Board Game Arena

ufm wrote: ↑22 March 2025, 05:06

kuzestanx wrote: ↑21 March 2025, 15:01

ufm wrote: ↑20 March 2025, 23:39 Actually, official tournaments have already adopted weighted scoring for years.
Unfortunately, BGA's rating system is not appropriate to handle these 'asymmetric' games with different odds. It simply can't do that.

yes yes and we can see it here https://bang.dvgiochi.com/northamerica/#Rules (page 4 of 8) and if you read the AVTactics study you must be wondering: is everything ok with this scoring system, is it balanced, how did they make it etc

Of course I read it. The problem is:

AVTactics wrote: ↑20 March 2025, 18:24 What can BGA Administration do?

As far as I understand, the Arena points system can be changed.

This is totally wrong.
BGA can't handle rating systems like this, rendering all attempts to directly assign points following role weights moot.

The best BGA can do is assigning ranks differently.
For example: If Law wins, instead of Law > everyone else, Law > Renegade (if Renegade lost at the final duel) > everyone else

I'd say there is more problems. I think Tournament's system system is unfair and unbalanced, and I think so knowing current win rates, see it in previous comments.

And also I think it is not a good idea to give points for rene's lose in final duel with sheriff. Now rene understands it is not necessary to win in 1v1v1, and outlaws now hate more the renegade

Then you have no one option left except voting for a different player number in Arena mode.

ufm wrote: ↑22 March 2025, 13:03 Then you have no one option left except voting for a different player number in Arena mode.

Could you explain why?

I already explained why. Rating calculation such as this is impossible with the current BGA framework:

AVTactics wrote: ↑20 March 2025, 18:24 Outlaws = 5 * 1,7 = 8,5 points for each player;
Sheriff & Vice = 5 * 3 = 15 points for each player;
Renegade = 5 * 14,3 = 71,5 points.

BGA's EAS adjustment always follows Elo rating system. It ignores the roles entirely and just compares the final rank inputs.

ufm wrote: ↑23 March 2025, 02:13 I already explained why. Rating calculation such as this is impossible with the current BGA framework:

AVTactics wrote: ↑20 March 2025, 18:24 Outlaws = 5 * 1,7 = 8,5 points for each player;
Sheriff & Vice = 5 * 3 = 15 points for each player;
Renegade = 5 * 14,3 = 71,5 points.

BGA's EAS adjustment always follows Elo rating system. It ignores the roles entirely and just compares the final rank inputs.

I missed your broad explanation of why it's better to have a 5-player arena. From AVtactics' research, I see that 5-player and 6-player modes are about equally unbalanced. So what's your point?

kuzestanx wrote: ↑23 March 2025, 17:19 I missed your broad explanation of why it's better to have a 5-player arena. From AVtactics' research, I see that 5-player and 6-player modes are about equally unbalanced. So what's your point?

I didn't say the player count must be changed.
What I meant is, the only choice given to players is either voting for a different player number or keeping the status quo, if they are not going to petition for rank adjustments according to the result.
As I said multiple times, directly assigning a specific number of points is simply impossible.

Before getting into game specific ratings (bang) the first problem that needs to be addressed is that how team rankings are calculated is flawed across the board. Furthermore, for bang bga never even fixed the problem of "tying" w/ your teammate that was supposed to be addressed for all team games over a year ago. https://boardgamearena.com/forum/viewtopic.php?t=33785

Currently players on the same team have different win probabilities. Thus, elo is distributed differently to players on the same team but why? If I play doubles tennis w/ Roger Federer I have a higher chance of winning than if my partner is an elderly person in a wheelchair and me and my partner have the same exact chance of winning the game.

ufm wrote: ↑23 March 2025, 02:13 I already explained why. Rating calculation such as this is impossible with the current BGA framework:

But bga is capable of fixing team elo in the manner I described above so hopefully this gets done!

ufm wrote: ↑22 March 2025, 13:03 Then you have no one option left except voting for a different player number in Arena mode.

kuzestanx wrote: ↑23 March 2025, 17:19 I missed your broad explanation of why it's better to have a 5-player arena. From AVtactics' research, I see that 5-player and 6-player modes are about equally unbalanced. So what's your point?

I don't believe ufm said any number in specific. There is an option of 4 players. The community is aware of the unbalance of 5+ players in bang. 4p is the most statistically balanced but yet it is the least popular (shrugs).

I am also disappointed that there are no intentions of adding bang duel which would add an option within the bang universe that solves this unbalance. Instead, several expansions are being added to the base game that doesn't offer the variety here to the bang universe as band duel would

ChiefPointThief wrote: ↑24 March 2025, 14:38 4p is the most statistically balanced but yet it is the least popular (shrugs).

Can you show the statistics you mention?

Wow, amazing insights and info/stats related to the game. Pretty nice work.

In my opinion I find the reward system for Arena proper as it is true chances for Renegade to win are lower but on the other side as AVTactics has mentioned the reward is higher than other roles.

What I find not very fair is deducted ELO in Arena as they should have separate logic between what you gain and what you lose. When you lose as Renegade would be more suitable to have a lower ELO loss as the chances are lower.

Introducing some complex logic with Risk Coefficient or considering when Rene dies in the game etc, would increase the Development complexity too much and I think Devs of this game wouldn't be very happy to explore as an option

AVTactics wrote: ↑24 March 2025, 22:50

ChiefPointThief wrote: ↑24 March 2025, 14:38 4p is the most statistically balanced but yet it is the least popular (shrugs).

Can you show the statistics you mention?

My main concern is how extremely unbalanced the renegade role is. I don't like the feeling of "oh great! I drew the renegade. This game will suck". Although the renegade still sucks in 4p it isn't nearly as bad as the other player counts. So you figure on average in 4p you should be the renegade 25% of the time. Considering there are 4p a winrate of 25% would be average. Me and a friend had this convo about the renegade last year so I manually checked my last 100 games to get an estimate of my %. But I think better statistics would come from sitewide so I just collected data from the site of the last 100 4p games. This was my findings.

Outlaws won 49%, Sheriff 34%, Renegade 17%. So although renegade still falls short a little it feels way more even for me personally and I feel like I have a legit chance when drawing the renegade in 4p as opposed to other player counts. I'm sure if statistics were taken for the site that renegade is <8% for the other player counts. Also you will see outlaws are even @ 4p with the sheriff being slightly over due to stealing wins from the renegade which makes sense.

Another thing to note as it pertains to arena and you advocating for a better rating system is the ability for players to get the renegade at an uneven rate. As mentioned before, last year I checked the last 100 games of me and a friend. In those games I averaged above the % of what I should for getting renegade and my friend was extremely below (only 5%). While renegade rate is 20% at 5p you may have one player who has a 5% rate in a arena season and another who has 35%. Without a weighted ranking the player w/ the %5 rate has a clear advantage over the other.

If you were questioning my comment about 4p being the least popular there are actually tons of 4p games played. I didn't do a count of all games to compare but I know when I suggested 4p in arena it was extremely shot down and a lot of my friends thinks it sucks also

Devs if you could share info on game player count % that would be great.