Points for winning a game.

[Ceaseless]

Well, in Arena, if a platinum player beats a gold player (in a 2-player game), the gain is 2 medals.

I find it odd that beating a gold player AND a bronze player in a 3-player game gives fewer medals than just beating the gold player in a 2-player game.

Why? With 2 players, you have to beat the opponent without help. With 3 players, you do not.

True, the bronze player could end up accidentally helping the platinum player with bad plays. But this could also help the gold player. I would think the overall effect would be neutral. And there's the possibility the bronze player could win.

Seems tough to advance if a 3-player game win only gives one medal.

It could help either player. So adding a bronze player to the mix doesn't necessarily mean a player should get more for the win. Since advancing is a relative thing, I don't think this makes advancing such an issue, as everyone would have to handle this. If it's a matter of league gap issues, then league filters can help with that at least.

[FrankJones]

Putting aside whether the player should get more medals for the 3-player win than for the 2-player; I still find it odd that the 3-player win would result in fewer medals gained.

[tbhp]

Isn't that odd though? I thought it was +2 medals for beating someone one tier below. In this game, only one medal total was given for a game in which there were two opponents, one of whom was only one tier below. That win alone should result in 2 medals, right? And I would think winning against the other player would also be at least one medal for a total of at least 3.

The way it works for multiplayer games probably isn't cumulative. Otherwise players could complete their new league in just one game in games that are 5+ players.

I mean, I know leagues don't matter much anyway, but it's more fun if you have to play at least a few games.

[tbhp]

Well, in Arena, if a platinum player beats a gold player (in a 2-player game), the gain is 2 medals.

I find it odd that beating a gold player AND a bronze player in a 3-player game gives fewer medals than just beating the gold player in a 2-player game.

Why? With 2 players, you have to beat the opponent without help. With 3 players, you do not.

That's an interesting point. But the way number of players affects win probabilities for each rank of player is dependent on the mechanics of the game in question (and also the current meta of it I suppose, which in a past thread I think you referred to as "game politics").

So, I'm not sure how they could have turned that into an algorithm when it's not like there is one solution that fits all cases.

I think the answer is much more simple, something along the lines of "do we really want players to advance to the next league just because they got lucky and finished first in a couple of games?" They just wanted the fun to last a bit longer and for players to prove themselves a bit more before they advance to the next league. Then they decided on an arbitrary reducer to achieve that.

[Ingoschi]

Isn't that odd though? I thought it was +2 medals for beating someone one tier below. In this game, only one medal total was given for a game in which there were two opponents, one of whom was only one tier below. That win alone should result in 2 medals, right? And I would think winning against the other player would also be at least one medal for a total of at least 3.

But in a multiplayer you get the mean value, so your total is divided by the number of opponents. So in your example it's (2 + 1) / 2 = 1.5
So it looks like it was rounded down to 1.

[tbhp]

Isn't that odd though? I thought it was +2 medals for beating someone one tier below. In this game, only one medal total was given for a game in which there were two opponents, one of whom was only one tier below. That win alone should result in 2 medals, right? And I would think winning against the other player would also be at least one medal for a total of at least 3.

But in a multiplayer you get the mean value, so your total is divided by the number of opponents. So in your example it's (2 + 1) / 2 = 1.5
So it looks like it was rounded down to 1.

That's possible, I guess the confusion is that unlike Elo, league points don't just add up in a multiplayer game, and in cases where they can round down, they will and the decimals just disappear.

Elo considers that you won against all the players you finished ahead of and lost against all the players you finished behind. There is also a reducer but it's a fixed one (I think like 50% or 66%), it's not based on the number of players to create a mean. It also keeps track of the decimals even though it shows as a round number. So the way the two things work is not at all comparable.

But like I said earlier, the site can handle league points however they want, because it's not what really matters anyway. What actually matters is your hidden Arena score. The way you climb and eventually reach Elite League, at which point your score will be shown, is very arbitrary. The choices made in this regard are more about the experience that they want to offer players while they climb.

[BarnardsStar]

Elo considers that you won against all the players you finished ahead of and lost against all the players you finished behind. There is also a reducer but it's a fixed one (I think like 50% or 66%), it's not based on the number of players to create a mean. It also keeps track of the decimals even though it shows as a round number. So the way the two things work is not at all comparable.

The reducer is not a fixed number. It is based on the number of players. Details are in this thread but the gist is this: each game has a "recommended number of players" (which I gather is a value which used to be visible when that thread was created 6 years ago, but now I cannot find where it might be indicated); below the recommended number of players, the reducer is calculated as (p/2)/(p-1) - at 2 players this works out to 1/1, but at higher values it scales down accordingly.

However, *above* the recommended number of players, the reducer is calculated as (r/2)/(p-1) where r is the recommended number of players, not the actual number of players. The thread I linked to above uses 7 Wonders as its example, where apparently the recommended number of players is 4. So for 2 players the reducer is 1/1 (not really a reducer at all); at 3 players it is (3/2)/2 or ¾. At 4 players it is ⅔ and then for 5, 6, and 7 it is 2/(p-1). I've checked with recent games of 7 Wonders and these ratios still hold 6 years later. They'll be different for other games depending on where their "recommended number of players" is set at.