EPTTL Mafia Database - Page 24
| Forum Index > TL Mafia |
|
Onegu
United States9699 Posts
| ||
|
marvellosity
United Kingdom36161 Posts
On January 12 2016 11:18 Onegu wrote: Need a catagory about who loses games. I will bring you all down as both alignments. People need to listen to onegu more. And lynch onegu less. maybe people would do that if you played the game more until then, enjoy the status quo | ||
|
Artanis[Xp]
Netherlands12971 Posts
On January 12 2016 08:14 GlowingBear wrote: I don't know, I think I've been lynched in 8 of my 9 games as scum, and I won 8 games lol Every single game that I won as mafia I carried, each time that I died we lost miserably  | ||
|
Alur
Denmark3900 Posts
On January 10 2016 03:13 kitaman27 wrote: Ace, BloodyC0bbler, bumatlarge, and Caller all failed to play a game in 2015, leaving only Chezinu and RebirthOfLeGenD on the list of players who have played game every year since 2015. TL Mafia dead game. Unrelated: How regularly are newbie games hosted? Considering trying out this mafia thing. | ||
|
GreYMisT
United States6736 Posts
On January 13 2016 08:10 Alur wrote: TL Mafia dead game. Unrelated: How regularly are newbie games hosted? Considering trying out this mafia thing. I believe we always try to have a newbie game running or in the works | ||
|
kitaman27
United States9245 Posts
On January 13 2016 08:10 Alur wrote: TL Mafia dead game. Unrelated: How regularly are newbie games hosted? Considering trying out this mafia thing. XD I'd expect signups for the next one to open in the next week or two. LoneMeow mentioned he was thinking about hosting the next one. | ||
|
Onegu
United States9699 Posts
Maybe that is why mafia numbers are so far down, Onegu played 28 games... | ||
|
Alakaslam
United States17336 Posts
| ||
|
iamperfection
United States9646 Posts
| ||
|
Acrofales
Spain18290 Posts
Just one thing: how is there a win % for people with 0 gave played? Better visualization for that last graph: scatterplot! Also, can you compute Pearson's R statistic (using all data, not the avgs)? Looks like there might be a faint negative correlation. If that's statistically significant, it would be weird. If it's not, it probably means there's just more luck involved than Marv would like you to believe. Also would be interesting to compare experience vs win% to other games. There's probably a dataset about poker winnings somewhere. But it'd be a lot of work. Might get a scientific publication out of it if done well, though. In any case, you're curating a pretty cool data set here. Great job! | ||
|
justanothertownie
16325 Posts
On January 15 2016 19:09 Acrofales wrote: Awesome Stat job! Just one thing: how is there a win % for people with 0 gave played? Better visualization for that last graph: scatterplot! Also, can you compute Pearson's R statistic (using all data, not the avgs)? Looks like there might be a faint negative correlation. If that's statistically significant, it would be weird. If it's not, it probably means there's just more luck involved than Marv would like you to believe. Also would be interesting to compare experience vs win% to other games. There's probably a dataset about poker winnings somewhere. But it'd be a lot of work. Might get a scientific publication out of it if done well, though. In any case, you're curating a pretty cool data set here. Great job! The win percentage you have in mafia is at best loosely correlated to your skill or experience. First of all because it depends very much on your teammates and if you die n1 you cannot really win the game for your team as town for example. And second of all of course also because of luck in certain situations. | ||
|
sicklucker
Canada16987 Posts
| ||
|
kitaman27
United States9245 Posts
On January 16 2016 06:11 sicklucker wrote: I know stats somewhat and im sure the variance in mafia games means your win percentage will mean very little in your lifetime unless you play like a game everymonth for your entire life. Stats dont mean much do you really think marv is an elite scum player that does not give up day one and wins 9/10 of his games? Hence the disclaimer in the OP: "Please note that these stats are just for entertainment purposes. Mafia is obviously a team game so its difficult to put much weight into individual statistics, especially with a small sample size"  On January 15 2016 22:38 justanothertownie wrote: The win percentage you have in mafia is at best loosely correlated to your skill or experience. First of all because it depends very much on your teammates and if you die n1 you cannot really win the game for your team as town for example. And second of all of course also because of luck in certain situations. I'd say that there is a much stronger correlation to skill, than experience, although measuring "skill" in any quantitative way would be difficult. I'd probably have to look at something like a relationship between win percentage from games 1-x vs the outcome of game x+1, which is a bit too complicated. There are definitely a lot of factors that go into determining the outcome though, so you're probably right that it would be difficult to come up with anything worthwhile. On January 15 2016 19:09 