BETA: Arena mode (Season 0 - BETA)

[asd123321]

It was an obnoxious stunt to have it start with no Accept or Reject, a 4 player game I would not have done. That
cost me 19 ELO.

[Ankit91]

In 7 wonders, I lost 5 places on the ladder in gold league. How can I lose 5 places in a match? I thought, maximum one could gain/lose is 3.

[Waterd103]

Players below certain karma shouldn't be able to join arena games. Right now there is a player joining and quitting games 1 minute in, boosting players rank. He has 0 karma

[nmego]

I was skeptical of the system at first, but now I find it quite fun. it has a bunch of flaws and it could be improved to be much better. But I quite like it, thanks BGA!

[sprockitz]

Initial thoughts on EAS.

It looks like EAS is just akin to Elo, with a K factor about 20. I really think we need game specific K factors...a 5 minute game should have a lower K factor than an hour long game. The longer games need a bigger K factor...with a K factor of 20, top players will need at least 100 games to approach their true rating. For hour long games this is an unreasonable ask if you are truly looking to crown the best players at a game instead of just the most active good player. I’d propose adjusting the K factor based on the length of a typical game at that player count.

Something continuous would work as well, but if the desire is a few discrete steps this looks reasonable to me
Average play time in minutes: K factor
60+: K = 60
40-59: K = 40
25-39: K = 30
15-24: K = 20
<15: K = 15

Longer games are also typically more skill based so a bigger K factor will help sort people faster and getting them playing games against similarly skilled opponents faster

Right now the levels seem kinda meaningless if you need to crank out tons of games to get your EAS to a competitive level. More meaningful games (against those ranked near you) should be the goal.

Another idea is to rank players by peak EAS achieved during the season. This keeps people playing without risking their rating, but if they lose a few they would fall well off their peak EAS and need many strong games to make it up.

[Waterd103]

I think it's insane that as a gold player i lose 5 points vs a bronze, but only win 1 vs them, wth. Bronze players aren't that bad. This will lead for me to not accepting lower level matches or playing only when i there is high traffic to be able to match gold or higher players.

It means i need to win 85% of hte games vs bronze players, just to keep even, lol.

[bluemutant]

Wouldn't a decaying ELO achieve many of the stated objectives of the Arena feature? The decay would serve as motivation to play more. You could display both a decayed "arena" ELO and a normal, non- decaying version. The current system has decay but only on a three month time frame.

It seems weirdly hard to move up a level (and will frequently be more a matter of getting lucky 5 games in a row rather than true skill) but then immediately have a floor slid under you.

It does makes sense to eliminate some choice of opponents but that could be implemented just as easily with a decaying ELO system

[spedia2116]

I do not really think that climbing up the leagues will be a matter of volume. Once the leagues will have settled in a few weeks and if the matchmaking will work, everyone will notice differences in skill level in gold and bronze league. Achieving 10 points in gold league will be quite tough and definitely a matter of skill.

In theory and in some games you are right. However in games where luck is involved I have got a lot of doubts. For example in games like 7 wonders (quick 4 player game) or even more like Saboteur. No relegation means that you have to win only 3-4 games to advance to higher level. It's not so hard in games where luck is one of the games element. Bad players playing a lot together will finally advance, lowering the higher league overall level - because they can't lose their position. More bad players in silver league MM together leads to more bad players in gold league and so on.
If you roll the dice and you get 1 or 2 4 times in a row you will advance to higher level. If you can't be relegated it's a matter of time when you achieve that. Not only of course, but it's kind of simplification.

Another thing is that the system allows you to play with people from other leagues. I know that's was designed to speed up matchmaker when there is lack of people in your league. However I don't think that it is the problem. If I can't play arena games because there are no opponent from my high league I can play ELO game in meantime. The point of being in elite league is to play with others from the same league. That's basically the whole point of dividing into leagues. You have to climb to play with better players. And better players have to climb to not play with bad ones. So playing in gold league I can play for example with people from silver league and from brown - because I will be playing at time where there is lack of players from gold league. And I can get better results from someone who is playing with players all the time from his league.
I can give you example. I played in one game at the evening and I lost 8 times in a row. Then I played same game at the morning and I won 3-4 times in a row and I advanced to higher league. And now I can lose every game and it doesn't matter that I'm bad in that game. I will be better from others who joined later when the competition was higher. I advanced because of luck and because I started to play early and nothing now can change it.

[Waterd103]

No relegation means that you have to win only 3-4 games to advance to higher level.

Have in mind that once you are in gold or higher, playign against lowre levels only nets you 1 dot, Im right now in gold in some games and i have to get 1 dot at a time (and dont get me on ever losing those games, you lose 5 dots.)

[Waterd103]

I hope its planned the feature to be able to filter to obsserve only arena games of certain rankings, for example i would like to watch replays or active games of Elite players in a specific game.