Board Game Arena

Stroom wrote: ↑19 December 2023, 06:53

h_illes wrote: ↑19 December 2023, 01:52

Stroom wrote: ↑18 December 2023, 23:31 Your original claim is pointless. Other players are playing with the same constraints as well. So overall, it will even out and the rankings are still comparable. Elo is not a score that is supposed to increase indefinitely. It is a very rough approximation of your skill. Just play the games and let it rise as your own skill level increases... not by gaming the system by avoiding players who are always beating you or playing only weaker opponents who you are more likely to beat.

Let me try to explain. Assume two players play against each other regularly. (They might play against others as well, we'll get to that in a moment, but let's just focus on their games against each other for now.) Their skill level remains the same, with no improvement or degradation. Then their ELO should remain roughly the same, right? Only fluctuating a bit due to the randomness of recent results. Well, with the current ELO power levelling fix, both their ELO will go down in the long run (as games are negative-sum instead of zero-sum). And if they play against other players as well, their ELO will go down even more, since this is true for every pairing. Actually, with the power levelling fix, the total ELO of all player bases in all games will go down simply from playing. That's just wrong.

Elo is meaningful only if you play with a wide variety of players. A group of players only playing amongst themselves and their elo going up or down, is not comparable to the whole population.

h_illes wrote: ↑19 December 2023, 07:00

Stroom wrote: ↑19 December 2023, 06:53

h_illes wrote: ↑19 December 2023, 01:52
Let me try to explain. Assume two players play against each other regularly. (They might play against others as well, we'll get to that in a moment, but let's just focus on their games against each other for now.) Their skill level remains the same, with no improvement or degradation. Then their ELO should remain roughly the same, right? Only fluctuating a bit due to the randomness of recent results. Well, with the current ELO power levelling fix, both their ELO will go down in the long run (as games are negative-sum instead of zero-sum). And if they play against other players as well, their ELO will go down even more, since this is true for every pairing. Actually, with the power levelling fix, the total ELO of all player bases in all games will go down simply from playing. That's just wrong.

Elo is meaningful only if you play with a wide variety of players. A group of players only playing amongst themselves and their elo going up or down, is not comparable to the whole population.

Yes, I was addressing that, too. If you play against a wide variety of players, ELO will go down in every single matchup separately in the long run. Same for everyone else, too. For the entire player base of a given game.

h_illes wrote: ↑14 December 2023, 15:57 Another issue with the power levelling fix is the following situation. Take two strong players who play against each other regularly (for whatever reason, small player field etc.). They both win close to 50% of the games between them, but sometimes one of them gets a bit of a win streak going, sometimes the other. With the current fix, during a streak, the loser gets full points reduced, while the winner gets less than full points added. So in the long run, both of their ratings go down - just from playing each other. This is a serious WTF, and a strong motivation against playing. Please don't do this. The current power levelling fix is a bad mistake which should be fixed as soon as possible. Go with the K factor reduced over a win streak instead, that achieves a similar effect without the harm.

h_illes wrote: ↑14 December 2023, 15:57 Another issue with the power levelling fix is the following situation. Take two strong players who play against each other regularly (for whatever reason, small player field etc.). They both win close to 50% of the games between them, but sometimes one of them gets a bit of a win streak going, sometimes the other. With the current fix, during a streak, the loser gets full points reduced, while the winner gets less than full points added. So in the long run, both of their ratings go down - just from playing each other. This is a serious WTF, and a strong motivation against playing. Please don't do this. The current power levelling fix is a bad mistake which should be fixed as soon as possible. Go with the K factor reduced over a win streak instead, that achieves a similar effect without the harm.

Stroom wrote: ↑19 December 2023, 06:33
These are two completely different things. I care about elo being displayed because then I can better assume the skill level of a player. But the actual number of who is good or bad, changes over time. But the calculations changes I do not care about. If everyone is playing by the same rules, over time everyone's elo will still be comparable. Maybe today I would think that 900 elo is an indication of bearable game partner but tomorrow it might be 800. Doesn't matter i nthe grand scheme of things. If power play and elo decay affect everyone equally, it won't really matter.

The actual implementation isn't really explained well but people get very defensive without any mathematical analysis, whether it actually becomes worse or not... Not by individual games but rather by the overall system. If you only play games with a handful of players, then your elo calculation does not become comparable to the whole elo system. You are meant to play with a variety of players.

If you have 2 groups completely separated from each other and they play the game differently, one group's top player and another group's top player might end up with the same elo... but you can not compare these elos with one another as one group could just be playing on a different level. But for individual elo calculations, they might have the same score. Now if those 2 top players would play, you might see that actually there is a skill difference and the player in one group easily defeats the other group's top player. Similar things happen when you are used to playing in real life and suddenly play online - much more variety in games and more developed strategies.

