poll elo system

[skibbel]

"Random teams, 15/15 all games"

Is not close an "elo" system.

And the resulting "score" doesn't not reflect anyones skill level, cause the score of 2 players of equal skill level will differ by the number of games played only.

E.g. 2 players win 66% of the games.

Player A has played 20 games. He will have ~1100 score.
Player B has played 100 games. He will have ~1500 score.

Imagine that at a larger scale...

[dweiler]

Any change made will indeed be for next season.

About 15/15. When I proposed it I never said it's an 'Elo' system, the reason it 'looks' like Elo is just because the bots are arranged like this (and for example for DC penalties). It's more every win is 1 point and every loss 1 point off.

From what I hear online people complain most about inactivity. People just want to play games. A system where you get 15 for each win makes keep playing more attractive. So the question is what you want, an active league where the winner is someone who is good AND active, or an Elo system where people often dodge, are discouraged playing for low Elo, and someone more inactive but perhaps more skilled wins.

[FadingSuns]

Any change made will indeed be for next season.

About 15/15. When I proposed it I never said it's an 'Elo' system, the reason it 'looks' like Elo is just because the bots are arranged like this (and for example for DC penalties). It's more every win is 1 point and every loss 1 point off.

From what I hear online people complain most about inactivity. People just want to play games. A system where you get 15 for each win makes keep playing more attractive. So the question is what you want, an active league where the winner is someone who is good AND active, or an Elo system where people often dodge, are discouraged playing for low Elo, and someone more inactive but perhaps more skilled wins.

I definitely choose an active league. Really we all already know who are the best and the worst players here, independent of elo. There is nothing to be proved on that matter. I see really more important to keep this league alive somehow. Last season it was much much easyer to play a game than this one. And this is basically the only reason of why i chose the option 15/15 random teams, because i know that league will be more active.

[SLSGuennter]

Any change made will indeed be for next season.

About 15/15. When I proposed it I never said it's an 'Elo' system, the reason it 'looks' like Elo is just because the bots are arranged like this (and for example for DC penalties). It's more every win is 1 point and every loss 1 point off.

From what I hear online people complain most about inactivity. People just want to play games. A system where you get 15 for each win makes keep playing more attractive. So the question is what you want, an active league where the winner is someone who is good AND active, or an Elo system where people often dodge, are discouraged playing for low Elo, and someone more inactive but perhaps more skilled wins.

I definitely choose an active league. Really we all already know who are the best and the worst players here, independent of elo. There is nothing to be proved on that matter. I see really more important to keep this league alive somehow. Last season it was much much easyer to play a game than this one. And this is basically the only reason of why i chose the option 15/15 random teams, because i know that league will be more active.

You arent here to long, so you cant know ... in the summer the league activity always drops heavily, and will rise again next autumn, no matter what score-system is used.

[MarshMallows]

Since I've been in LIHL it has never dropped this low.

[TinSoldier]

well the problem with the current system is that it can scale to 22/8, which is pointless because at that point the payoff for for DC penalty is better than actually losing lol. That is the main difference I see between option 1 and 3.

[SLSGuennter]

well the problem with the current system is that it can scale to 22/8, which is pointless because at that point the payoff for for DC penalty is better than actually losing lol. That is the main difference I see between option 1 and 3.

It does not scale like this, it just SHOWS it does. If you calculate it after the game ends, you lost/won an other amount of Elo then shown ingame.

[TinSoldier]

no if you look at the elo scales diablo posted in the announcement thread it is possible to have a 22/8 game with 700 elo difference. its unlikely of course that a game is that ill balanced but its still a possibility @guennter

[SLSGuennter]

Well, yeah ... but with "unlikely" you mean: It never happens, literaly. Does it?
If the 4 highest would play together vs the 4 lowest atm, that would result in just 500 Elodiff (i did calc ca. 509) ... So how would 700 happen? @iistyles

1.) and 3.) are just the same (except for 2s) ... so its like 12 votes for the same

[TinSoldier]

Its still a flaw in the system guennter maybe its not possible now but it can still happen. The point is there should be no way for a game elo to be worth more than the DC penalty. Either increase the DC pen or change the scale

[Nore]

So, any update on this now? The poll has been up for long enough imo.

[TinSoldier]

just hang on mick is in a 1200 game atm he'll get back to you when hes done @nore

[dweiler]

Like I said, anything that happens will be for next season, which is still almost a month away.

As I see it now, this poll practically ties between Random teams 15/15 and Random with Elo (13 votes vs 12 votes- with Elo has 2 options with small differences), so with such a small margin I am rather inclined to leave it this way, because otherwise we can keep changing every time someone changes their mind again.

[Diablo_]

Seems like people have calmed down about the Elo system update and seeing it's basically a tie anyway there's no need to change it again for now.