A Theoretical way to improve Casual and implement Ranked

Feedback/Suggestions
#21

ELO is fundamentally flawed in team games and should never be used, but developers continue to misuse it because it is recognizable and “easy”. ELO only works correctly in 1v1 games.

Somebody’s ELO in a game like ToS or ToL would mean literally nothing. There is too much random chance and unbalanced distribution of power between the classes to provide an equal amount of influence for each game.

Newark suggests a simple +1 or -1 rating system which is only correct under the Law of Large Numbers theory. In reality, reaching your true ELO with this simple method would require an enormous amount of games. Likely thousands.

1 Like
#22

winrate

still has huge problems but less effort

#23

Zero effort but zero meaning

#24

how about you just display number of wins and number of losses
the ratio between wins and losses could be used to measure general skill, but at the same time, even with a ~50/50 ratio, someone whos played a lot more games is generally better

#25

There is still a flaw to this way of ranking, as you may carry your team while playing. The point of ranked is to match you with people of your own skill, and not be carried or carry most of the time.

#26

And yet, there is no other ranking system that works in a game like Throne of Lies, unless there is one I don’t know of.

#27

There really isn’t – at least, not that I know of. I know that Town of Salem’s ranking system takes into account which role you play as. Neutral killing roles receive far more points for a win and lose less points for a loss. When you are Town, you receive a small amount of points for a win and lose slightly more points than a NK.

#28

Ideally, our theoretical ELO system would factor in;

  1. Which class you started as
  2. How long you lived
  3. If you were converted or not
  4. The ELOs of your fellow allies and enemies

Pretty complex system.

#29

but when people see the 2 numbers, they can take that sort of thing into effect

#30

No one said this would be easy, but I agree with you, though it will be a lot of roles and you will need to re-evaluate everything which is very difficult.

#31

And? You simply can get boosted or xmcheat your way up, or even play a lot of games. It’s not how people think about your skill level, it’s about your actual skill level that matters.

#32

Before any more comments arise, I thought this might be an interesting thing to talk about that can potentially be tweaked to fit Throne of Lies:

Overwatch: 1-100 rank. 50 is average. Tries to mimic a percentile. Pros Very easy to see who is better than other players at a certain time. Straightforward. Actually is a bell-curve as opposed to LoL. Cons It does not accurately represent a percentile. No one is rank 100. Rank 70+ is the top 1% of players, but it looks like top 30% erroneously. This could be tweaked in the future though.

Any ideas how we can try to implement and tweak this for Throne of Lies?

#33

You misspelt “Hippolytus” as “No one”. May wanna correct that pal.

#34

It’s from Reddit :stuck_out_tongue_winking_eye:

#35

winrate * (games played/2)?

#36

Nailed it :joy:

1 Like
#37

still better than Elo in Mafia lmao

#38

and by playing more games, people will get better
and the people who would hack those value arent the kind of people that would put some losses there too. they would just give themselves a ton of wins, making them easy to tell apart from non-hackers, or they can tell from their skill level when playing, and report them if neccesary

#39

Really think you have missed my point .

I play 160 games , I play 90 games as bd 10 games as king , 20 games start as evil 40 games neutral

you play 160 games , you will play 90 games as bd 10 games as king 20 games start as evil 40 games neutral

I win 90 of those games , you win 70 of those games . I am higher ranked than you .

Because players are randomly assigned each game ,win ratio is the only important factor, as EVERYONE over a long enough window will have the same distribution of starting roles .

Sure it maybe harder to win as bd than as cult , but that’s ok because everyone gets an equal amount of time as each .

Also not to muddy the waters but the devs have said that about 50 -50 split between good and evil wins .

1 Like
#40

newarkinvaders is 100% right.