Stroom wrote: ↑19 December 2023, 07:34

h_illes wrote: ↑14 December 2023, 15:57 Another issue with the power levelling fix is the following situation. Take two strong players who play against each other regularly (for whatever reason, small player field etc.). They both win close to 50% of the games between them, but sometimes one of them gets a bit of a win streak going, sometimes the other. With the current fix, during a streak, the loser gets full points reduced, while the winner gets less than full points added. So in the long run, both of their ratings go down - just from playing each other. This is a serious WTF, and a strong motivation against playing. Please don't do this. The current power levelling fix is a bad mistake which should be fixed as soon as possible. Go with the K factor reduced over a win streak instead, that achieves a similar effect without the harm.

That's no proof tho. Just your general opinion.

"To prevent this, any consecutive wins against the same opponent will reduce the amount of ELO points awarded from this victory. This reduction will be gradually higher as more consecutive wins are accumulated. This way, we encourage you to challenge other players to keep the original spirit of the ELO: show your skill at a game compared to the whole community, not just the same ones."

Common sense would say that they are just decreasing the K factor for consecutively winning against the same person. How much the K factor is changed and after how many games - have to find out. Maybe it only kicks in after 5 games and the change is like... 20% less. In that case, I don't find it as an issue. If it is after 2 games and it's 80%... yeah, bad.

So if you usually would win/lose 10 points for winning, if you keep beating the same player, the K value would be decreased. And the other player will also lose less points as you win more. Let's say now you only win 5 points. He also loses 5. So it is still zero-sum. But if the other player wins, the K value is reset to 10 and he wins 10. But before that you had already won 10, 5 against him so you are still ahead.

Edit: Seriously... most players here are strongly negative and start to echo chamber off of each other's posts... without reading the original post anymore. Or not even trying to reason with the changes. Only afraid of the changes, not trying to do actual calculations of how it would affect the whole system. Let's give it a chance and see what happens. The system would be different and more volatile than other elo systems but I do not see it as horribly broken as other players.

lordalx wrote: ↑13 December 2023, 10:41 → Red Thumbs for competitive play
One last thing that is going to be changed in a few weeks will be the impact of the Red Thumbs.
You can currently give a Red Thumb to any player with which you had a bad game experience. This prevents you from competing against them in ranked games or the Arena, but not in Tournaments.
This system can, and unfortunately is, used to prevent some players from competing against the best players. Considering that the Red Thumb system is being abused this way, on purpose or not, we will update the way Red Thumbs are impacting the matchmaking for the next season.

estroncio wrote: ↑19 December 2023, 08:55 Matchmaking is broken and this is a mitigation.
The Arena system thinks a 200 ELO player with 30% win rates (in a 1vs1 game) is an "Elite" player. The outcome cannot be a surprise to anyone, any half good player will wipe out this "Elite" player in a totally pointless game. Twenty minutes of life wasted.

At least you could reduce this issue my red thumbing weak players. Not anymore apparently, FML.

anykidrated wrote: ↑19 December 2023, 09:05

estroncio wrote: ↑19 December 2023, 08:55 Matchmaking is broken and this is a mitigation.
The Arena system thinks a 200 ELO player with 30% win rates (in a 1vs1 game) is an "Elite" player. The outcome cannot be a surprise to anyone, any half good player will wipe out this "Elite" player in a totally pointless game. Twenty minutes of life wasted.

At least you could reduce this issue my red thumbing weak players. Not anymore apparently, FML.

What game does this happen in?

+---------------------+-----------+--------+-------------+
| Date                | Table Id  | O. ELO | O. Win Rate |
+---------------------+-----------+--------+-------------+
| 2023-12-15 at 09:08 | 450316142 |    279 |         31% |
| 2023-12-15 at 12:39 | 450316142 |    250 |         34% |
| 2023-12-13 at 16:11 | 449700951 |    126 |         24% |
| 2023-12-13 at 13:31 | 449653533 |    166 |         38% |
| 2023-12-11 at 08:40 | 448852721 |    231 |         36% |
| 2023-12-07 at 11:53 | 447436429 |    219 |         34% |
| 2023-12-04 at 23:18 | 446578965 |    259 |         54% |
| 2023-12-01 at 08:50 | 445267753 |    137 |         21% |
+---------------------+-----------+--------+-------------+

+---------------------------+-------------------------+
| I'm okay playing with     | My level                |
| I never want to play with | Near my level (+/- 1)   |
| I never want to play with | Around my level (+/- 2) |
| I never want to play with | Far from my level       |
+---------------------------+-------------------------+