|
Abstract
It's long been debated whether mafia performance can be evaluated objectively. Faced with discussion of stats, some people say, "some people make more intangible contributions to their team," or, "win percentage doesn't really show how good you are." Those people probably lost a lot of games. I'm here to show you who was really good in 2016. Not those sucky "contributed a lot to the town, even if he put his vote in the wrong place" players, I mean the truly skilled ones, who deftly navigate the field of possible candidates and happen to place their vote on the right guy.
What other way to do this than to analyze who voted who, and whether they were right or wrong? Let's take a look at the data.
Methodology
I collected data from (almost) all mafia games in 2016, beginning with Unoriginal Name Mini Mafia and going all the way to Liquidmania Qualifier #1. We'll get to the exceptions later.
To calculate, objectively and indisputablely, skill as town, I collected the number of times a player voted scum as town and divided it by the number of days the player was alive as town. Then I multiplied it by 100 to make it a percent.
To calculate skill as scum, I collected the number of times a player was voted as scum and divided it by the number of days the player was alive as scum. Then I multiplied it by 1 to keep it the same.
There were some days/votes that were not counted. Here is a rundown of them:
- Days or votes on days when there was no traditional lynch. This includes no-lynches, days that never ended or ended before a lynch could occur, non-traditional lynches (e.g., mayor votes), and days with no lynch mechanic (e.g., most days in Palmar's Purge).
- Votes from factions (e.g., Benevolent Ghosts in Haunted Mansion)
- Anything in Unknown Mafia because the lynch documentation was very hard to find
If a townie did not vote on one day, the day was still counted; scum-on-scum votes were counted against the scum who received them.
To prevent extreme cases with very small sample sizes obscuring the important data, I set minimum eligibility requirements to be on the list. To be considered for the town list, a player had to vote scum more than 5 times and be alive as town for more than 10 days. To be considered for the scum list, a player had to be alive as scum for more than 5 days.
Results, etc
Below are the official, indisputable, objective, and otherwise perfect TL Mafia Skill Rankings:
Some thoughts:
(If you noticed, scum were voted much more than town voted scum. This is because of bussing. It appears about one-fourth of all votes scum receive come from their teammates.)
The most striking thing about the list is that several of the top players are not exactly feared by scum; and furthermore, there are many players (Palmar, marv come to mind) who typically die on D1 and yet did not even vote scum enough to be eligible. Part of the reason for this is that, because they tend to die so early, they tend to vote less per game and with less information than most players.
There were several players, some who are barely ineligible for this very reason, who have a very high success rate as town with a sample size of nine or so. Tubesock went 8-for-10, yet perplexingly most of his votes came on the day he was mislynched. The explanation seems to be that many playersplayed in a game where town dominated and got four or five scum in five days. Disformation, darthfoley, and LightningStrike all played in Star Wars: The Mafia Awaken and nailed scum every day because of how AFK they were. For this reason, perhaps a more telling metric would be correct votes in comparison to the rest of the town on that day., which sadly would require more effort.
Few players topped the scum list who had many scum games under their belt. The tendency of scum was to either play one or two games and lie low, like jat, or to play several but in one day accumulate several votes, like Shapelog. Sadly these rankings will be skewed in favor of sparser players with a couple lucky games until there are many with a large sample size, or unless we count days with many votes less harshly. Vivax had the most votes of all, because of his tendency to play in large games and get no votes until a single day of eight or more. Or maybe he's just bad at mafia, idk.
I would love a continued evaluation through 2017 and the very earliest and later parts of 2016, to see the numbers even out and converge to the averages, to find trends we may have missed, to see differences between years, and to evaluate whatever knew players arise. I don't exactly jump at the thought, seeing as this is rather monotonous, sometimes annoying work, but if someone else would like to help continue this to 2017 it would be greatly appreciated.
Find your own statistics and trends, for bragging rights or something, here.
|
I will gladly accept your nomination as the best town player on TL Mafia. Thank you Tumblewood, finally someone understands my jenius.
|
This is super cool, though. As someone who loves stats, this makes my heart soar.
|
Getting cop checked as scum seems to put a serious dent on your statistics Also CCing a blue and winning the initial claim wars seems to not help for some reason :/
|
Meanwhile having a real good/lucky game really boosts you. I am surprised at how good my rating for town is, but am 1 day short to make rankings.
def. interesting though.
edit: nm needs to play more so he can have 100% on 10 days.
|
United Kingdom30774 Posts
On January 21 2017 18:32 Artanis[Xp] wrote:Getting cop checked as scum seems to put a serious dent on your statistics Also CCing a blue and winning the initial claim wars seems to not help for some reason :/
I can tell you that's a lie.
|
On January 21 2017 20:52 Holyflare wrote:Show nested quote +On January 21 2017 18:32 Artanis[Xp] wrote:Getting cop checked as scum seems to put a serious dent on your statistics Also CCing a blue and winning the initial claim wars seems to not help for some reason :/ I can tell you that's a lie. Was a semiopen setup with a RBer and GF but no framers or millers so yeah
|
On January 21 2017 20:52 Holyflare wrote:Show nested quote +On January 21 2017 18:32 Artanis[Xp] wrote:Getting cop checked as scum seems to put a serious dent on your statistics Also CCing a blue and winning the initial claim wars seems to not help for some reason :/ I can tell you that's a lie.
