Board Game Arena

Dear BGA, it's unfathomable to me that we are in mid-2024 and in Arena, we still have the same K factor for each game on the site. The competition in games with some decent randomness is a joke. I recently started playing Arena again and I noticed how I start to dislike games I once liked a lot, just because you rub the randomness of these games on everybody's nose. They are already random enough, you don't need to make it extra-extra luck-based by implementing twice the K factor that is used for Elo.
I would love to play regularly in a healthy way, not in binges, which is incentivized by the system. You can't have the same K factor for Terra Mystica and Can't Stop. You simply can't. It's such a blunder. You are a cause of suffering, rage quits, starting to dislike favorite games. And the sad part is, it is soo unnecessary. You can't tell me that you cannot put different K factors for different games. It's an absolute insult to everyone who wants to play these games competetively.

I would really love to play the games regularly and not only once I have forgotten enough what a joke this Arena mode is. Yes, I can play outside Arena, but for some reasons, the best opponents tend to play more regularly in Arena and I get drawn into it. If that's your strategy, fine, but this is an extremely poor and careless choice. I am sure you can do better.

There is a simple fix - stop taking games with so much randomness as serious and competitive games. Just treat them like the braindead party games they are, you will probably be back to having fun with them instead of suffering. You are engaging with the game in a way that is causing you to suffer lol.

dschingis27 wrote: ↑03 July 2024, 07:16 . You are a cause of suffering, rage quits, starting to dislike favorite games. And the sad part is, it is soo unnecessary.

Completely true. You can choose not to play Arena. BGA gives you that option, but apparently you really want to make yourself miserable.

OP makes a valid point. The 2 replies above are a bit off. He already knows that he can play outside Arena, but BGA offers these games inside Arena. He obviously also likes to play these games competitively, otherwise he wouldn't be here making this statement. I can't see anything wrong this argument.

I'm guessing you already know the - make a suggestion and we'll upvote it thing.

I don't follow. What is the specific problem with K factor, and why is it the K factor's fault when you rage quit a game? Your penalty for rage quitting is the same as playing the game to completion so I don't see why the K factor is an issue. Also, your opponent's K factor doesn't affect how your ranking gets adjusted, so again I don't understand what the specific issue is here.

I do agree Arena has better quality opponents (less likely to find people who are going to run out of time or rage quit because they actually care about competing), but the point of Arena is not to be ranked #1 on the site, it's to have a more competitive environment. If the point was just to be ranked #1 then there wouldn't be anyone playing in the bronze-gold tiers because they don't stand a chance of being #1.

I often play Arena when I can (to avoid players who ignore the turn timer limits), and yes it's nice to see my rank go up, but my rank only serves to help find competitive matches. K factor probably works well for me because I tend to bounce around from game to game and don't stick with one long enough to get a proper placement in a season, as such if I shouldn't be in bronze tier it pulls me out of bronze tier faster. This also means if I'm in silver tier faster it doesn't penalize my opponent's rank as much if I beat them because I "should" be higher rank than I currently have.

I should explain a little bit more.

Cile wrote: ↑03 July 2024, 16:40 I don't follow. What is the specific problem with K factor, and why is it the K factor's fault when you rage quit a game?

The K factor determines the "recency bias" in the Arena score, it determines how severely the Arena score is changed after a single win or loss. If the K factor were 100000, you could only have a high Arena score if you won your last game. If the K factor were only 1, your Arena score would be very inelastic and it would depend on the last 1000s of games played. Counter to what some believe, it is not just a scaling factor, it is a true determinator on how much your recent games count compared to your older games. The current K factor for Arena is 40 which is twice as large as for Elo where it is 20. When I wrote "rage quit", I meant stopping a gaming session after a loss due to the realization how much the recent luck is important for your Arena score and how little influence you have, I haven't meant the abort during a match.

Cile wrote: ↑03 July 2024, 16:40 but the point of Arena is not to be ranked #1 on the site, it's to have a more competitive environment.

That's exactly what I am at. The current Arena experience for those competing for Top100, Top50, Top20 is just really bad. The Arena score and your rank jumps around just way too much to derive any lasting satisfaction from it. I don't care about #1, the #1 will always be hugely influenced by luck in every system. But I want some sense of accomplishment and meaning in the ranking. My wish would be, one season I am Top40, next season I am Top30, so I can see my progression. This is not achieved by randomly catapulting me into Top10 on Monday while letting me fall down to rank #200 on Tuesday. It doesn't make sense for a 3-months long competition. It shouldn't be that I can go from #10 to #100 (or from #100 to #10) in just 3 short games. It shows how much your Arena score is not based on your skill but a whole lot on the very recent luck. And that just makes the whole ranking, the whole point of the competition meaningless.

Of course my original post was a rant and the tone a little off. There are even more things to fix for Arena but the K factor is just such a simple and obvious fix to me that it's hard to understand why it stayed like this for the last 5 years.

dschingis27 wrote: ↑03 July 2024, 17:29 My wish would be, one season I am Top40, next season I am Top30, so I can see my progression.

This progression will not be visible in any "game" which is basically 'toss a coin, heads = player A wins, tails = player B wins', regardless of what elo system you use. The more swingy/jumpy nature of ranking that you described is in fact what one would expect in these types of "games".

buu441 wrote: ↑03 July 2024, 18:56

dschingis27 wrote: ↑03 July 2024, 17:29 My wish would be, one season I am Top40, next season I am Top30, so I can see my progression.

This progression will not be visible in any "game" which is basically 'toss a coin, heads = player A wins, tails = player B wins', regardless of what elo system you use. The more swingy/jumpy nature of ranking that you described is in fact what one would expect in these types of "games".

Yes, in extremely luck-dependent games like Can't Stop or Lucky Numbers. However, I am thinking about games like 7Wonders or Railroad Ink with some luck, some strategy. The design of the Arena score even increases the luck element on its own even beyond the luck that is already in the games. If the K factor were smaller, the swings in the rankings would be smaller as well.

dschingis27 wrote: ↑03 July 2024, 17:29 It shouldn't be that I can go from #10 to #100 (or from #100 to #10) in just 3 short games.

That's more of a factor of the number of players and their relative ranking than the K factor. If #10 to #100 are relatively close in rank it doesn't take much of a change to go from one end to the other. If you're playing enough to try to compete for a high rank in a season your K factor is going to quickly approach 1 anyway.

I proposed a while ago to use the 'Luck' rating in the k-factor formula. I just checked for Can't stop and Lucky number and it's at 4, so it would work.
And then I proposed to use the 'Interaction' part with 3+ players games to reduces the '% to win' of the highest elo (ie make it so that the highest elo lose less, and lowest win less ; but not less than if they got identical elo). This is because targetting a player is easier on those games (I believe).

Now the problem is how to find the right number? Do you apply it to all ranking?
I think you could.