Shared victory can still result in an Elo loss

[Space4]

Result | Elo before | Change | Elo after
Joint 1st | 458 | -2 | 456
Joint 1st | 176 | +12 | 188
3rd place | 152 | -9 | 143

The higher-rated player defeated the third-place player but tied with the much lower-rated co-winner. The Elo loss from that draw outweighed the Elo gained from the win, resulting in -2 Elo despite finishing joint first.

In games such as Nana Toridori, BGA applies a “Teammate: -100%” modifier between joint winners, preventing them from gaining or losing Elo against each other. This seems fair.

Why isn’t this modifier applied consistently in every game that allows shared victories?

[Silene]

Result | Elo before | Change | Elo after
Joint 1st | 458 | -2 | 456
Joint 1st | 176 | +12 | 188
3rd place | 152 | -9 | 143

The higher-rated player defeated the third-place player but tied with the much lower-rated co-winner. The Elo loss from that draw outweighed the Elo gained from the win, resulting in -2 Elo despite finishing joint first.

In games such as Nana Toridori, BGA applies a “Teammate: -100%” modifier between joint winners, preventing them from gaining or losing Elo against each other. This seems fair.

Why isn’t this modifier applied consistently in every game that allows shared victories?

when in a team, no matter what you do, you tie with them - so it makes no sense to evaluate the result vs each other. When not in the same team, you had the chance to outplay the other. when you tie with them instead, your result vs. them is a bit below expectations (when you have higher elo). so you lose some points. seems fair to me.

when on high elo, you need to consistently show better results if you wanna maintain that rank. if you can't do that, it probably means you're overrated at the moment (after a lucky streak usually). so it makes sense to easily lose some points and get back to a rating that reflects your skill more accurately.

remember elo is not a price. better to not care too much about it.

[Space4]

I understand the distinction between teammates and independent players, and I agree that the current calculation is logically consistent with Elo.

My concern is specifically about Arena matchmaking. In a normal game, I can restrict the Elo range of the opponents I play against. In Arena, I have much less control over who I am matched with. A highly rated player can therefore be placed against much lower-rated opponents and lose Elo despite officially winning the game.

I am not suggesting that joint winners should always be treated as teammates. I am suggesting a simple protection rule: a player who officially finishes first should never receive a negative Elo adjustment. If the normal calculation produces a negative result, it could simply be capped at zero.

I understand that this would make Elo slightly less mathematically accurate and could cause some rating inflation. However, from a competitive and player-experience perspective, losing Elo after winning feels fundamentally wrong—especially when the matchmaking system selected the opponents.