Acrofales wrote: Better visualization for that last graph: scatterplot! Also, can you compute Pearson's R statistic (using all data, not the avgs)? Looks like there might be a faint negative correlation. If that's statistically significant, it would be weird. If it's not, it probably means there's just more luck involved than Marv would like you to believe. Yeah scattersplot would be better, but I was being lazy with the 30 second graphing site XD The graph shows "experience", which is why a player with 0 games experience (their first game) could have a win percentage. The reason I'm using an avg instead of all data is that it gives a win percentage, rather than a binary result of "win" or "loss" for each sample point. Maybe I'll go a bit further the next time I'm really bored. On January 15 2016 19:09 Acrofales wrote: Might get a scientific publication out of it if done well, though. lolz | ||
|
LightningStrike
United States14277 Posts
| ||
|
sicklucker
Canada16987 Posts
| ||
|
Acrofales
Spain18290 Posts
On January 16 2016 06:58 kitaman27 wrote: Hence the disclaimer in the OP: "Please note that these stats are just for entertainment purposes. Mafia is obviously a team game so its difficult to put much weight into individual statistics, especially with a small sample size" I'd say that there is a much stronger correlation to skill, than experience, although measuring "skill" in any quantitative way would be difficult. I'd probably have to look at something like a relationship between win percentage from games 1-x vs the outcome of game x+1, which is a bit too complicated. There are definitely a lot of factors that go into determining the outcome though, so you're probably right that it would be difficult to come up with anything worthwhile. Yeah scattersplot would be better, but I was being lazy with the 30 second graphing site XD The graph shows "experience", which is why a player with 0 games experience (their first game) could have a win percentage. The reason I'm using an avg instead of all data is that it gives a win percentage, rather than a binary result of "win" or "loss" for each sample point. Maybe I'll go a bit further the next time I'm really bored. lolz Yeah, scientific publication would be a long shot, but people really like cleverly done stats. And some kind of comparative work between the influence of experience on performance in different games might be sufficiently hard to figure out that people haven't actually done this yet. Who knows what the Journal of Applied Statistics accepts (definitely baseball, I saw some). A comparison of Mafia and Dota could be cool: "the influence of experience on your win chances in multiplayer games". There, wrote a title for you. Rest is all yours  As for the yes/no: I didn't mean that average. You can average per person to give a win percentage -- although now that I think about it, a time series for some long-standing players in which you group together X games (or X period of time) and calculate the win percentage and see if there are trend lines: answering important questions like does Marv win more now than he did when he was a nublet? -- but tangent aside, averaging per person is fine, but you also averaged over all people who played 1, ..., N games, which seems unnecessary (in a scatterplot and for computing correlation it would be a sin). | ||
|
kitaman27
United States9245 Posts
On January 16 2016 10:34 Acrofales wrote: averaging per person is fine, but you also averaged over all people who played 1, ..., N games, which seems unnecessary (in a scatterplot and for computing correlation it would be a sin). Averaging per person? I'm not sure I follow. For a single person, you either win or lose your first game, you either win or lose your second game, etc. Even if you were to generate a scatter with 1000 players, its either going to be (X,0) or (X,1) for each data point, unless I'm misinterpreting what you're saying? | ||
|
Acrofales
Spain18290 Posts
On January 16 2016 10:59 kitaman27 wrote: Averaging per person? I'm not sure I follow. For a single person, you either win or lose your first game, you either win or lose your second game, etc. Even if you were to generate a scatter with 1000 players, its either going to be (X,0) or (X,1) for each data point, unless I'm misinterpreting what you're saying? Ah, I think I figured it out. We're talking about different things. I see what you're computing now. You are computing the likelihood of winning your Xth game, given the outcome of the X'th game of everybody else. This answers a subtly different question than I was expecting: what are your chances of winning, given a certain amount of experience? As opposed to what I was thinking, which is: are more experienced players more likely to win? | ||
|
raynpelikoneet
Finland43270 Posts
A comparison of Mafia and Dota could be cool: "the influence of experience on your win chances in multiplayer games". There, wrote a title for you. Rest is all yours I don't think this can be compared at all, since the other game (Dota), unless i am mistaken, has a matchamking system and the other game (mafia) doesn't, at least on TL. | ||
|
VayneAuthority
United States8983 Posts
| ||
| ||
