1ST ANNUAL TURBO MAFIA CHAMPIONSHIP
View Post #2 for updated Individual and Community Leaderboards.
[box=WHAT IS TURBO MAFIA?]A
turbo mafia game is a game of mafia that usually lasts between 45 to 90 minutes. Each game takes place in one game thread, and the games are automated, so everything will be not only fast-paced, but also very smooth. Day phases last 18 minutes and night phases last 6 minutes, so it's a game of speed reading, quick reactions and trusting your gut feelings, as opposed to regular length games where you would have time to do thorough re-reads and in-depth analysis. The fast pace makes these turbos well suited for people who don't have the time to commit to longer games, and people who are just looking for a quick fix!
We've run more than 800 turbos on Mafia Universe in the past couple of years, and while they can take some getting used to, we hope you'll give it a chance and join the fun!
[/box]
[box=WHO CAN PARTICIPATE?]
Everyone is eligible to participate. No matter what community you're from or whether or not you've ever played a turbo mafia game before, you are very welcome to play in our Championship games!
Find sign-ups by going to the
Turbo subforum or by taking a look at the front page:
We hope and expect to have multiple turbo games running each day. If you don't see an active sign-up, feel free to start one yourself or do an interest check on our
Discord server.
If you want to play in these turbos for the sake of fun, but you'd rather not appear on the leaderboards, tag @mhsmith0 in this thread and he'll make sure you don't show anywhere in the standings.[/box]
[box=HOW DOES THE CHAMPIONSHIP WORK?]By playing in Championship turbo games over the course of the next month, you rack up leaderboard points. The Championship season will end at
11:59 PM EST on March 19th. At that time, we will crown two individual winners: 1) the best player with 10-19 games played and 2) the best player with 20+ games played. You will need to play a minimum of 10 games to be eligible for any win.
If you have multiple accounts and wish to alternate between which one you play in Championship turbo games with, that is fine, but their records will be combined for leaderboard purposes. In the case of using an unouted alternate account as well as a primary account/outed alternate account, we will combine only at the end of the season. (Note that this means if you play both on an unouted alt and a primary account/outed alt, the unouted alt will be outed at the end of the season; however, if you only play on a single account that is an unouted alt, it will not be outed.)
In addition, you have the
option to represent a community. At the end of the season, we won't just crown individual winners; we'll also crown the best-performing community. By opting to represent a community, you'll be making an impact on the separate community leaderboard. Our hope is that this will add an extra layer of good-natured and fun competitiveness.
The points system:
The points system works on a few fundamental principles:
- How well did you do
- How much of your team's success (or failure) did you play a part of
- How difficult was the challenge you faced
[SPOILER=More details on the scoring system]
Principle #1: How well did you do
The first is easy: wins and losses. The more you win, the better you do, and the more you lose, the worse you do.
In addition, results are scored on average performance, so someone who plays 20 games and goes, say, 14-6, will be no better or no worse off than someone who plays 30 games and goes 21-9, 40 games and 28-12, and so on and so forth.
Principle #2: How much of your team's success (or failure) did you play a part of
The second ties into how we weigh results together. In particular, when you are a member of a small faction (i.e. scum), you are, in general, a relatively large part of your team's success or failure, while when you are a member of a large faction (i.e. town), the opposite is true. Similarly, in larger games, you will likely have less of an impact on your faction's result than you would in a smaller game.
What we've done with this is assign "Game Shares". All members of a faction will receive an equal portion of that faction's outcome. To take an easy example, in the Vig 10'er setup (8 town, 2 scum), each town member will receive 1/8 of a win or 1/8 of a loss, and each scum member will receive 1/2 of a win or 1/2 of a loss, depending on the outcome.
In practical terms, this means that every game result is zero sum: the amount that one faction gains will be equal to the amount that the other faction loses. If a group of people play two turbos in a night and scum wins both, then anyone who went 1-1 (one scum win, one town loss) would have had a good night, which balances out the fact that a bunch of people will have gone 0-2 and had a bad night. And vice-versa if you get a pair of town wins instead.
As an example of how this calculation works, take a look at the google doc:
https://docs.google.com/spreadsheets...it?usp=sharing
Here I've taken the last 12 turbos that have been in championship-eligible format, processed them through the model, and come up with outputs based on principles 1 and 2 (we'll get to #3 later).
All the way at the right, in tab "Player Statistics-Summary", you can see that I've ordered players by their calculated rankings. All players with perfect records (for a sample size this small, there are a bunch) are at the top, and the remainder generally correlates to winning percentage. Exceptions come from divergent town and scum records; since scum games get greater weight than town games, average performance can differ from overall winning percentage when you do particularly well or poorly as scum.
Actual leaderboards will split out by those fully eligible (20+ games), those partially eligible (10-19 games) and everyone else. However, in this case no one had 10+ games in the given sample, so instead you see everyone tossed together in one big table.
Principle #3: How difficult was the challenge you faced
This is where the regression model (see turbo ratings at
https://www.mafiauniverse.com/forums...o-Ratings-Data ) comes into play.
In the turbo championships, there are four different formats of games (each of which has different levels of town vs scum power), and player lists will (or should) vary substantially in terms of not just overall player quality, but in terms of specific aptitude for particular alignments.
The turbo championships ranking system takes these differences into account, and will adjust the value of a result based on how easy or difficult it was. Rand town with a really strong town team, and a weak wolf team, in a format where town does unusually well? You're going to get relatively low credit for that win (but still more than zero). Pick up a really impressive win? You're going to get outsize credit for that win. And similar principles hold for your losses, just in reverse.
If you go back to the google doc link, those values were all calculated without making any difficulty adjustments; essentially, every result was treated as if it was against a 50% win or loss chance. Once the model has enough data to start generating ratings, you'll see scores start to adjust for the difficulty of particular results.
In the end, rankings will still correlate strongly to winning percentages, but they'll be adjusted to give greater weight to your performance in games where you're scum, and they'll be adjusted to give more credit (or less demerit) for results in relatively difficult games, and the inverse for easier games.[/SPOILER]
The community leaderboard points system:
If at least 4 people from any given community have played 10+ games (and thus are "ranked"), we will calculate the average score for that community's players, and that will be the community's score. In other words, if Players A (0.6), B (0.65), C (0.35) and D (0.40) are representing Community Gollywag, then Gollywag's community score would be 0.5.
We will use the "Community" profile tags and your posts here in this thread to help us discern who belongs to which community, but if you either wish to NOT represent a community or if you're representing a different community than is indicated by your profile, let us know and we'll make the necessary adjustments to the leaderboard.[/box]
[box=WHAT ARE THE CHAMPIONSHIP SETUPS?]
"Cop 9'er"
"Cop 13'er"
"Vigilante 10'er"
"Mountainous 11'er"
All Championship turbos must meet the following criteria:
- The turbo must be run and completed between the time of this post being made and 11:59 PM EST on March 12th.
- The turbo must be either a Cop 9'er, Cop 13'er, Vigilante 10'er or Mountainous 11'er (click spoilers above for more information).
- The turbo must have the community standard phase settings of 18/6 (18-minute days, 6-minute nights).
- The turbo must also enable the following settings: Majority ON from Day 2 onwards, No Lynch ON, Votes locked at LYLO, Tied votes are resolved via randomization.
- No anonymous turbos.
Note that a turbo doesn't necessarily need to be tagged as a "Championship turbo" to count for the standings. If all of the above criteria is met, then it's a sanctioned Championship turbo. If even just one of the criteria isn't met (such as days being shorter), then it doesn't count.[/box]