🌿ELO System Updates

[TricksyFox]

Just trying to find ways to honor their initial objective of preventing cheating through power-levelling vs. puppet-opponents - just without the implemented side-effects of draining honest players and avoidance of making new friends.

This is very reasonable, but I think you're running into people who are already frustrated because the pages and pages of well intentioned and thoughtful alternative proposals early on in this thread have gone entirely ignored by BGA for weeks.

[GeoPaladin]

Just trying to find ways to honor their initial objective of preventing cheating through power-levelling vs. puppet-opponents - just without the implemented side-effects of draining honest players and avoidance of making new friends.

This is very reasonable, but I think you're running into people who are already frustrated because the pages and pages of well intentioned and thoughtful alternative proposals early on in this thread have gone entirely ignored by BGA for weeks.

It's well-intentioned on their part and I hope I'm not coming across as hostile in my responses.

However, I would not find such a compromise satisfying as it still interferes with the progression system. I'm a competitive player and I enjoy playing Arena mode. If I am randomly paired with a player where I cannot gain or lose any meaningful amount of elo, I might as well have played a friendly game with a random player - which is the opposite experience of what I'm looking for when I ask to be paired in an Arena game. I would prefer not to do this.

If the system must stay, it should be tuned so that people who play a wide range of opponents aren't being penalized. That would address my concerns while still keeping the system intact.

There's my $0.02 anyway.

[Zozoken]

Instead of reducing K-Factor after a streak rather than giving a penalty - which has been suggested many times as a fix - it might be even more elegant and simpler to implement to reduce the K-factor already when playing someone many times (regardlessof streaks). So there would be no need to track results. As a consequence, games with friends and other frequent opponents would have only little influence on each other's ELO, which I think is a good thing.

This is a good solution if the BGA team insists on lowering the impact of players repeatedly playing each other. It would allow us to play timed matches with friends without having to worry about gaining or losing too much elo. And of course, reducing the K-factor shouldn't apply to blind match-making in Arena.

[KenFin1]

I'm confused. What exactly was wrong with the old ELO system in the first place?

[ab in dc]

For what it's worth I'm totally fine with reducing the K factor after a certain number of games, as long as it's done consistently. There's nothing magical about K=20 and a lower number just makes the system a bit more stable, which would seem to address the original concerns about power leveling. It means repeat games are played for slightly lower stakes, but that's all. No imbalance in either direction.

[h_illes]

A lower K factor would effectively slow down changes. Everyone's ELO would still converge to the value reflecting their skill, just slower, and once it got there, it would remain more stable. Players would have to rely more on skill if they want to reach the top 10 (or whatever top spot) rather than via a lucky streak. Personally I consider that a good thing. Just make sure the K factor doesn't get too small, as then all games are just friendly games practically.

[6element]

So this lowered K factor would be applied conditionally when playing against relatively frequent opponents, not just "after game 5000"? Because if the goal of the feature is to reduce cheating / manual Elo corrections due to cheating - cheating can start without waiting for 5000 games.

I wonder if lowered K factor against opponent A, but not against opponent B at the same table is easy to implement.

PS: The thread got a million views.

[ChiefPointThief]

We can take the topic of cheating out of conversation. I can’t take it serious when bga is still doing nothing to take cheaters off of the arena champions list/xp list. People who are caught cheating are not even restricted. You don’t see a report if you go to their page. But for a light insult like telling someone they are being a sore loser you do. While I am happy that they crackdown on insults to an extent so that we don’t have a site of trolls spewing nonsense everywhere cheating is a far major offense imo. Also these changes are NOT stopping cheaters.

[The Nuzzer]

...bga is still doing nothing to take cheaters off of the arena champions list/xp list...

How do you cheat in Arena mode to become Champion? I thought the matching was random. Asking for a friend

[ChiefPointThief]

...bga is still doing nothing to take cheaters off of the arena champions list/xp list...

How do you cheat in Arena mode to become Champion? I thought the matching was random. Asking for a friend

Have multiple accounts and que at the same time. I’ve only played a fraction of games on this site. Yet in games I played there were about 6-7 people proven or suspected of cheating at the top of arena boards last season. Some of these ppl are just extremely sloppy and that’s why they got caught but the actual # of ppl cheating overall in arena is probably ridiculous. These new changes to elo are targeting regular elo but does nothing for aes which appears to be the main reason for ppl cheating. Team games, games w/ 2,4,5,6 players, games with a luck factor, strategic games, all were targets of cheating that I saw. I wouldn’t advocate for manipulating aes though.
As a arena player who plays a lot of games with a luck factor it is already grueling considering that losses are inevitable and when you reach a certain point you have to win 4/5 games for every loss. I couldn’t imagine additional obstacles like reducing pts in arena after consecutive wins.