Fuck you
|
Yes, I did certainly "lie low " as mafia last year. Not like I had to put in most of the work in both games
|
|
What is it you don't like about my method?
It's the name, ain't it? Too highfalutin, Bragons.
|
Btw. If you aim to improve your ratings:
Obviously for both alignments this doesn't deliver a perfect skill representation. But it is especially skewed in case of the townies. While as a single player you can win a game as scum by not being lynched you cannot win by only voting mafia as town. You identified the problem of regularly being shot early in the game already which favors people who mafia likes to have in the game. Other problems are that you need to convince people (which is most likely impossible to include in a ranking as simple as this one) and more importantly not being mislynched (which could be included by adding a "voted as town" category - you only have to evaluate its importance compared to the scumvoting if you want a single "skill value" in the end). You do not help your team by voting mafia off wagon or by getting votes yourself.
|
On January 22 2017 01:54 justanothertownie wrote: Btw. If you aim to improve your ratings:
Obviously for both alignments this doesn't deliver a perfect skill representation. But it is especially skewed in case of the townies. While as a single player you can win a game as scum by not being lynched you cannot win by only voting mafia as town. You identified the problem of regularly being shot early in the game already which favors people who mafia likes to have in the game. Other problems are that you need to convince people (which is most likely impossible to include in a ranking as simple as this one) and more importantly not being mislynched (which could be included by adding a "voted as town" category - you only have to evaluate its importance compared to the scumvoting if you want a single "skill value" in the end). You do not help your team by voting mafia off wagon or by getting votes yourself. going into this project the aim was not to determine absolutely or even accurately who is the best, but to examine the data for interesting trends and bragging rights. the trouble is that a good town player should, naturally, create more mafia lynches, but, especially with such a small sample size, is in distinguishing between good town play and poor mafia play. adding a votes received category for town would be sensible if I wanted to go back through all the old games, but even that is susceptible from interference of people just being afraid of someone's scum game. if I really wanted to evaluate players' true skill, I would just make power rankings the old-fashioned way.
|
On January 21 2017 22:16 justanothertownie wrote:Yes, I did certainly "lie low " as mafia last year. Not like I had to put in most of the work in both games would you prefer "flew under the radar"? I didn't mean to imply getting carried or anything
|
On January 22 2017 02:53 Tumblewood wrote:Show nested quote +On January 21 2017 22:16 justanothertownie wrote:Yes, I did certainly "lie low " as mafia last year. Not like I had to put in most of the work in both games would you prefer "flew under the radar"? I didn't mean to imply getting carried or anything Don't worry. I didn't think that you implied anything like that.
On January 22 2017 02:51 Tumblewood wrote:Show nested quote +On January 22 2017 01:54 justanothertownie wrote: Btw. If you aim to improve your ratings:
Obviously for both alignments this doesn't deliver a perfect skill representation. But it is especially skewed in case of the townies. While as a single player you can win a game as scum by not being lynched you cannot win by only voting mafia as town. You identified the problem of regularly being shot early in the game already which favors people who mafia likes to have in the game. Other problems are that you need to convince people (which is most likely impossible to include in a ranking as simple as this one) and more importantly not being mislynched (which could be included by adding a "voted as town" category - you only have to evaluate its importance compared to the scumvoting if you want a single "skill value" in the end). You do not help your team by voting mafia off wagon or by getting votes yourself. going into this project the aim was not to determine absolutely or even accurately who is the best, but to examine the data for interesting trends and bragging rights. the trouble is that a good town player should, naturally, create more mafia lynches, but, especially with such a small sample size, is in distinguishing between good town play and poor mafia play. adding a votes received category for town would be sensible if I wanted to go back through all the old games, but even that is susceptible from interference of people just being afraid of someone's scum game. if I really wanted to evaluate players' true skill, I would just make power rankings the old-fashioned way. I get that. I just wanted to mention that you could make this a little more meaningful with the "voted as town" stuff even if you just use the data of 2016. Of course everything would be more telling if you included even older data but I can see how this is a lot of work as is.
|
Mafia KP used on you as town should also be a good indicator, especially on N1/N2.
|
On January 22 2017 04:55 Artanis[Xp] wrote: Mafia KP used on you as town should also be a good indicator, especially on N1/N2. Yep. The problem with this are blue claims or at least mafia bluehunting/medic dodging but in general it should be. It would expand the method though since it isn't tied to voting.
|
United Kingdom30774 Posts
I'd argue that it takes more skill to be voted as mafia but survive longer too. It should be something like days survived/total game days x100 and ignore the votes since that's not a good metric at all imo. Surviving is the true mafia skill.
|
On January 22 2017 07:40 Holyflare wrote: I'd argue that it takes more skill to be voted as mafia but survive longer too. It should be something like days survived/total game days x100 and ignore the votes since that's not a good metric at all imo. Surviving is the true mafia skill. Yes, I agree. The hardest part of being scum is turning the lynch around once influential people really want to murder you.
|
It is late so I didn't read the thread but I have a question.
Are only the votes at the end of the day counted?
I'll read the thread tomorrow.
|
|
|
|