FFA games influencing team game balance

Currently FFA game results influence balancing of teams in team games. FFA games are vastly different in playstyle compared to team games, one requiring practically no ability to cooperate with your team and the other being almost completely reliant on it. It is not logical to alter team compositions based on performance in FFA games, where vastly different goals and skills are at play.

From my understanding, FFA offers minimal risk in regards to losing rank compared to gaining it, since loses result in very minor rank loss and victories in comparatively massive gain. Making it a very popular way to increase ones rank without actually playing at a higher level, resulting in sudden high ranked players surprisingly underperforming in team games when faced with having to cooperate with their teams towards victory.

A solution would be a separate FFA ladder and a team vs team ladder, which would lead to more balanced team games. FFA players would see their proper ranks for FFA, having a general idea of how good which player actually is in that environment, fostering competitive behavior.

Note that I have not said which mode is "most difficult" or other such nebulous terms, I have said that team games and FFA games are vastly different in amount of cooperation required. I am saying that you winning a FFA game should have no effect on your ranking in team games, because of the inherently different requirements for these games.

I understand that you're a predominantly FFA player that has grinded a ton of rank in such games, I'm not saying that it's not a worthy achievement or less difficult than playing team games. I'm saying that such games are not representative of team games where you have to cooperate with your team towards victory. Of course exceptional players can succeed at learning both of these tasks, but that does not change the fact that they are completely different abilities.

Nobody in their right mind would suggest that the 1v1 ladder be merged with team ladder, because yet again they are vastly different games. Similarly, FFA is vastly different to 1v1 and team games. I also understand that the main developers, veteran players and even most moderators are FFA regulars, so I do not expect the slightest bit of enthusiasm about such an idea as it is so ingrained in current status quo. I am however hopeful that at least some rational reasoning can be seen in further arguments.

I would also like to note that you yourself advocated for such separation in https://zero-k.info/Forum/Thread/25060?postID=207267#207267

I did not make any assumption about any "50% average win chance". Please do not assume what I have not written. A victory in a FFA will net you a much larger reward than the loss from a defeat, which logically means you only need to win a small fraction of your FFA games to consistently rise in rank. Rank which is then used to balance team games.

Again, I made no claim which game mode requires more skill or is more difficult. Game mode difficulty is in no way an argument for merged ladders of vastly different game modes. I am not taking away your achievements in FFA games or saying they are somehow inferior, I am saying the are different from team games and should be separately ranked as such.

I will not debate semantics about who assumed what with you further, as to me it feels as you're posting in bad faith in that regard.

In FFA games, any and all rank gains and losses are completely irrelevant to the mode itself. The mode does not require any balancing between players to function, it introduces no mechanical penalties for good players or bonuses for bad players. If there was no ranking system at all, FFA games wouldn't change at all as the ranks for FFA are only for record keeping rather than maintaining "fair" games.

In contrast, team games absolutely rely on balanced player teams to properly function. Which requires a ranking system in place. In a team game, with higher rank you are much more likely to be placed on a team with less experienced players beside you. All these players very directly impact each other in terms of shared income, reclaiming/feeding the enemy and cooperating in attacks or strategies.

How much do FFA results realistically affect team games though? Looking through the replays, there are only a few FFA games a day, and there are even days when no games take place. Every game can only have one winner, so whatever effect there is should logically be minuscule to the point of being unnoticeable?

Granted, FFA is extremely reliant on various non-skill based factors unlike 1v1/teams, but anyone who can consistently win those games should have no trouble keeping up with the above average team player.

Assuming perfect balance, FFA actually doesn't count nearly as much for team rating as I'd like it to. The "weight" of each game in your rating history is inversely proportional to the number of players in that game. This means, that big FFAs which are very hard to win count only very little towards team rating. Additionally, FFAs are modeled as team games of the winner against the losers. This doesn't take into the account the fact that the win chance is split among many more than just two teams.

So, besides FFA wins being a more reliable indicator of skill than teams (solo performance), FFAs are already weighted much less than they probably should be. Keep in mind that FFA losses on the other hand are one of the least informative game results, as you are expected to lose.

As for the transferability of the FFA <-> Teams skill set, you'd need much more data per player for a split rating to be effective. The 1v1 <-> Teams split already only exists so that players can play team games without affecting their competitive rating.

TinySpider you say that the ffa player loses a little rating. It wasn’t always like that. I lost 125 elo once in 5 games, 2550 -> 2425, 2550 then it was top 6-7 position. 2550 i got playing only team games. When I finally won in ffa i got back 27 elo. Where are my remaining 100 elo TinySpider?
I remind you that these 100 elo are an indicator of my skill in a team game. You know, if I continued to play ffa, then my elo would drop it below, to about 2200, maybe 2100
If I came to a team game with 2100, I would get a balance with a weaker opponent, because my real elo 2500

Anarchid already wrote to you why this happens.

In fact, I for the fact that there would be a separate ffa rating, Just to make ffa more popular

It seems to me quite likely that there exist many players of more-or-less equal skill as teamgame players who have widely disparate FFA skill. I would have to run the numbers to know whether these two players would have significantly different casual rating in practice, or whether making the FFA games not affect casual elo would improve their ratings' accuracy.

[Spoiler]

We want FFA ELO!

quote: The more games you exclude the less accurate the rating becomes. This is true pretty much always for zero k.

Would it be easy for you to run a test seeing how much less (or more) accurate the lobsterpot predictions become when you only have lobsterpot games?