much in the way of composing statistics today. I am thinking about expanding to score 90-95% of games, just leaving out the very weird mechanics and blatantly unbalanced ones. Then cutting down to only include games within the past year instead of two years to cut down on the number of games I need to go through. Bc I feel like that would work better.

[Taly]

FYI Aronis, I stumbled upon a Micro I forgot about before my hiatus. I won/survived as town in that game - it's in my wiki. ;P

[AlwaysInnocent]

Thanks for the input. As town not getting lynched isn't a super huge factor, but as scum I think that is crucial. If you never get lynched as scum, you'll win every game. And I realize this system has its flaws, but I feel like the individual player can have a much bigger impact on their rating then if it was just based on wins/losses, because I can lead town in lynching scum D1 then get NK'ed on N1 and then the town tanks and I get a loss despite playing rlly well.

And I'm not going to force ppl to participate in a rating system. So if you want to ignore it, that's fine.

I have not finished any games on this forum yet, so even if I wanted to participate, I could not (this does not mean that I am a total newbie, though). Do not get me wrong: I like that people think about this kind of stuff. I thought a lot about this, too. Especially when I played chess for a while, I became convinced that Elo is probably one of the best rating systems in existence. Of course, Mafia is different. It is indeed a team game, and games are not guaranteed to be balanced. It is thus difficult to measure an individual's performance by means of a rating system. I can see why people would approach it like you do. Even though I do not agree with the approach, I like that people think about it in the first place. Still, in my view, we cannot rely on anything else than win rate, if we want to have ratings and want those ratings to reflect a player's skill level. Otherwise we risk imposing our own biases about the game on the rating system.

[Fro99er]

basically I want advanced statistics that back up what I've known all along which is that I'm the best mafia player in the world

I have a PhD in stats. I can verify this.